Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for optimizing a plan for one or more robots using a process definition graph. One of the methods includes receiving a process definition graph for a robot, the process definition graph having a plurality of action nodes, and the plurality of action nodes including a plurality of motion nodes that were previously split from a single motion node due to a conflict with a second motion node representing a second motion to be performed by another robot; determining that the conflict with the second motion node no longer exists; and in response to determining that the conflict with the second motion node no longer exists, modifying the process definition graph including combining the plurality of motion nodes into a new single motion node representing all of the motions of the plurality of motion nodes.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for planning by work volumes to avoid conflicts. One of the methods includes receiving a process definition graph for a robot that includes action nodes, wherein the action nodes include (1) transition nodes that represent a motion to be taken by the robot from a respective start location to an end location and (2) task nodes that represent a particular task to be performed by the robot at a particular task location. An initial modified process definition graph that ignores one or more conflicts between respective transition nodes as well as one or more conflicts between respective transition nodes and task nodes is generated from the process definition graph. A refined process definition graph that ignores conflicts between transition nodes and recognizes conflicts between transition nodes and task nodes is generated from the initial modified process definition graph.

Abstract: This specification describes systems, methods, devices, and other techniques for planning workspaces for automated fabrication processes. A computing system facilitates planning by receiving a set of parameters for planning a layout of a workspace for an automated fabrication process, and generating a plurality of candidate workspace layouts, including selecting, for each candidate workspace layout, (i) one or more robots for performing tasks in the automated fabrication process and (ii) corresponding locations for the one or more robots within the workspace. The system determines an optimal workspace layout based on the plurality of candidate workspace layouts, generates a workspace layout specification for the optimal workspace layout, and provides the workspace layout specification to one or more second computing systems.

Abstract: A system for event prediction using schema networks includes a first antecedent entity state that represents a first entity at a first time; a first consequent entity state that represents the first entity at a second time; a second antecedent entity state that represents a second entity at the first time; and a first schema factor that couples the first and second antecedent entity states to the first consequent entity state; wherein the first schema factor is configured to predict the first consequent entity state from the first and second antecedent entity states.

Abstract: A computer-implemented method for surface modeling includes: receiving one or more polarization raw frames of a surface of a physical object, the polarization raw frames being captured with a polarizing filter at different linear polarization angles; extracting one or more first tensors in one or more polarization representation spaces from the polarization raw frames; and detecting a surface characteristic of the surface of the physical object based on the one or more first tensors in the one or more polarization representation spaces.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for computing interpolated robot control parameters. One of the methods includes receiving, by a real-time bridge from a control agent for a robot, a non-real-time command for the robot, wherein the non-real-time command specifies a trajectory to be attained by a component of the robot and a target value for a control parameter, wherein the control parameter controls how a real-time controller will cause the robot to react to one or more external stimuli encountered during a control cycle of the real-time controller. The real-time bridge provides the one or more real-time commands translated from the non-real-time command and interpolated control parameter information to the real-time controller, thereby causing the robot to effectuate the trajectory of the non-real-time command according to the interpolated control parameter information.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributing skill templates for robotic demonstration learning. One of the methods includes receiving, from the user device by a skill template distribution system, a selection of an available skill template. The skill template distribution system provides a skill template, wherein the skill template comprises information representing a state machine of one or more tasks, and wherein the skill template specifies which of the one or more tasks are demonstration subtasks requiring local demonstration data. The skill template distribution system trains a machine learning model for the demonstration subtask using a local demonstration data to generate learned parameter values.

Abstract: According to one embodiment of the present disclosure, an imaging system includes: an image sensor including a plurality of subpixels grouped into a plurality of pixels; a polarization system including: a rotatable linear polarizer; and a polarizer mask including a plurality of polarizer filters, the polarizer filters being aligned with corresponding ones of the subpixels, the subpixels of a pixel of the plurality of pixels being located behind polarizer filters at different angles of linear polarization; and imaging optics configured to focus light from a scene onto the image sensor.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for distributed robotic demonstration learning. One of the methods includes receiving a skill template to be trained to cause a robot to perform a particular skill having a plurality of subtasks. One or more demonstration subtasks defined by the skill template are identified, wherein each demonstration subtask is an action to be refined using local demonstration data. On online execution system uploads sets of local demonstration data to a cloud-based training system. The cloud-based training system generates respective trained model parameters for each set of local demonstration data. The skill template is executed on the robot using the trained model parameters generated by the cloud-based training system.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for rule execution in an online robotics system. One of the systems includes an execution engine subsystem and an execution memory subsystem. The execution engine receives rules having types and subtypes that represent a particular entity in an operating environment of a robot, provides subscription requests to the execution memory subsystem, and receives events emitted by the execution memory subsystem. The an execution memory receives subscription requests from the execution engine subsystem, receives new observations, converts the new observations into fact updates, performs pattern matching with the fact updates against the patterns of the subscription requests, and emits events to the execution engine subsystem for patterns that have been matched by the fact updates.

Abstract: Systems and methods for adjusting an exposure parameter of an imaging device are disclosed. A first exposure level of the imaging device is identified, and a first image of a scene is captured via the imaging device at the first exposure level. The first image of the scene comprises a plurality of polarization images corresponding to different degrees and angles of polarization. Each of the polarization images comprise a plurality of color channels. A gradient for the first image is computed based on the plurality of the polarization images, and a second exposure level is computed based on the gradient. A second image of the scene is captured based on the second exposure level, where the gradient of the second image is greater than a gradient for the first image.

Abstract: Methods, systems, and computer programs stored on computer storage devices, for coordinating movements of robots are disclosed. One of the methods includes, for each robot in a group of robots, identifying a set of tasks assigned to the robot and generating a plurality of candidate motion plans. The method further includes, for each candidate motion plan: (i) generating a 3D model that represents a volume of space through which the robot would move in executing the sequence of motions represented by the candidate motion plan, and (ii) determining a score for the candidate motion plan. The method further includes determining conflicts between candidate motion plans of different robots, selecting a motion plan from the candidate motion plans based on the score for the selected motion plan and the conflicts, and providing the selected motion plans for execution by the group of robots.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for performing real-time code analysis. One of the methods includes receiving a request to perform real-time code analysis on source code, wherein the request identifies one or more target source code elements. A plurality of real-time annotation values occurring in the source code for the target source code elements are identified. A call graph is generated for the one or more target source code elements. The call graph is traversed to identify one or more real-time violations, wherein each real-time violation is an instance in the source code that violates one or more real-time safe criteria. An output is provided that identifies one or more of the real-time violations occurring in the source code.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for performing automated safety procedures for a robot. One of the methods includes receiving, by a robotic control system for a robot, a request to execute an automated safety procedure by a safety control subsystem for the robot. Each step of the automated safety procedure is iterated until an end of the automated safety procedure is reached, including if a step requires a new safety configuration, a respective safety configuration for the step is obtained and activated before performing one or more automatic actions for the step.

Abstract: A system and method for machine understanding, using program induction, includes a visual cognitive computer including a set of components designed to execute predetermined primitive functions. The method includes determining programs using a program induction engine that interfaces with the visual cognitive computer to discover programs using the predetermined primitive functions and/or executes the discovered programs based on an input.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for planning robotic movements to perform a given task while satisfying object pose estimation accuracy requirements. One of the methods includes generating a plurality of candidate measurement configurations for measuring an object to be manipulated by a robot; determining respective measurement accuracies for the plurality of candidate measurement configurations; determining a measurement accuracy landscape for the object including defining a high measurement accuracy region based on the respective measurement accuracies for the plurality of candidate measurement configurations; and generating a motion plan for manipulating the object in the robotic process that moves the robot, a sensor, or both, through the high measurement accuracy region when performing pose estimation for the object.

Abstract: A method for preemptive control planning can include: optionally determining a graph that defines dependencies between behaviors; executing an execution plan for an executing behavior; generating a set of predictions with the executing behavior; propagating the set of predictions to child behaviors of the executing behavior; determining child execution plans and child predictions with the child behaviors; repeating the above for descendant behaviors of each child behavior; determining a realized world state associated with completion of the execution plan; and executing a child execution plan associated with a world state prediction matching the realized world state.

Abstract: The system can include: a container 110, a set of sensors 120, and a controller 130. The system can optionally include a robot 140. However, the system 100 can additionally or alternatively include any other suitable set of components. The system functions to monitor and/or maintain a fullness level of a container. The system can additionally or alternatively function to enable robotic picking out of the container (e.g., in a pick-and-place setting). The system can additionally function to maintain candidate objects within reach of the robot's end effector to increase robot uptime while minimizing the extent of the robot's required motion (e.g., in the z-axis).

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for optimizing a plan for one or more robots using a process definition graph. One of the methods includes receiving a process definition graph for a robot, the process definition graph having a plurality of action nodes. One or more of the action nodes are motion nodes that represent a motion to be taken by the robot from a respective start location to an end location. It is determined that a motion node satisfies one or more splitting criteria, and in response to determining that the motion node satisfies the one or more splitting criteria, the process definition graph is modified. Modifying the process definition graph includes splitting the motion node into two or more separate motion nodes whose respective paths can be scheduled independently.

Abstract: A method for characterizing a pose estimation system includes: receiving, from a pose estimation system, first poses of an arrangement of objects in a first scene; receiving, from the pose estimation system, second poses of the arrangement of objects in a second scene, the second scene being a rigid transformation of the arrangement of objects of the first scene with respect to the pose estimation system; computing a coarse scene transformation between the first scene and the second scene; matching corresponding poses between the first poses and the second poses; computing a refined scene transformation between the first scene and the second scene based on coarse scene transformation, the first poses, and the second poses; transforming the first poses based on the refined scene transformation to compute transformed first poses; and computing an average rotation error and an average translation error of the pose estimation system based on differences between the transformed first poses and the second poses.