Designing an object in an interactive design environment is performed by capturing information from interactions of a user, arranging the information into a knowledge graph. When a user changes an aspect of a design, at least one other aspect of the design is identified that is affected by the user's change. By examining the knowledge graph impacts are identified and visualized. A user interface of the interactive design environment provides the ability to edit the knowledge graph. The visualization may be displayed to a user in real time. The visualization may include a radar chart indicating affected aspects of the design. The visualization may highlight a violation of a constraint, or range of a metric based on past designs in the captured interactions. If a design metric falls outside the range of past designs a detailed evaluation of the change to the first aspect of the design may be performed.
G06F 30/12 - CAO géométrique caractérisée par des moyens d’entrée spécialement adaptés à la CAO, p. ex. interfaces utilisateur graphiques [UIG] spécialement adaptées à la CAO
2.
CLOSED-LOOP DATA GENERATION FOR FINE-TUNING GRASP NEURAL NETWORKS
A system includes a grasping neural network that is trained so that it is configured to determine locations for a robot to grasp objects. The system can obtain a first object dataset that defines a plurality of images that represent a plurality of objects. The system can generate a first plurality of parameters that arrange the plurality of objects into a first training dataset representative of scenes. The system can determine first key performance indicators that indicate whether the neural network improved responsive to the first training dataset. Based on the first key performance indicators, the system, in particular a Bayesian optimizer, can adjust the first plurality of parameters so as to define an updated plurality of parameters. Based on the updated plurality of parameters, the system can generate a second training dataset that improves a performance of the neural network, so as to define a closed-loop architecture.
In a virtual commissioning of an industrial automation system, an industrial system or process is simulated by a simulation engine (118) utilizing a digital twin (112) of the industrial system or process. A virtual data acquisition system (120) utilizes a simulation snapshot (128) to generate data communicated as a model input (140) to a virtual inference system (124) having an AI model (138) deployed on it. The virtual inference system (124) processes the model input (140) using the AI model (138) to generate a model output (142) communicated to a virtual automation controller (126). The virtual automation controller (126) processes the model output (142) using automation logic (116) running on it to generate control data (144) communicated to the simulation engine (118). A closed-loop behavior of the automation logic (116) responsive to interaction with the AI model (138) is thereby validated. Communications between the virtual system components (118, 120, 124, 126) are based on data models and/or runtime interfaces that match communications between corresponding physical system components (102, 104, 106, 108) of the industrial automation system.
A method for identifying quality deviations in a machine learning application, implementable by corresponding systems and computer-readable mediums, may include training (205), based on a world-truth paradigm (202), a first machine learning application (MLA) (212) to generate a world-truth prediction model (216); labeling (209), via the first MLA using the world-truth prediction model, intended-use training data (224) and field data (228); training (211) a second machine learning application (MLA) (240) using the labeled intended-use training data to generate an intended-use model (246); generating (219), via the trained second MLA, using the labeled field data and the intended-use model, at least one prediction (256) having an associated confidence level; and determining at least one quality deviation (258) of the intended-use model from the world-truth paradigm or of the at least one prediction and from the world-truth paradigm.
Methods and systems for identifying cybersecurity threats construct a graph from a CISA security alert and traverse the graph to identify a cybersecurity threat by monitoring data in the system and updating the state of the graph. Tactics in the CISA alert are represented by nodes in the graph, techniques in the CISA security alert represented by edges in the graph Traversing the graph is performed by receiving information from the system and for each graph representing an attack model comparing the attack model graph to the received information to identify a match and updating the graph state based on the match. Updated graphs may be displayed to a user showing the state of an attack vector in the graph. The state of the attack vectors may be updated with information from the system and displayed in a color corresponding to a traffic signal.
A method for providing cyber security measures in a shared network estimates in a probabilistic model, a likelihood of a cybersecurity event occurring in a system. Based on the estimated likelihood selectively applies some but not all of the cyber security measures. Based constraints of system resources available for operations a combination discretionary security measures are selected for execution. During runtime of the system, security risk of an aspect of the system is periodically evaluated causing reconfiguration of the security measures for execution based on the estimated security risk and system resources available for operations. Timed automata corresponding to security threats are assigned a risk score associated with the security threat based on one or more states of the automata. An aggregation of multiple risk scores produces a trust score for the overall system. A number of security measures may be selected and allocated based on the trust score.
Methods for controlling a train and corresponding systems and computer-readable mediums. A method (400) for controlling a train (200) includes executing (402), by a computer system (100), a software-defined train control system (210). The method includes executing (404), by the software-defined train control system (210), a plurality of train control apps (212, 214, 216, 218, 220, 222, 224), and executing (406) a middleware module (230) that enables communications between the plurality of train control apps (212, 214, 216, 218, 220, 222, 224) and train hardware and subsystems (290) of the train (200). The method includes controlling (406), using the plurality of train control apps (212, 214, 216, 218, 220, 222, 224), the train hardware and subsystems (290) of the train (200), thereby controlling the train (200).
An augmented reality (AR) computing device can identify a robotic device and a pose of the robotic device. Based on identifying the robotic device and the pose of the robotic device, the AR computing device can obtain a world model of the robotic device. While performing an operation of the robotic device, the robotic device can be displayed on the user interface of the AR computing device. In an example aspect, a geometry of the robotic device can be overlayed in a simulation position on the user interface that also displays the actual position of the robotic device. Offsets can be determined between the world model and the physical environment, for instance using the AR computing device. In various examples, the world model is adjusted based on the offsets, so as to calibrate the world model for the operation.
In a computer-implemented method for supporting vision-based inspection of manufactured parts on a production line, a defect library is created in simulation utilizing real images of manufactured part samples with defects, by digitization and parameterization of enclosed defect geometries of different defect types obtained from the real images. Synthetic defects are generated by randomizing parameters defining the enclosed defect geometries of the defect types in the defect library. A 3D superimposition of the synthetic defects is carried out at randomized positions on a CAD model of a target part to generate a number of defect model objects, each including a digital representation of the target part containing one or more of the synthetic defects. Simulated camera images are then rendered using the generated defect model objects and a digital twin of an inspection environment having one or more cameras, to create a dataset for training a machine learning model.
A method of optimizing a process having a plurality of potential inputs, comprising selecting a first set of inputs from the plurality of potential inputs, providing the first set of inputs from the to a first optimization process, running an objective function on the first set of inputs to produce a value corresponding to the set of inputs, providing the value to a second optimization process, running an acquisition function in the second optimization process to select a new candidate set of inputs from the plurality of potential inputs, and providing the selected new candidate set of inputs to the first optimization process. In one embodiment, the inputs are a set of lattice kernels for constructing a structural object. A Bayesian optimization is used to select sub-sets of kernels from the set of inputs. The inputs are provided to a topology optimization for evaluation.
A telemetry system is configured to monitor temperature of a wafer disposed in a vacuum. The system can include a plurality of sensors configured to measure temperature at a plurality of locations across the wafer. The system can further include a plurality of sensor signal conditioning circuits. Each sensor signal conditioning circuit can be coupled to a respective sensor of the plurality of sensors. The plurality of sensor signal conditioning circuits can be configured to convert a resistance from each sensor of the plurality of sensors into low frequency sinusoids that define a respective frequency that is proportional to the resistance. The system can further include a multiplexor configured to generate a composite modulating signal from the low frequency sinusoids. The system can further include a frequency modulation (FM) transmitter configured to send the composite modulating signal outside of the vacuum.
Processes for operating an autonomous vehicle (100) and corresponding systems and computer-readable mediums. A process (300) includes performing a path planning process (320) using vehicle position and perception information (352) to produce an iterative reference path (322), and performing a position control function (330) using the iterative reference path (322) and the vehicle position and perception information (352) to produce velocity and steering commands (332). The process includes controlling a vehicle propulsion system (340) to move the autonomous vehicle (100) along the iterative reference path (322) based on the velocity and steering commands (332). The process includes determining (350) updated vehicle position and perception information (352) that includes identification of any objects in or near the iterative reference path (322). The process includes using the updated vehicle position and perception information (352) for subsequent iterations of the path planning process (320).
A surrogate model for predicting computational fluid dynamics simulations uses a computational fluids dynamics (CFD) simulator and a feedforward neural network. CFD simulator receives first input parameters of ambient conditions and operating conditions for an indoor space and generates heat maps and airflow maps of the indoor space. Target region data and an image background map are extracted from the seed images. Feedforward neural network is trained as a surrogate machine learning (ML) based model to learn normalized CFD prediction data based on the first input parameters. Training is based on a loss function comparison of the normalized CFD prediction data to the normalized target region data. The surrogate model generates normalized CFD prediction data from received second input parameters of ambient conditions and operating conditions for the indoor space. CFD prediction images are generated by denormalizing the prediction data and combining with the image background map.
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
G06F 30/28 - Optimisation, vérification ou simulation de l’objet conçu utilisant la dynamique des fluides, p. ex. les équations de Navier-Stokes ou la dynamique des fluides numérique [DFN]
A comprehensive digital twin or metaverse model of a system includes a plurality of models that each represent a particular aspect of the system. A three- dimensional (3D) engine renders a scene of the model based on two or more of the models. The scene is embodied as a 3D visualization of some or all of the system and is presented to user or stakeholder via a user interface. The user interface enables a stakeholder to provide information about the system that is incorporated into the digital twin of the system. Additional information is received and included in the appropriate model representing the aspect of the system relating to the new information. In addition to stakeholders, additional information relating to the system may be generated by outside sources or applications. A set of importers and adapters receive the information and process the information for use in the digital twin model.
G06F 30/13 - Conception architecturale, p. ex. conception architecturale assistée par ordinateur [CAAO] relative à la conception de bâtiments, de ponts, de paysages, d’usines ou de routes
G06F 30/20 - Optimisation, vérification ou simulation de l’objet conçu
15.
AUTOMATED AERIAL DATA CAPTURE FOR 3D MODELING OF UNKNOWN OBJECTS IN UNKNOWN ENVIRONMENTS
System and method are disclosed for multi-phase process of automated data capture for photogrammetry and 3D model building of an unknown object (311) in an unknown environment. Planner module (152) generates a flight plan (413) for a camera drone (110) to fly autonomously on a flight path along a virtual polygon grid (302) defined above the target object (311) during a survey phase. Model builder computer (153) receives a point cloud dataset (321) captured by LiDAR sensor on camera drone (301) during survey flight and constructs low resolution 3D mesh (331) of the target object (311). Planner module (152) generates a flight path (413) for camera drone inspection phase with virtual waypoints surrounding the target object (311) at a marginal distance from the surface defined by the low resolution 3D mesh (331). Model builder (153, 163) builds a high resolution 3D model (422) of the target object (311) using photogrammetry processing of high resolution images captured by camera drone (411, 412) during inspection phase.
G06T 17/20 - Description filaire, p. ex. polygonalisation ou tessellation
G01C 11/02 - Dispositions de prises de vues spécialement adaptées pour la photogrammétrie ou les levers photographiques, p. ex. pour commander le recouvrement des photos
G05D 1/689 - Interaction avec des charges utiles ou des entités externes dirigeant des charges utiles vers des cibles fixes ou en mouvement
Current approaches to controlling robots from multiple vendors typically requires multiple software systems that define vendor-exclusive fleet manager or dispatch systems. Autonomous devices (e.g., robots, drones, vehicles) can be controlled from multiple vendors that use multiple locally sourced map. For example, maps from individual robots can be translated to a base map that can be used to command and control hybrid fleets of robots.
G05D 1/86 - Surveillance des performances du système, p. ex. modules d’alarme ou de diagnostic
G05D 1/246 - Dispositions pour déterminer la position ou l’orientation utilisant des cartes d’environnement, p. ex. localisation et cartographie simultanées [SLAM]
G05D 1/692 - Commande coordonnée de la position ou du cap de plusieurs véhicules impliquant une pluralité de véhicules disparates
17.
SYSTEM AND METHOD FOR TRAINING A PHYSICS INFORMED NEURAL NETWORK WITH REYNOLDS AVERAGED NAVIER STOKES FORMULATION OF TURBULENT FLOWS
Physics informed neural networks (PINNs) can exploit automatic differentiation and incorporate the underlying PDEs to approximate a solution field. PINNs combine differential equations, such as compressible and incompressible Navier-Stokes equations, with experimental data or high-fidelity numerical simulations. Conventional approaches to training PINNs involve introducing data and PDE losses simultaneously at the start of the training phase, often with equal weight multipliers. It is recognized herein, however, that such training approaches often result in noisy training losses, slow convergence, and high validation error.
G06F 30/28 - Optimisation, vérification ou simulation de l’objet conçu utilisant la dynamique des fluides, p. ex. les équations de Navier-Stokes ou la dynamique des fluides numérique [DFN]
G06N 3/044 - Réseaux récurrents, p. ex. réseaux de Hopfield
Current approaches to obstacle detection can result in some obstacles being undetected, for instance obstacles that are higher or lower than the height of a two-dimensional (2D) laser, or obstacles defining particular thicknesses, thereby resulting in a collision or a risk of collision between the robot or autonomous device and the undetected obstacle. Using a three- dimensional (3D) camera or 3D sensor input, such as a depth camera, an automated guided vehicle (AGV), drone, or robot can detect obstacles at various heights along various trajectories of the robot, such that the robot can avoid collisions with objects in various unknown and dynamic environments, or use the extra data information for navigation and path planning.
A system for vision-based inspection of manufactured parts on a production line includes an image renderer that generates multiple synthetic camera images of a part utilizing a CAD model of the part by simulating the inspection environment and the one or more cameras based on randomization of a number of operational and environmental parameters. An image labeler generates annotations i the synthetic camera images by automatically detecting nominal surface features on the part in the synthetic camera images. An image blender randomly superimposes background textures and/or artifacts on a part surface, which are obtained from a set of real images, on the synthetic camera images. Subsequently, a model trainer uses synthetic camera images with the generated annotations to train a machine learning model for detecting nominal surface features on real images of manufactured parts on the production line.
G06V 10/70 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique
20.
DETERMINING LOCATION AND SIZING OF A NEW POWER UNIT WITHIN A CURRENT SYSTEM ARCHITECTURE OF A POWER SYSTEM OR A GRID
A system determines a location and a size of a new power generation or power regulating unit within a current system architecture of a power system including a plurality of power generation units. The system comprises a controller including a processor and a memory, computer-readable logic code stored in the memory which, when executed by the processor, causes the controller to execute a hybrid algorithm as a combination of a data-driven algorithm and a model-based algorithm to determine an optimal location and size of the new power generation or power regulating unit. The data-driven algorithm encodes a location and a size information. The controller to enable the model-based algorithm to optimize performance of a selected location and size of the new power generation or power regulating unit, which is based on a linearized system or a nonlinear system to provide guidance for the data-driven algorithm to incorporate physical rules and verify a new system architecture.
G06F 30/18 - Conception de réseaux, p. ex. conception basée sur les aspects topologiques ou d’interconnexion des systèmes d’approvisionnement en eau, électricité ou gaz, de tuyauterie, de chauffage, ventilation et climatisation [CVC], ou de systèmes de câblage
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
G06F 111/06 - Optimisation multi-objectif, p. ex. optimisation de Pareto utilisant le recuit simulé, les algorithmes de colonies de fourmis ou les algorithmes génétiques
System and method of classification of manufacturing defects and anomaly detection based on a neural network model applying physical measurement data. The system includes a graph engine and a graph neural network (GNN). The graph engine generates a graph of measurement points, a measurement matrix, and weight matrix of connected node distances. The GNN model includes an encoder that encodes the measurement matrix and the weight matrix to generate a latent node representation, from which a decoder determines node reconstructions. Unsupervised training converges by minimizing node reconstruction loss, the reconstruction performed using a decoder that decodes the latent node representation. Supervised training converges by minimizing reconstruction of binary labels of annotated measurement inputs, the reconstruction performed by a softmax layer that translates the latent node representation.
A method for implementing a real-time controller for an automation system includes starting a code generator to transform a model representation of a controller into controller code for execution on a target device operating on an open real-time platform. A configuration specification is read by the code generator, which includes data definition mapped to an input and output in the model representation of the controller and execution definition specifying a real-time requirement for execution on the target device. The configuration specification is parsed by the code generator to generate interface code configured to use an API provided by an inter-app communication layer in the target device based on the data definition, and to use an API provided by a real-time execution service running on the target device based on the execution definition. The controller code and interface code are linked to generate a control app deployable on the target device.
A system includes two or more physics solvers to generate one or more physical field variables, and one or more objective values based on received design variables. Two or more sensitivity computational modules compute a sensitivity field based on the mesh and the physical field variables. A design optimizer module (240) generates an updated design comprising one or more new design variables by executing an optimization of the design domain based on the sensitivity field, the objective values, one or more optimization requirements, one or more design constraints, and one or more manufacturing constraints. An iterative topology optimization is performed by iterations of the design optimizer sending updated design variables to the two or more physics solvers and sensitivity computational modules and receiving new sensitivity fields from the two or more sensitivity modules based on the updated design variables.
Methods for implementing an automation system (152) and corresponding systems (100) and computer-readable mediums (126). A method includes starting (602) a system orchestrator (168) and reading (604) a system configuration definition (156) and system manifest (158). The method includes reading (606) app configuration definitions (156) and app manifests (158) for one or more apps (154) based on the system configuration definition (156). The method includes defining (610) app configuration definitions (156) for one or more apps (154) based on the system configuration definition (156) and executing (614) the one or more apps (154) based on the app configuration definitions (156) and the app manifests (158). The method includes controlling (618) at least one external physical device (570) based on the one or more executing apps (154).
G05B 19/042 - Commande à programme autre que la commande numérique, c.-à-d. dans des automatismes à séquence ou dans des automates à logique utilisant des processeurs numériques
25.
VARIATIONAL QUANTUM ALGORITHMS FOR ACOUSTICS SOLVER
A comprising a first quantum computing device having a quantum processing unit (QPU) and a second computing device having a central processing unit (CPU). An original quantum state of the system is estimated in the QPU and is decomposed into a set of expected values of a plurality of Pauli terms. Then, in the CPU the Pauli terms are summed. The result of the sum is optimized in the general-purpose computer to generate an updated set of parameters representative of the quantum state of the system. Iteratively the quantum state is updated to generate a lowest eigenvalue representing the base acoustic energy of the system. The summing of the plurality of Pauli terms represents a Hamiltonian of the system. The ground state of the acoustic energy may be calculated repeatedly over a frequency spectrum in the frequency domain.
System and method are disclosed for simulating manufacturing of a part by material removal. A simulation module executes a simulation of material removal by a cutting tool based on a sparse voxelization of a workpiece material and a discretized set of tool surface points. A voxelization module generates a sparse voxelization of the workpiece material and a discretization of tool surface points. A tool physics module computes reaction forces based on a set of marked tool surface points and marked material voxels and computes resulting temperature increase used to predict resulting tool wear over the course of a material removal step. The tool wear information is exported to the simulation module using a feedback loop.
A robotic visual and tactile inspection system can perform surface inspection to identify different defects on various large industrial components (e.g., different sizes, geometries, smoothness, textures, etc.) that are used in various industries (e.g., manufacturing, transportation, aerospace, defense, power, process industries, etc.). The system can conduct a visual inspection using a RGBD camera to detect defect areas. A vision and tactile guided robot control system can be configured to control a robot to perform a tactile inspection, thereby detecting and identifying various defects as well as various attributes of the defects in various surfaces.
A method for solving partial differential equations in a fluid dynamics problem with an incompressible fluid, includes in a central processing unit (CPU), performing a first set of calculations from the set of PDEs, in a quantum processing unit (QPU), performing a second set of calculations from the set of PDEs, and solving for the solution to the PDEs based in the solution of the second set of calculations from the QPU. The set of PDEs may be a Navier Stokes or Euler equation. The second set of calculations performed on the QPU may calculate a pressure vector for the incompressible fluid. The pressure gradient may be calculated based on a Poisson problem. The Poisson problem may be solved using a Hamiltonian simulation or by using a Variational Quantum Algorithm (VQA). The second set of calculations may be performed in an external kernel routine executed by the QPU.
AI systems can define industrial AI models that perform, for example and without limitation, pattern recognition, trend prediction, or deviation identification. Furthermore, embodiments can automatically identify immediate and responsive actions to the detected deviations, and can automatically verify whether the identified actions are successful. Without being bound by theory, but by way of example, in some cases, embodiments described herein enable operators to monitor AI applications and respond to deviations within minutes, whereas current approaches might require data scientists to monitor and respond, and such a response might take days.
A computer-implemented method for online calibration of power system model against a power system includes iteratively approximating the power system model, at sequential optimization steps, around a moving design point defined by parameter values of a set of calibration parameters of the power system model. At each optimization step, an approximated system model is used to transform a dynamic input signal into a model output signal, which is compared with measurement signals obtained from measurement devices installed in the power system that define an actual power system output signal generated in response to the dynamic input signal. Parameter values of the calibration parameters adjusted in a direction to minimize an error between the model output signal and the actual power system output signal. The power system model is calibrated against the power system based on resulting optimal values of the calibration parameters.
A computer-implemented method for building an object detection module uses mesh representations of objects belonging to specified object classes of interest to render images by a physics-based simulator. Each rendered image captures a simulated environment containing objects belonging to multiple object classes of interest placed in a bin or on a table. The rendered images are generated by randomizing a set of parameters by the simulator to render a range of simulated environments. The randomized parameters include environmental and sensor-based parameters. A label is generated for each rendered image, which includes a two-dimensional representation indicative of location and object classes of objects in that rendered image frame. Each rendered image and the respective label constitute a data sample of a synthetic training dataset. A deep learning model is trained using the synthetic training dataset to output object classes from an input image of a real-world physical environment.
G06V 10/774 - Génération d'ensembles de motifs de formationTraitement des caractéristiques d’images ou de vidéos dans les espaces de caractéristiquesDispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p. ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]Séparation aveugle de source méthodes de Bootstrap, p. ex. "bagging” ou “boosting”
G06V 10/764 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant la classification, p. ex. des objets vidéo
G06V 10/776 - ValidationÉvaluation des performances
G06V 20/70 - Étiquetage du contenu de scène, p. ex. en tirant des représentations syntaxiques ou sémantiques
It is recognized herein that current approaches to training deep neural networks to perform grasp computations lack capabilities and efficiencies related to determining grasps for diverse objects at different scales with different gripper configurations, among other shortcomings. An end-to end neural network model can be trained to that can accurately determine grasp candidates on diverse objects at different scales, for various different gripper configurations having different sizes.
System and method include a neural network-based model (403) trained to generalize a small number of simulations for predicting outcomes of similar but different tasks related to CFD. A first variation applies a multi-task learning approach using a multi-head output layer, each head trained with data of a unique CFD task simulation. A second variation applies meta-learning with bi-level optimization by minimization of PDE loss as well as initial and boundary condition losses. A third variation employs a task encoder (405) for assisting a surrogate model (403) in learning from simulated task datasets by encoding the task data to a parameter domain useful to the surrogate model (403).
G06F 30/17 - Conception mécanique paramétrique ou variationnelle
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
G06F 30/28 - Optimisation, vérification ou simulation de l’objet conçu utilisant la dynamique des fluides, p. ex. les équations de Navier-Stokes ou la dynamique des fluides numérique [DFN]
34.
NEURAL NETWORK-GUIDED SYNTHESIS OF MANUFACTURABLE GEOMETRIC TWINS
Computer-aided drafting application provide a visual representation of a design and may generate surfaces that are difficult to manufacture. A method of receiving an organic shape (211) generated in a CAD application utilizes a neural network to generate a geometric twin (209) of the organic shape (211) using basic geometric shapes that are easier to manufacture. The neural network includes multiple layers (210, 220, 230) where each layer receives a shape library and collection of shape operations (201, 203, 205). At each layer, additional shapes are added to an intermediate model (211, 221) to approximate a target organic shape. An output layer provides a geometric twin (209) that is compared to the original organic shape (211) with a loss function (213). The measured difference is used to iteratively generate geometric twins until an updated geometric twin (209) is produced that is within an acceptable tolerance of the original shape (211) and may be manufactured using available automated manufacturing processes.
G06F 30/17 - Conception mécanique paramétrique ou variationnelle
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
G06F 119/18 - Analyse de fabricabilité ou optimisation de fabricabilité
35.
MULTI-ASSET PLACEMENT AND SIZING FOR ROBUST OPERATION OF DISTRIBUTION SYSTEMS
A method for adding assets to a distribution network includes using a placement generation engine to generate discrete placements of assets to be added to the distribution network subject to asset-installation constraint(s). Each placement is defined by a mapping of an asset, from multiple assets of different sizes, to a placement location defined by a node or branch of the distribution network. Each placement is used to update an operational circuit model of the distribution network for tuning control parameters of one or more controllers of the distribution network for robust operation over a range of load and/or generation scenarios. A cost function is evaluated for each placement based on a simulated operation. Parameters of the placement generation engine are iteratively adjusted based on the evaluated cost functions to arrive at an optimal placement and sizing of assets to be added to the distribution network.
A system (100) for enabling trusted on-demand distributed manufacturing includes a first interface (120) configured to receive product data from a product data source (110-A, 110-B, 110-N), a second interface (130) configured to exchange manufacturing data from a manufacturing data source (140-A, 140-B, 140-N), a trust anchor module (170) configured via computer executable instructions to add a physically unclonable function (PUF) to a product, generate a non-fungible token (NFT) to represent the product on an NFT platform (172), a matching module (174) configured via computer executable instructions to match the product characteristics with requirements of a buyer and manufacturing capabilities of a producer, and a traceability module (176) to verify success of a manufacturing process.
G05B 19/4155 - Commande numérique [CN], c.-à-d. machines fonctionnant automatiquement, en particulier machines-outils, p. ex. dans un milieu de fabrication industriel, afin d'effectuer un positionnement, un mouvement ou des actions coordonnées au moyen de données d'un programme sous forme numérique caractérisée par le déroulement du programme, c.-à-d. le déroulement d'un programme de pièce ou le déroulement d'une fonction machine, p. ex. choix d'un programme
37.
HIGH-LEVEL SENSOR FUSION AND MULTI-CRITERIA DECISION MAKING FOR AUTONOMOUS BIN PICKING
In described embodiments of method for executing autonomous bin picking, a physical environment comprising a bin containing a plurality of objects is perceived by one or more sensors. Multiple artificial intelligence (AI) modules feed from the sensors to compute grasping alternatives, and in some embodiments, detected objects of interest. Grasping alternatives and their attributes are computed based on the outputs of the AI modules in a high-level sensor fusion (HLSF) module. A multi-criteria decision making (MCDM) module is used to rank the grasping alternatives and select the one that maximizes the application utility while satisfying specified constraints.
G06T 7/73 - Détermination de la position ou de l'orientation des objets ou des caméras utilisant des procédés basés sur les caractéristiques
G06V 10/764 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant la classification, p. ex. des objets vidéo
G06V 10/80 - Fusion, c.-à-d. combinaison des données de diverses sources au niveau du capteur, du prétraitement, de l’extraction des caractéristiques ou de la classification
G06V 10/82 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant les réseaux neuronaux
A computer-implemented method for automatically creating a digital twin of an industrial system having one or more devices includes accessing a triple store that includes an aggregated ontology of graph-based industrial data synchronized with the one or more devices. The triple store is queried for a specified device to extract, from the graph-based industrial data, structural information of the specified device defined by a tree comprising a hierarchy of nodes. For each node, a neural network element is assigned based on a mapping of node types to pre-defined neural network elements. The assigned neural network elements are combined based on the tree topology to create a digital twin neural network. The triple store is then queried to extract, form the graph-based industrial data, real-time process data gathered from the specified device at runtime and use the extracted real-time process data to tune parameters of the digital twin neural network.
G05B 13/02 - Systèmes de commande adaptatifs, c.-à-d. systèmes se réglant eux-mêmes automatiquement pour obtenir un rendement optimal suivant un critère prédéterminé électriques
39.
Method of Producing and a Photonic Metasurface for Performing Computationally Intensive Mathematical Computations
The Penn State University-College of Earth & Mineral Sciences (USA)
Inventeur(s)
Chi, Heng
Xu, Huijuan
Rodriguez, Alejandro
Houyou, Mohamed El Amine
Reinhart, Wesley
Molesky, Sean
Chao, Pengning
Abrégé
According to aspects of embodiments described herein, an optical computing device (100) comprises a plurality of input waveguides (101), a photonic meta-surface (103) in contact with the plurality of input waveguides, and a plurality of output waveguides (105) in contact with the transformational meta-surface. The optical computing device may be configured to perform a mathematical operation may be a matrix multiplication. A computer-implemented method (300) of designing an optical computing device includes a plurality of input waveguides, a photonic meta-surface, and a plurality of output waveguides, the method includes exciting each input waveguide one-by-one (303) and measuring the energy at the input region and the output region (305) to determine a contribution of the current input waveguide. The sum of contributions (307) of all input waveguides are compared to a target transformation (315) to determine a loss value used to update a set of design parameters (317).
G06E 1/04 - Dispositions pour traiter exclusivement des données numériques agissant sur l'ordre ou le contenu des données maniées pour effectuer des calculs en utilisant exclusivement une représentation numérique codée, p. ex. représentation binaire, ternaire, décimale
G02B 1/00 - Éléments optiques caractérisés par la substance dont ils sont faitsRevêtements optiques pour éléments optiques
G02B 27/00 - Systèmes ou appareils optiques non prévus dans aucun des groupes ,
40.
MULTI-CORE PROCESSING WITH SOFTWARE-BASED SCHEDULER AND PLANNER
A multi-core processing system includes a processing unit having a plurality of distinct processing cores, and a memory coupled to the processing unit that includes instructions defining modules executable by the processing unit. The modules include an application program configured to execute one or more processes at runtime, where each process is defined by one or more user-level threads. The modules further include a software scheduler configured to schedule execution of the user-level threads on the processing cores. The software scheduler accesses a hardware limit metric and/or a real time performance metric associated with individual processing cores monitored during runtime. The software scheduler dynamically allocates processing cores to the user-level threads based on the monitored hardware limit metric and/or the real time performance metric associated with the individual processing cores. The user-level threads are dispatched for execution on respective processing cores based on the allocation.
System and method ensure stability of power transmission system or power grid with mixed generation sources of renewable and non-renewable types while providing a best combination of stability services that minimizes both cost of generation and guaranteed reliability. An energy market management (EMM) module derives an energy market dispatch for a mix of generation sources including renewable and conventional types based on a steady state optimization of supplier cost and a power demand forecast. Dynamic Security Assessment (DSA) module performs a dynamic contingency assessment of stability for the power transmission system based on the energy market dispatch and sends feedback to the EMM module with results of the dynamic assessment of stability. EMM module clears the energy market dispatch on a condition that feedback indicates a stable dynamic assessment. For unstable assessment, data driven, model- based and/or rule-based algorithms provide recommendations on how EMM module can find stable market solutions.
SIEMENS INDUSTRY SOFTWARE NETHERLANDS B.V. (Pays‑Bas)
SIEMENS CORPORATION (USA)
Inventeur(s)
Valenzuela Del Rio, Jose
Mirabella, Lucia
Motheau, Emmanuel
Arvanitis, Elena
Lancashire, Richard
Chatrath, Karan
Ritmeijer, Peter
Abrégé
A method for modeling airbag designs includes training a reduced order parametric model (104) of an airbag system for evaluating different airbag designs. Training the model includes applying a high-order multi-physics model (102) to an airbag system in the design space to produce snapshots (103) of performance over time. A reduced order model (104) constructs reduced bases (105) from the high-order snapshots (103) and a set of modal coefficients (107) are projected. A fit regression model (108) produces the parametric model (109). The parametric model is built and trained in an offline process, while the evaluating of an airbag design is done online. The reduced order model may be created using proper orthogonal decomposition a greedy algorithm and the regression model may be implemented in a machine learning model or in a Gaussian Process. Evaluating an airbag design uses parameters describing an airbag, or an environment where the airbag operates to the parametric model.
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
B60R 21/00 - Dispositions ou équipements sur les véhicules pour protéger les occupants ou les piétons ou pour leur éviter des blessures en cas d'accidents ou d'autres incidents dus à la circulation
To implement a joint detection, localization, segmentation and classification of anomalies in images, an autoencoder is trained on images in a training dataset based on an attention loss, which encourages the encoder network to generate attention maps maximizing focus on normal regions of images. An image segmentation model is trained by extracting feature representations using the encoder network from images in the training dataset, and alternatively optimizing cluster labels of pixels in a forward process and optimizing the feature representations by backpropagation of a segmentation error. The encoder network is then tuned in a few-shot learning process for classifying anomalies by extracting feature representations using the encoder network from anomalous images in support and query sets created from the training dataset, using support set feature representations to learn embeddings of anomaly classes, and updating the encoder network based on a classification error computed using query set feature representations.
Recovering symbols from a stripped binary includes representing the stripped binary as a plurality of graph of graphs (GoG) representations, converting the plurality of GoGs into a plurality of expressive representations of each function in the stripped binary, training a machine learning (ML) model using the expressive representations, and determining a missing symbol of at least one of the functions based on an output of the ML model. Information relating to functions is and interactions between functions are used to train an ML to determine similarity of two functions. Based on similarities, missing symbols may be inferred and used to enable updates to the stripped binary file where source code is not available.
A computer-implemented method for online calibration of power system model against a power system includes iteratively approximating the power system model to a time-domain system model, at sequential optimization steps, as a function of parameter values of a set of system parameters. At each optimization step, an error is measured between a time series of a model output in response to the dynamic input signal and a time series of measurement signals obtained from the measurement devices defining an actual power system response to the dynamic input signal, summed over a number of discretized points in time with a defined sampling interval. A sequential optimizer is used to adjust parameter values of the calibration parameters and a system state to minimize the measured error, constrained by a discretization of the time-domain system model based on the sampling interval, to thereby determine optimal values of the calibration parameters.
H02J 13/00 - Circuits pour pourvoir à l'indication à distance des conditions d'un réseau, p. ex. un enregistrement instantané des conditions d'ouverture ou de fermeture de chaque sectionneur du réseauCircuits pour pourvoir à la commande à distance des moyens de commutation dans un réseau de distribution d'énergie, p. ex. mise en ou hors circuit de consommateurs de courant par l'utilisation de signaux d'impulsion codés transmis par le réseau
46.
SYSTEMS AND METHODS FOR OPERATION OF ELECTRICAL INVERTERS
Methods for controlling an inverter performed by an inverter control system and related systems and media. A method (300) includes receiving (302) a forecasted active power (102) and a reactive power profile (122) for an inverter (195), and determining (304), based on the forecasted active power (102) and the reactive power profile (122), a power loss prediction (106) over time for the inverter (195). The method includes determining (302) a temperature prediction (110) over time for the inverter (195) and performing (308) a rainflow analysis process (112) to produce temperature fatigue cycle data (114) for the inverter (195). The method includes determining (310) a damage prediction (118) over time for the inverter (195) and determining (312) a reactive power profile (122) over time for the inverter (195). The inverter (195) can thereafter be controlled (314) to produce power according to the reactive power profile (122).
Methods for performing a black start in a power system and corresponding systems. A method (200) includes starting (100) a first anchor grid-forming inverter (122) in a first microgrid (110) and starting (100) a second anchor grid-forming inverter (122) in a second microgrid (120). The first microgrid (110) is connected to a first bus (118) and the second microgrid (120) is connected to a second bus (128). The method includes measuring (202) a first voltage at the first bus (118) and measuring a second voltage at the second bus (128), determining (204) a voltage difference between the first bus (118) and the second bus (128), and determining (204) whether the voltage difference is within a voltage difference threshold (350). The method includes measuring (206) a first phase angle at the first bus (118) and measuring a second phase angle at the second bus (128), determining (208) a phase angle difference between the first bus (118) and the second bus (128), and determining (208) whether the phase angle difference is within a phase angle difference threshold (360). The method includes, when the voltage difference is within the voltage difference threshold (350) and the phase angle difference is within the phase angle difference threshold (360), then operatively connecting (212) the first bus (118) to the second bus (128).
H02J 3/38 - Dispositions pour l’alimentation en parallèle d’un seul réseau, par plusieurs générateurs, convertisseurs ou transformateurs
H02J 3/44 - Synchronisation d'un générateur pour sa connexion à un réseau ou à un autre générateur avec moyens pour assurer une séquence de phase correcte
48.
MODULAR METHOD FOR EXTENDING 5G SERVICES WITH ZERO TRUST SECURITY
Applying cybersecurity to a network includes creating a modular implementation of CISA's Zero-Trust maturity model and extending a fifth-generation (5G) core. Each module corresponds to one of the five pillars in the CISA Zero-Trust maturity model. A first module verifies and enforces the access of a user based on the user's identity. A second module maintains a complete inventory of every authorized device and prevents, detects and responds to incidents involving the authorized devices. A third module protects the network by encrypting all DNS and HTTP traffic, isolating traffic flows and monitoring user activities. A fourth module directs to applications and workload of the network to test applications and generate vulnerability reports periodically. A fifth module for monitoring data on the network by classifying, encrypting all data and maintaining logs of every access to data.
A computer-implemented method and system are provided for composition of an adaptive model to enhance a physics-based baseline simulation model of a physical system. An adaptive model is configured as a statistical model trained with real input and output signals of a modified physical system and simulated input and output signals of a physics-based baseline simulation model of the physical system. The baseline simulation model is based on the physical system prior to modification and runs in tandem with the trained adaptive model. The adaptive model is embedded and executed in an edge computing device that receives the real input and output signals.
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
50.
SYNCHRONOUS USER AUTHENTICATION AND PERSONALIZATION OF HUMAN-MACHINE INTERFACES
A human-machine interface (HMI) system defines a camera and a plurality of modules. The HMI system can capture a video feed of a production area in vicinity of the HMI system, so as to define an input video. The HMI system can detect a face in the input video, so as to define a detected face. Furthermore, the HMI system can determine whether the detected face in the input video represents a spoof attack. When it is determined that the detected face does not represent the spoof attack, the HMI system can determine whether the face belongs to an authorized user of the HMI system. In an example, after determining that the face belongs to an authorized user of the HMI system, the HMI system can allow access to the production area corresponding to particular access rights of the authorized user.
Methods, systems, and apparatuses are configured to generate new industrial asset models that can be efficiently queried. The new industrial asset models can be tailored to specific users and represented in an RDF knowledge graph. In particular, the RDF graph can define variables that correspond to OPC variables. The RDF knowledge graph queried via a query that is compliant with a GraphQL API.
Methods and systems are disclosed for generating, maintaining, and tracking programmatically accessible information related to safety requirements for a system design. For example, a safety computing system can automatically generate reports and continuously update such reports, so as to define the current state of safety engineering at any given time, automatically or on-demand.
A system can define semantic markers that denote or indicate where various objects can be placed, how various objects can be assembled, or where various objects can be gripped by a robot. An example user interface can indicate to users how objects and grippers are intended to be placed or positioned.
G06F 30/12 - CAO géométrique caractérisée par des moyens d’entrée spécialement adaptés à la CAO, p. ex. interfaces utilisateur graphiques [UIG] spécialement adaptées à la CAO
G05B 19/418 - Commande totale d'usine, c.-à-d. commande centralisée de plusieurs machines, p. ex. commande numérique directe ou distribuée [DNC], systèmes d'ateliers flexibles [FMS], systèmes de fabrication intégrés [IMS], productique [CIM]
Programmable logic controllers (PLCs) are monitored for cybersecurity through leveraging a connection-oriented network protocol. Connection-oriented network protocol link is established between background service and each of a plurality of PLCs. PLC generates a constant cyclical pulse signal having frequency set to align with scan cycles of control program executed by the PLC. Process control modules of control program are executed to control process facility field devices. PLC sends a series of heartbeat messages on the link synchronized to the pulse signal, each heartbeat message containing a time value at which the heartbeat message occurs. Background service monitors the heartbeat messages from each PLC, detects potential cyberattack incident for PLC in response to any interruption of heartbeat messages from PLC, and sends an alert to upper control layer analysis unit to analyze cause for the interruption of heartbeat messages.
G05B 19/05 - Automates à logique programmables, p. ex. simulant les interconnexions logiques de signaux d'après des diagrammes en échelle ou des organigrammes
G06F 21/55 - Détection d’intrusion locale ou mise en œuvre de contre-mesures
55.
A SYSTEM FOR OPTIMIZATION OF COMPLEX SYSTEMS USING DATA-DRIVEN MODELING OF CROSS-DISCIPLINE INTERACTION
A method of optimizing a multi-discipline system by constructing a multi-disciplinary model built by discovering a first latent space of a first discipline then discovering at least a second latent space of at least a. The multi-disciplinary model contains information relating to common characteristics and interactions across multiple disciplines. Inputting a desired response for one of the first and second discipline to the multi-disciplinary model will generate an output representative of a set of design parameters for the first discipline and the second discipline. In an embodiment, a field from a first discipline may be used as input to the multi-disciplinary model, which generates a second field from a discipline other than the first discipline. A desired response for input to the multi-discipline model may be received from a user via a user interface.
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
An automated model based guided digital twin synchronization system is described. The system comprises visual sensors configured to acquire raw visual data of a physical 3D scene content from a real site, a database to provide a 3D model of the physical 3D scene content, a processor and a memory for storing computer-executable instructions executed by the processor. The instructions comprise an automated machine learning model based logic to: generate and maintain an as-built digital twin of the assets and large‐scale infrastructures present in the physical 3D scene by ingesting the raw visual data to a common and binding structured representation, compare the as-built digital twin representation obtained from the real site against corresponding an as- planned digital twin to determine spatio‐temporal differences between the as-built and the as-planned digital twins, and update automatically the as-planned digital twin to reflect any changes to the physical 3D scene based on the spatio‐temporal differences.
A computer-implemented method of engineering design includes performing a simulation for a design under test of the design interacting in a real-world environment, in a computer processor, generating simulation data from performance of the simulation, storing the generated simulation data in a computer memory, extracting the generated simulation data to train a first inverse model neural network, and generating a plurality of design parameters from the inverse model neural network. A visualization representative of the generated plurality of design parameters for display to a user allows an expert user to evaluate the suggested design parameters and select some or all of the suggested design parameters. For existing designs more data is available for the simulation than for a new design. The inverse model for a new design may be augmented by transferring stored knowledge in a pre-existing inverse model to the model for the new design.
A computer-implemented method and system are provided for modeling a nonlinear system having on-demand nonlinear components. Controller design engine generates an initial nonlinear system model having system components and controllers represented as interconnected system blocks. A system component assessment identifies which system blocks are linear blocks and nonlinear blocks. A linear component engine linearizes each nonlinear block by performing a linear system approximation. A simulation of the system is performed using inputs and parameters for an expected operating range. Response of each system block is monitored for whether one or more output values exhibit an active nonlinearity. Nonlinear component engine converts linear blocks to nonlinear blocks for any linear blocks with a detected active nonlinearity. The process iterates until no new active nonlinearity is detected.
G05B 13/04 - Systèmes de commande adaptatifs, c.-à-d. systèmes se réglant eux-mêmes automatiquement pour obtenir un rendement optimal suivant un critère prédéterminé électriques impliquant l'usage de modèles ou de simulateurs
G05B 17/02 - Systèmes impliquant l'usage de modèles ou de simulateurs desdits systèmes électriques
59.
LARGE-SCALE BATTERY ENERGY STORAGE SYSTEM SITING AND SIZING FOR PARTICIPATION IN WHOLESALE ENERGY MARKET USING HYPERPARAMETER OPTIMIZATION
According to a method to aid installation of large-scale BESS in a power network, a hyperparameter optimization engine generates a feasible configuration of BESS defined by location and sizing parameters subject to an installation constraint. A power system simulation engine conducts an energy market simulation for the power network with the generated configuration over a defined simulation horizon to determine a configuration value. The power system simulation engine comprises one or more subroutines characterizing the impact of the BESS on the market clearing mechanism, to compute an expected generation cost associated with the BESS. The configuration value is determined based on the expected generation cost and a total installation cost of the BESS. The hyperparameter optimization engine is iteratively executed to generate an updated configuration based on the configuration values of previous configurations, to determine a final configuration of BESS defined by a set of optimal location and sizing parameters.
H02M 7/66 - Transformation d'une puissance d'entrée en courant alternatif en une puissance de sortie en courant continuTransformation d'une puissance d'entrée en courant continu en une puissance de sortie en courant alternatif avec possibilité de réversibilité
60.
INDUSTRIAL TASKS SUPPORTED BY AUGMENTED VIRTUALITY
A method of creating instructions for performing an industrial task in an augmented reality (AR) application includes querying a knowledge base constructed from existing data resources relating to the industrial task. An instruction set for the industrial task is generated and formatted as input for an AR application. The formatted instructions correspond to objects represented in a CAD model. Information relating to the formatted instructions may be displayed along the objects from the CAD model file. Objects represented in the CAD model file may correspond to real-world objects in the user's field of view. In an AR application, the formatted instructions may be overlaid on a real-world object in the user's field of view. The order of relevant objects to display from the CAD model may be obtained by the AR application by constructing the knowledge base in part on the CAD model file and relationships about the components.
G06F 9/44 - Dispositions pour exécuter des programmes spécifiques
G06F 3/04815 - Interaction s’effectuant dans un environnement basé sur des métaphores ou des objets avec un affichage tridimensionnel, p. ex. modification du point de vue de l’utilisateur par rapport à l’environnement ou l’objet
G06F 30/00 - Conception assistée par ordinateur [CAO]
G06T 19/00 - Transformation de modèles ou d'images tridimensionnels [3D] pour infographie
G06V 20/00 - ScènesÉléments spécifiques à la scène
G06V 20/20 - ScènesÉléments spécifiques à la scène dans les scènes de réalité augmentée
61.
METHOD FOR GENERATING INSTRUCTIONS USING 3D MODEL COMPONENT RELATIONSHIP EXTRACTION
A method of generating instructions in an industrial setting comprising analyzing 3D/CAD file and other formats extracting information relating to an industrial task; aligning the information to a knowledge graph containing ontological and other semantic information associated with the industrial task; inferring from reasoned geometric relationships the steps of task execution with refined step ordering; applying knowledge to task ordering to infer instruction order fine-tuning; and generating instructions for performing the task(s) providing steps for target entities in AR, image, audio, video, and/or automated sentence format. The information may be extracted using natural language processing, deep/machine learning, or other artificial intelligence techniques including pattern recognition representative of instructions having inherent ground truth validation, correct by design based on engineering specifications. To identify semantic information relating to the industrial task, action words and other relevant entities may be extracted from various data sources such as from text and image tagging.
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
G06Q 10/0832 - Marchandises spéciales ou procédures de manutention spéciales, p. ex. manutention de marchandises dangereuses ou fragiles
62.
METHOD OF AUTOMATED CREATION OF GRAPHQL AND API SERVER FOR A OPC UA DERIVED RDF GRAPH
A computer implemented method for extracting information from an OPC UA specification represented in a RDF knowledge graph includes presenting a query compliant with a GraphQL API in a GraphQL server, receiving the GraphQL query; in the GraphQL server, converting the GraphQL query to a query compliant with SPARQL, extracting information from the RDF knowledge graph using the SPARQL query, and formatting the result of the SPARQL query in a format desired by the user. Additionally in some embodiments, the GraphQL server comprises a GraphQL schema containing information identifying objects and relationships between objects, the GraphQL schema representative of the OPC UA specification; and a GraphQL resolver for receiving a GraphQL query and extracting requested information from the RDF knowledge graph. According to an embodiment, the format desired by the user is javascript object notation (JSON).
In a unified data management system for industrial automation, a query request is received via a unified query API layer configured to expose a common interface for accessing data from a persistent storage layer including multiple data stores that ingest data respectively from multiple data sources in an automation system. The data stores include different access APIs and data schema. Query parameters are extracted from the unified query API layer to build a dedicated query structured for a specific data store by a query builder utilizing a semantic layer. The semantic layer can map queries to individual data stores based on context information derived from query parameters and provide access to individual data stores based on their respective access API and data schema. The dedicated query is executed directly on the specific data store by a query engine and a query result is returned via the unified query API layer.
Tools can generate semantic information that indicates the purpose and contents of messages that are transmitted on a given network. In particular, for example, forensic tools described herein can discriminate between security issues, bugs, performance limitations, user errors, and the like.
G06F 21/55 - Détection d’intrusion locale ou mise en œuvre de contre-mesures
H04L 43/062 - Génération de rapports liés au trafic du réseau
H04L 43/08 - Surveillance ou test en fonction de métriques spécifiques, p. ex. la qualité du service [QoS], la consommation d’énergie ou les paramètres environnementaux
H04L 43/0811 - Surveillance ou test en fonction de métriques spécifiques, p. ex. la qualité du service [QoS], la consommation d’énergie ou les paramètres environnementaux en vérifiant la disponibilité en vérifiant la connectivité
Robots might interact with planar objects (e.g., garments) for process automation, quality control, to perform sewing operations, or the like. It is recognized herein that robots interacting with such planar objects can pose particular problems, for instance problems related to grasping garments. An end effector can include a clamp configured to move from an open position to a closed position in which the clamp holds a planar object against a surface. The end effector can further include a sensor control system comprising at least one sensor. The sensor control system can be configured to detect when the planar object is disposed against the surface. Furthermore, responsive to detecting that the planar object is disposed against the surface, the sensor control system can move the clamp from the open position to the closed position, so as to press the planar object against the surface.
According to disclosed embodiments for configuring an image acquisition system with one or more cameras for vision-based inspection of parts on a production line, a simulation engine renders synthetic images of a part on the production line acquired by the one or more cameras, based on a 3D model of the part and a configuration of the image acquisition system defined by optimizable parameters. A surface coverage measurement engine uses an output of the simulation engine to measure blind spots on a part surface for individual cameras and therefrom determine a measure of visible surface coverage on the 3D model of the part. An optimization engine generates an updated configuration of the image acquisition system by updating the optimizable parameters based on evaluation of an optimization objective defined by the measured visible surface coverage. The above process is iteratively executed to determine a final configuration of the image acquisition system.
System and method enable development of an industrial metaverse used in engineering by computer aided design. A metaverse pipeline includes a plurality of metaverse entities including at least one design application and one simulation application configured to develop a virtual environment metaverse representation of an industrial facility. A metaverse knowledge graph describes semantics and relationships between the metaverse pipeline entities and is configured to exchange information across the software applications of the metaverse pipeline entities of various platforms while developing the virtual environment metaverse. The metaverse knowledge graph enables synchronization across metaverse development platforms, such that updates to one platform are transferrable, whereby the virtual entities and environment can be assimilated according to new edits.
Bin picking refers to a robot grasping objects that can define random or arbitrary poses, from a container or bin. The robot can move or transport the objects, and place them at a different location for packaging or further processing. It is recognized herein, however, that current approaches to robotic picking lack efficiency and capabilities. In particular, current approaches often do not properly or efficiently identify certain clearances associated with a given robot to execute various grasps, due to various technical challenges in doing so. During runtime of a robot, various clearance dimensions associated with the robot executing a grasp are determined. In particular, for example, during runtime the robot can determine trajectories for executing grasps of objects in bins without colliding with the bin, for instance walls of the bin.
System and method for synchronizing information model changes between hierarchical systems of a smart factory are disclosed. A model generator instantiates objects for an aggregated information model for each of a plurality of subsystems in the smart factory based on an OPC UA standardized model. A synchronization engine maintains a global timestamp variable for a last successful synchronization performed by a parent OPC UA server for syncing to a second aggregated information model instance stored in a lower level OPC UA server. Model nodes are annotated with a node-wise timestamp in response to a modification to the node. All nodes subject to modification are pushed to a priority queue as a serialized node set format having annotation extensions that include synchronization-relevant data. The first and second aggregated information model are synchronized by updating nodes with the synchronization-relevant data.
A computer-implemented method for failure classification of a surface treatment process includes receiving one or more process parameters that influence one or more failure modes of the surface treatment process and receiving sensor data pertaining to measurement of one or more process states pertaining to the surface treatment process. The method includes processing the received one or more process parameters and the sensor data by a machine learning model deployed on an edge computing device controlling the surface treatment process to generate an output indicating, in real-time, a probability of process failure via the one or more failure modes. The machine learning model is trained on a supervised learning regime based on process data and failure classification labels obtained from physics simulations of the surface treatment process in combination with historical data pertaining to the surface treatment process.
A system and method for analysis of potential cybersecurity threats in an engineered system combines a functional simulation of the system with a cyberattack information layer. The cyberattack information layer informs the simulation with respect to the effects of a potential cyberattack on devices and links in the system. An iterative AI-based sequential decision-making optimization identifies a sequence of attacker steps for carrying out a cyberattack which has the greatest impact on a key performance indicator relating to operation of the system. The cyberattack information layer includes information relating to a topology and devices in the system from a computer network perspective and includes information on an amount of effort required to carry out possible attacks affecting each device or communication link in the system. The decision-making optimization may be run iteratively between the simulation and the cyberattack information layer to find a most impactful sequence of attacker actions.
In some cases, images and depth maps can define bins with objects in random configurations. It is recognized herein that current approaches to training deep neural networks to perform grasp computations lack capabilities and efficiencies, such that the resulting grasp computations and grasps can be imprecise or cumbersome, among other shortcomings. Synthetic depth images can be labeled with grasp annotations that are generated based on heuristic-based analyses, so as to define annotated synthetic datasets. The annotated synthetic datasets can be used to train neural networks to determine the best grasp locations for different objects arranged in a variety of positions with respect to each other.
An AI-based method for visual inspection of parts manufactured on a shop floor includes acquiring a set of real images of nominal parts manufactured on the shop floor to create training datasets. A self-supervised pre-trainer module is used to pre-train a loss computation neural network in a self-supervised learning process using a first dataset on pretexts defined by real-world conditions pertaining to the shop floor. The first dataset is labeled by automatically extracting pretext-related information from image metadata. A main anomaly trainer module is used to train a main anomaly detection neural network to reconstruct a nominal part image from an input manufactured part image in an unsupervised learning process using a second dataset. The main anomaly training measures a perceptual loss between an input image and a reconstructed image by measuring a difference between feature representations thereof at one or more layers of the pre-trained loss computation neural network.
A system (100) for reducing carbon dioxide emissions using trusted on-demand distributed manufacturing, with a processor (102) and a memory (104), includes a first interface (120) configured to receive product data from a product data source (110-A, 110-B, 110-N), a second interface (130) configured to exchange manufacturing data from a manufacturing data source (140-A, 140-B, 140-N), a matching module (174) configured via computer executable instructions to match the product data with the manufacturing data based on manufacturing characteristics for producing a product (154) as described in the product data, a carbon footprint module (180) configured via computer executable instructions to determine a traditional carbon footprint and an actual carbon footprint of the product (154), and determine a carbon dioxide reduction based on the traditional and the actual carbon footprint of the product (154), and wherein the system (100) is configured to automatically transmit the carbon dioxide reduction to a certification body (240).
G06Q 10/04 - Prévision ou optimisation spécialement adaptées à des fins administratives ou de gestion, p. ex. programmation linéaire ou "problème d’optimisation des stocks"
G06Q 10/06 - Ressources, gestion de tâches, des ressources humaines ou de projetsPlanification d’entreprise ou d’organisationModélisation d’entreprise ou d’organisation
A method for exploring a design space for a design includes, establishing a baseline geometric design and creating a baseline mesh based on the baseline geometric design (101), introducing a new geometric design (111) and creating a new geometric design mesh which is compatible with the baseline mesh by morphing the latter mesh (123). A design solution is generated based on the compatible mesh to create a snapshot solution (127). Snapshots are then used to train and create a reduced order model, ROM (130). Once the ROM is created, a new design alternative may be analysed by applying a set of design parameters to the ROM. A compatible mesh may be understood to have the same topology and dimensions of the baseline mesh. The process for morphing the new design meshes and creating the ROM may be implemented as an add-on workflow (120) in cooperation with a computer-aided engineering, CAE, application.
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
G06F 30/23 - Optimisation, vérification ou simulation de l’objet conçu utilisant les méthodes des éléments finis [MEF] ou les méthodes à différences finies [MDF]
G06F 30/28 - Optimisation, vérification ou simulation de l’objet conçu utilisant la dynamique des fluides, p. ex. les équations de Navier-Stokes ou la dynamique des fluides numérique [DFN]
A grid-forming photovoltaic (PV) system and method for both islanded connection and grid-connected mode is provided. An inverter converts PV array voltage to a voltage usable as a power source to an electric power system load. Active power-frequency droop controller regulates a modulator that generates drive signals to drive the inverter. Proportional integral controller generates a frequency shift value that adjusts output of the active power-frequency droop controller to yield a phase angle control for modulation of the inverter drive signals. A control mode switch selects among a plurality of control modes for operation of the proportional integral controller. A model-free control algorithm controls the control mode switch, including a control mode (221b) for synchronizing the PV system with the grid in which proportional integral controller (a) detects offset between inverter output voltage and grid output voltage and (b) generates the frequency shift value.
System and method optimize recyclability of an electronic device during manufacturing design A manufacturing design software produces engineering bill of materials, manufacturing bill of materials, and bill of process for the manufacturing design. Recycling process plan engine constructs a recycling process plan for the electronic device according to the manufacturing design. A virtual model of recycling processes is constructed by mapping needed skills to corresponding recycling equipment using a library of recycling equipment information. Recycling process plan engine uses the virtual model to simulate the recycling plan and optimizes each recycling process, and the overall sequence of recycling processes, according to an objective function. Evaluator module receives key performance indicator values from a virtual model simulation and calculates the value of the objective function based on key performance indicators. Recycling plan is optimized by iteration of simulating the process and varying parameters to most improve the objective function.
System and method simulate power distribution system reconfigurations for multiple contingencies. Decision tree model is instantiated as a graph with nodes and edges corresponding to simulated outage states of one or more buses in the power distribution system and simulated states of reconfigurable switches in the power distribution system, Edges related to each outage are disconnected. A reconfiguration path is determined with a plurality of switches reconfigured to a closed state by an iteration of tree search algorithms. A simulation estimates feeder cable and transformer loading and bus voltages on the reconfigured path for comparing against constraints including system capacity ratings and minimum voltage. Further iterations identify additional candidate reconfiguration paths which can be ranked by total load restoration
An edge device can be configured to perform industrial control operations within a production environment that defines a physical location. The edge device can include a plurality of neural network layers that define a deep neural network. The edge device be configured to obtain data from one or more sensors at the physical location defined by the production environment. The edge device can be further configured to perform one or more matrix operations on the data using the plurality of neural network layers so as to generate a large scale matrix computation at the physical location defined by the production environment. In some examples, the edge device can send the large scale matrix computation to a digital twin simulation model associated with the production environment, so as to update the digital twin simulation model in real time.
Dynamic slicing technology can be performed to assign network resources for various robotic repairing subtasks having different priorities, while satisfying various real-time and high throughput requirements. In some cases, M/M/1 queuing theory is applied so as to efficiently improve the utilization of network resources, reduce queuing latency and propagation latency, and balance the load among different network slicing, so as to perform robotic repairing operations.
In an example aspect, a first object (e.g., an electronic component) is inserted by a robot into a second object (e.g., a PCB). An autonomous system can capture a first image of the first object within a physical environment. The first object can define a mounting interface configured to insert into the second object. Based on the first image, a robot can grasp the first object within the physical environment. While the robot grasps the first object, the system can capture a second image of the first object. The second image can include the mounting interface of the first object. Based on the second image of the first object, the system can determine a grasp offset associated with the first object. The grasp offset can indicate movement associated with the robot grasping the first object within the physical environment. The system can also capture an image of the second object. Based on the grasp offset and the image of the second object, the robot can insert the first object into the second object.
A method for automating a process plant includes configuring field components of the process plant as a swarm of self-organizing agents where each agent communicates with adjacently physically connected agents. The method includes reacting on requests generated by requesting agents by responsible agents assigned based on a prioritization mechanism established using a topological representation of the process plant. The prioritization mechanism is established by assigning a penalty value to each request and forwarding the request via one or more paths using the topological representation by incrementing the penalty value at an agent that forwards or reacts on the request depending on the type of request and/or agent. Thereby, a penalty matrix is created indicating how much penalty would be incurred if a particular request is reacted on by a given responsible agent. The penalty matrix is used to assign requests to responsible agents to minimize a total penalty value.
A computer-implemented method for controlling an autonomous machine includes processing sensor data streamed via a plurality of calibrated sensors by a plurality of perception modules to extract perception information from the sensor data in real time. The extracted real time perception information from the plurality of perception modules is fused by a context awareness module to create a blackboard image, which is a representation of an operating environment of the autonomous machine derived from fusion of the extracted perception information using a controlled semantic, defining a context of the autonomous machine. A stream of blackboard images, representing a time evolving context of the autonomous machine, is processed by an action evaluation module, using a control policy, to output a control action to be executed by the autonomous machine. The control policy includes a learned mapping of context to control action represented by blackboard images created using the controlled semantic.
G06V 10/774 - Génération d'ensembles de motifs de formationTraitement des caractéristiques d’images ou de vidéos dans les espaces de caractéristiquesDispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant l’intégration et la réduction de données, p. ex. analyse en composantes principales [PCA] ou analyse en composantes indépendantes [ ICA] ou cartes auto-organisatrices [SOM]Séparation aveugle de source méthodes de Bootstrap, p. ex. "bagging” ou “boosting”
G06V 10/80 - Fusion, c.-à-d. combinaison des données de diverses sources au niveau du capteur, du prétraitement, de l’extraction des caractéristiques ou de la classification
G06V 10/82 - Dispositions pour la reconnaissance ou la compréhension d’images ou de vidéos utilisant la reconnaissance de formes ou l’apprentissage automatique utilisant les réseaux neuronaux
G06V 20/58 - Reconnaissance d’objets en mouvement ou d’obstacles, p. ex. véhicules ou piétonsReconnaissance des objets de la circulation, p. ex. signalisation routière, feux de signalisation ou routes
It is recognized herein that deep-learning approaches to anomaly detection can require a large amount of training data to properly learn the task. It is further recognized herein that capturing images of anomalies can be particularly costly or impractical or, in some cases, impossible. For example, by definition, anomalies can be rare and, therefore, gathering enough samples to train a convolutional neural network can be tedious. Annotating anomalies that are depicted can also be an expensive and time-consuming task. In various examples, realistic synthetic images are generated that include plausible and annotated surface defects (anomalies). Such synthetic images are used to train an efficient anomaly segmentation network in a fully supervised manner.
A computer-implemented method and system are provided for system reconfiguration of redundant components of an autonomous system. Failure mechanism modeling module (113) predicts future degradation states for each failure mode of the system components based on a current configuration of system components responsive to a selected reconfiguration action, mission requirement, failure mechanism parameters, and current degradation states. A set of mission requirement capability variables is defined by the predicted degradation states and current system configuration. A set of reward values is generated for time increments spanning the mission, each reward value quantifying a goal function that estimates current and future reward for a reconfiguration action being evaluated at a time instance. One or more reconfiguration actions associated with a maximum value function that sums up the estimated current and future reward are determined. Uncertainty of degradation states and degradation parameters is propagated as a complete probability density.
Robots might interact with planar objects (e.g., garments) for process automation, quality control, to perform sewing operations, or the like. It is recognized herein that robots interacting with such planar objects can pose particular problems, for instance problems related to detecting the planar object and estimating the pose of the detected planar object. A system can be configured to detect or segment planar objects, such as garments. The system can include a three-dimensional (3D) sensor positioned to detect a planar object along a transverse direction. The system can further include a first surface that supports the planar object. The first surface can be positioned such that the planar object is disposed between the first surface and the 3D sensor along the transverse direction. In various examples, the 3D sensor is configured to detect the planar object without detecting the first surface.
In various software applications, wizards can assist users in navigating through selected, predefined workflows for automating some repetitive interaction patterns. It is recognized herein, however, that such conventional wizards are generally limited to few use cases that are hard-wired in a given wizard's implementation, such that user still needs to master the respective tool's complexity. In an example aspect, a computing system within an automation system can determine a goal state associated with a target application. The system can extract an initial state from the target application. Based on knowledge obtained by the system, the system can generate a plurality of plans. The plans can define respective sequences of actions for reaching the goal state from the initial state. The system can render options on a user interface of the target application. Based on the options, the system receives selections related to the sequences of actions so as to define selected actions. The system can perform the selected actions until the goal state is reached.
G05B 15/02 - Systèmes commandés par un calculateur électriques
G06Q 10/06 - Ressources, gestion de tâches, des ressources humaines ou de projetsPlanification d’entreprise ou d’organisationModélisation d’entreprise ou d’organisation
88.
SYSTEM AND METHOD FOR ENABLING SCALABLE PROCESSING THROUGHPUT OF HIGH-VOLUME DATA ON INDUSTRIAL CONTROLLERS
A system for processing data in an industrial environment includes a sensor that communicates sensor data acquired from a factory floor as a multicast data stream where each frame is tagged with a sequence identifier indicative of a position of that frame in the data stream. A network switch distributes the data stream to a group of multicast endpoints. The system includes a controller with a plurality of modularly connected processing units. Each processing unit is further connected to the network switch and configured as a respective endpoint of the group of multicast endpoints. Each processing unit selectively processes frames of the data stream in dependence of the sequence identifier of individual frames by executing a data processing module, to produce an output result associated with each processed frame. The controller assembles the output result and the sequence identifier of each processed frame from the plurality of processing units.
H04L 12/28 - Réseaux de données à commutation caractérisés par la configuration des liaisons, p. ex. réseaux locaux [LAN Local Area Networks] ou réseaux étendus [WAN Wide Area Networks]
H04L 67/10 - Protocoles dans lesquels une application est distribuée parmi les nœuds du réseau
H04L 67/12 - Protocoles spécialement adaptés aux environnements propriétaires ou de mise en réseau pour un usage spécial, p. ex. les réseaux médicaux, les réseaux de capteurs, les réseaux dans les véhicules ou les réseaux de mesure à distance
H04L 67/289 - Traitement intermédiaire fonctionnellement situé à proximité de l'application consommatrice de données, p. ex. dans la même machine, dans le même domicile ou dans le même sous-réseau
H04L 67/5651 - Conversion ou adaptation du format ou du contenu d'applications en réduisant la quantité ou la taille des données d'application échangées
H04L 65/611 - Diffusion en flux de paquets multimédias pour la prise en charge des services de diffusion par flux unidirectionnel, p. ex. radio sur Internet pour la multidiffusion ou la diffusion
H04L 65/75 - Gestion des paquets du réseau multimédia
G05B 19/042 - Commande à programme autre que la commande numérique, c.-à-d. dans des automatismes à séquence ou dans des automates à logique utilisant des processeurs numériques
G05B 19/418 - Commande totale d'usine, c.-à-d. commande centralisée de plusieurs machines, p. ex. commande numérique directe ou distribuée [DNC], systèmes d'ateliers flexibles [FMS], systèmes de fabrication intégrés [IMS], productique [CIM]
H04L 69/00 - Dispositions, protocoles ou services de réseau indépendants de la charge utile de l'application et non couverts dans un des autres groupes de la présente sous-classe
36 - Services financiers, assurances et affaires immobilières
42 - Services scientifiques, technologiques et industriels, recherche et conception
Produits et services
Financing services; bill payment services Software as a service (SaaS) services featuring software for providing financial services and for processing electronic payments; Platform as a service (PaaS) featuring computer software platforms for providing financial services and for processing electronic payments
90.
GRAPH CONVOLUTIONAL REINFORCEMENT LEARNING WITH HETEROGENEOUS AGENT GROUPS
A system and method adaptively control a heterogeneous system of systems. A graph convolutional network (GCN) that receive a time series of graphs representing topology of an observed environment at a time moment and state of a system. Embedded features are generated having local information for each graph node. Embedded features are divided into embedded states grouped according to a defined grouping, such as node type. Each of several reinforcement learning algorithms are assigned to a unique group and include an adaptive control policy in which a control action is learned for a given embedded state. Reward information is received from the environment with a local reward related to performance specific to the unique group and a global reward related to performance of the whole graph responsive to the control action. Parameters of the GCN and adaptive control policy are updated using state information, control action information, and reward information.
G05B 13/02 - Systèmes de commande adaptatifs, c.-à-d. systèmes se réglant eux-mêmes automatiquement pour obtenir un rendement optimal suivant un critère prédéterminé électriques
91.
AUTOMATED AERIAL DATA CAPTURE FOR 3D MODELING OF UNKNOWN OBJECTS IN UNKNOWN ENVIRONMENTS
System and method are disclosed for multi-phase process of automated data capture for photogrammetry and 3D model building of an unknown object (311) in an unknown environment. Planner module (152) generates a flight plan (413) for a camera drone (110) to fly autonomously on a flight path along a virtual polygon grid (302) defined above the target object (311) during a survey phase. Model builder computer (153) receives a point cloud dataset (321) captured by LiDAR sensor on camera drone (301) during survey flight and constructs low resolution 3D mesh (331) of the target object (311). Planner module (152) generates a flight path (413) for camera drone inspection phase with virtual waypoints surrounding the target object (311) at a marginal distance from the surface defined by the low resolution 3D mesh (331). Model builder (153, 163) builds a high resolution 3D model (422) of the target object (311) using photogrammetry processing of high resolution images captured by camera drone (411, 412) during inspection phase.
Current approaches to controlling robots from multiple vendors typically requires multiple software systems that define vendor-exclusive fleet manager or dispatch systems. Autonomous devices (e.g., robots, drones, vehicles) can be controlled from multiple vendors that use multiple locally sourced map. For example, maps from individual robots can be translated to a base map that can be used to command and control hybrid fleets of robots.
G01C 21/00 - NavigationInstruments de navigation non prévus dans les groupes
G01C 21/28 - NavigationInstruments de navigation non prévus dans les groupes spécialement adaptés pour la navigation dans un réseau routier avec corrélation de données de plusieurs instruments de navigation
G01C 25/00 - Fabrication, étalonnage, nettoyage ou réparation des instruments ou des dispositifs mentionnés dans les autres groupes de la présente sous-classe
G05D 1/02 - Commande de la position ou du cap par référence à un système à deux dimensions
G05D 1/08 - Commande de l'attitude, c. à d. élimination ou réduction des effets du roulis, du tangage ou des embardées
G01C 21/20 - Instruments pour effectuer des calculs de navigation
93.
AN AUTOMATION SYSTEM AND A METHOD THAT PROVIDE A GLOBAL VIEW FOR MANAGING ANY INDUSTRIAL CONTROLLER IN A NETWORK
A system and a method provide a global view of an automation system for any industrial controller in a network. The method comprises providing a distributed version control runtime system for managing industrial controller process images in that an automation engineering process provides non-linear workflows. The method further comprises providing an engineering system having a first industrial controller program database of an industrial controller program. The method further comprises providing a first industrial controller having a first process image including a second industrial controller program database of an industrial controller program and a first historian database. The method further comprises providing a second industrial controller having a second process image including a third industrial controller program database of an industrial controller program and a second historian database. The method further comprises managing versions of the automation code and versions of the data of the industrial controller process image.
G05B 19/05 - Automates à logique programmables, p. ex. simulant les interconnexions logiques de signaux d'après des diagrammes en échelle ou des organigrammes
94.
BAYESIAN OPTIMIZATION FOR MATERIAL SYSTEM OPTIMIZATION
A method of optimizing a process having a plurality of potential inputs, comprising selecting a first set of inputs from the plurality of potential inputs, providing the first set of inputs from the to a first optimization process, running an objective function on the first set of inputs to produce a value corresponding to the set of inputs, providing the value to a second optimization process, running an acquisition function in the second optimization process to select a new candidate set of inputs from the plurality of potential inputs, and providing the selected new candidate set of inputs to the first optimization process. In one embodiment, the inputs are a set of lattice kernels for constructing a structural object. A Bayesian optimization is used to select sub-sets of kernels from the set of inputs. The inputs are provided to a topology optimization for evaluation.
G06F 111/20 - CAO de configuration, p. ex. conception par assemblage ou positionnement de modules sélectionnés à partir de bibliothèques de modules préconçus
G06F 113/10 - Fabrication additive, p. ex. impression en 3D
95.
SYSTEM AND METHOD FOR SUPPORTING EXECUTION OF BATCH PRODUCTION USING REINFORCEMENT LEARNING
A computer-implemented method for supporting execution of batch production by a production system includes, over a sequence of steps: acquiring a system state defined by a shop floor status, inventory status and a demand of the product types, and processing the system state using a reinforcement learned policy including a deep learning model to output a control action defining an integer batch size of a selected product type. The control action is determined by using learned parameters of the deep learning model to compute logits for a categorical distribution of predicted product types and a categorical distribution of predicted batch sizes from the system state. The logits are processed to transform the categorical distribution of predicted product types into an encoding of the selected product type and reduce the categorical distribution of predicted batch sizes into an integer batch size, for producing a next batch on the shop floor.
G06Q 10/06 - Ressources, gestion de tâches, des ressources humaines ou de projetsPlanification d’entreprise ou d’organisationModélisation d’entreprise ou d’organisation
G06Q 10/04 - Prévision ou optimisation spécialement adaptées à des fins administratives ou de gestion, p. ex. programmation linéaire ou "problème d’optimisation des stocks"
96.
CONTROLLER FOR AUTONOMOUS AGENTS USING REINFORCEMENT LEARNING WITH CONTROL BARRIER FUNCTIONS TO OVERCOME INACCURATE SAFETY REGION
System and method are disclosed for approximating unknown safety constraints during reinforcement learning of an autonomous agent. A controller for directing the autonomous agent includes a reinforcement learning (RL) algorithm configured to define a policy for behavior of the autonomous agent, and a control barrier function (CBF) algorithm configured to calculate a corrected policy that relocates policy states to an edge of a safety region. Iterations of the RL algorithm safely learn an optimal policy where exploration remains within the safety region. CBF algorithm uses standard least squares to derive estimates of coefficients for linear constraints of the safe region. This overcomes inaccurate estimation of safety region constraints caused by one or more noisy observations of constraints received by sensors.
G06N 5/00 - Agencements informatiques utilisant des modèles fondés sur la connaissance
G06N 20/10 - Apprentissage automatique utilisant des méthodes à noyaux, p. ex. séparateurs à vaste marge [SVM]
G05D 1/00 - Commande de la position, du cap, de l'altitude ou de l'attitude des véhicules terrestres, aquatiques, aériens ou spatiaux, p. ex. utilisant des pilotes automatiques
G06N 3/00 - Agencements informatiques fondés sur des modèles biologiques
97.
MACHINE LEARNING-BASED ENVIRONMENT FAIL-SAFES THROUGH MULTIPLE CAMERA VIEWS
A computing system may include a fail-safe learning engine configured to access camera data captured by multiple cameras positioned within an environment during a learning phase, generate training data based on the camera data captured by the multiple cameras, and construct a human detection model using the training data. The computing system may also include a fail-safe trigger engine configured to access camera data captured by the multiple cameras positioned within the environment during an active phase, and the camera data captured during the active phase may include a target object. The fail-trigger engine may further be configured to provide, as an input to the human detection model, the camera data that includes the target object and execute a fail-safe action in the environment responsive to the determination, provided by the human detection model, indicating that the target object is a human.
A system determines a location and a size of a new power generation or power regulating unit within a current system architecture of a power system including a plurality of power generation units. The system comprises a controller including a processor and a memory, computer-readable logic code stored in the memory which, when executed by the processor, causes the controller to execute a hybrid algorithm as a combination of a data-driven algorithm and a model-based algorithm to determine an optimal location and size of the new power generation or power regulating unit. The data-driven algorithm encodes a location and a size information. The controller to enable the model-based algorithm to optimize performance of a selected location and size of the new power generation or power regulating unit, which is based on a linearized system or a nonlinear system to provide guidance for the data-driven algorithm to incorporate physical rules and verify a new system architecture.
G06F 30/18 - Conception de réseaux, p. ex. conception basée sur les aspects topologiques ou d’interconnexion des systèmes d’approvisionnement en eau, électricité ou gaz, de tuyauterie, de chauffage, ventilation et climatisation [CVC], ou de systèmes de câblage
G06F 30/27 - Optimisation, vérification ou simulation de l’objet conçu utilisant l’apprentissage automatique, p. ex. l’intelligence artificielle, les réseaux neuronaux, les machines à support de vecteur [MSV] ou l’apprentissage d’un modèle
G06F 111/06 - Optimisation multi-objectif, p. ex. optimisation de Pareto utilisant le recuit simulé, les algorithmes de colonies de fourmis ou les algorithmes génétiques
G06F 111/20 - CAO de configuration, p. ex. conception par assemblage ou positionnement de modules sélectionnés à partir de bibliothèques de modules préconçus
A system and method are disclosed for preliminary design validation of a target design entity. A realistic dynamic environment module simulates dynamic conditions of an environment related to the target design entity. An interactive platform module drives a user interface for interactive placement and modification of environmental scenario assets obtained from an asset library, interactive placement of the target design entity within the virtual environment, and intuitive modification of shape and material properties of the target design entity. Assets are updated for use by the realistic dynamic environment module and automatically converted to boundary conditions. Physics simulator module executes physics simulations that augment environmental scenario assets by incorporating expected physics behavior, the effect of such behavior on other entities, and the impact of other physics aspects generated by other entities onto the target design entity.
It is recognized herein that current approaches to autonomous operations are often limited to grasping and manipulation operations that can be performed in a single step. It is further recognized herein that there are various operations in robotics (e.g., assembly tasks) that require multiple steps or a sequence of motions to be performed. To determine or plan a sequence of motions for fulfilling a task, an autonomous system that includes a robot can perform object recognition, pose estimation, affordance analysis, decision-making, probabilistic task or motion planning, and object manipulation.
