Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media for evaluating robot learning. In some implementations, one or more computers receive object classification examples from a plurality of robots. Each object classification example includes (i) an embedding that a robot generated using a machine learning model, and (ii) an object classification corresponding to the embedding. The object classification examples are evaluated based on a similarity of the received embeddings with respect to other embeddings. A subset of the object classification examples is selected based on the evaluation of the quality of the embeddings. The subset of the object classification examples is distributed to the robots in the plurality of robots.

Abstract: An example method includes determining a path to be followed by a vehicle through an environment. The path includes an ordered sequence of positions. The method also includes determining an intersection between a first object in the environment and a first area planned to be occupied by the vehicle while moving along the path and, in response, sequentially testing the ordered sequence of positions to identify a first ordinal position in the ordered sequence of positions, where the first ordinal position corresponds to a second area planned to be occupied by the vehicle while moving along the path, and where the second area is within a threshold distance of the first object. The method additionally includes trimming the path to remove (i) the first ordinal position and (ii) any positions subsequent thereto and causing the vehicle to stop at an end of the trimmed path.

Abstract: Aspects of the disclosure provide an optical communication system. The system may include a receiver lens system configured to receive a light beam from a remote optical communication system and direct the light beam to a photodetector. The system may also include the photodetector. The photodetector may be configured to convert the received light beam into an electrical signal, and the photodetector may be positioned at a focal plane of the receiver lens system. The system may also include a phase-aberrating element arranged with respect to the receiver lens system and the photodetector such that the phase-aberrating element is configured to provide uniform angular irradiance at the focal plane of the receiver lens system.

Abstract: Methods for designing a mode-selective optical device including one or more optical interfaces defining an optical cavity include: defining a loss function within a simulation space encompassing the optical device, the loss function corresponding to an electromagnetic field having an operative wavelength within the optical device resulting from an interaction between an input electromagnetic field at the operative wavelength and the one or more optical interfaces of the optical device; defining an initial structure for each of the one or more optical interfaces, each initial structure being defined using a plurality of voxels; determining values for at least one structural parameter and/or at least one functional parameter of the one or more optical interfaces by solving Maxwell's equations; and defining a final structure of the one or more optical interfaces based on the values for the one or more structural and/or functional parameters.

Abstract: Training and/or use of a machine learning model for placement of an object secured by an end effector of a robot. A trained machine learning model can be used to process: (1) a current image, captured by a vision component of a robot, that captures an end effector securing an object; (2) a candidate end effector action that defines a candidate motion of the end effector; and (3) a target placement input that indicates a target placement location for the object. Based on the processing, a prediction can be generated that indicates likelihood of successful placement of the object in the target placement location with application of the motion defined by the candidate end effector action. At many iterations, the candidate end effector action with the highest probability is selected and control commands provided to cause the end effector to move in conformance with the corresponding end effector action.

Abstract: A method for analyzing electroencephalogram (EEG) signals is disclosed. Information associated with two or more options is presented to a user. EEG signals from a sensor coupled to the user are received contemporaneously to the user receiving information associated with the two or more options. The EEG signals are processed in real time to determine which one of the options was selected by the user. In response to determining which one of the options was selected by the user, an action from one or more possible actions associated with the information presented to the user is selected. An output associated with the selected action is then generated.

Abstract: Systems, methods, devices, and other techniques are described for planning motions of one or more robots to perform at least one specified task. In some implementations, a task to execute with a robotic system using a tool is identified. A partially constrained pose is identified for the tool that is to apply during execution of the task. A set of possible constraints for the unconstrained pose parameter are selected for each unconstrained pose parameter. The sets of possible constraints are evaluated for the unconstrained pose parameters with respect to one or more task execution criteria. A nominal pose is determined for the tool based on a result of evaluating the sets of possible constraints for the unconstrained pose parameters with respect to the one or more task execution criteria. The robotic system is then directed to execute the task, including positioning the tool according to the nominal pose.

Abstract: A technique for optimizing a physical device includes receiving an initial description of the physical device that describes the physical device with voxels that each describes one or more structural parameters of the physical device. The initial description includes a characterization including a desired output signal generated at an output region of the physical device in response to a source signal at a source region of the physical device. A field response is forward simulated from the source region to the output region to generate a forward simulated output signal. Structural parameter weights of the voxels are adjusted with an adaptive algorithm configured to reduce an error between the forward simulated output signal and the desired output signal. The structural parameters of the voxels are revised based upon the adjusting. The forward simulating, adjusting, and revising are iteratively repeated and a revised/optimized description of the physical device is generated.

Abstract: Systems and methods related to roadmaps for mobile robots are provided. A computing device can determine a roadmap of an environment. The roadmap can include lanes and a designated region that is adjacent to a first lane of the plurality of lanes and suitable for robotic traversal when unoccupied. The computing device can determine a first route between first and second points in the environment that uses the first lane. The computing device can send a direction to use the first route to a first robot. The computing device can receive, from the first robot, sensor data indicative of an occupied status of the designated region. The computing device can determine a second route between the first and second points through the designated region based on the occupied status of the designated region. The computing device can send a direction to use the second route to a second robot.

Abstract: A physical voxel, a volumetric testbed, and method for physically simulating a photonic device are described herein. The volumetric testbed comprises a simulation stage and a controller. The simulation stage includes a three-dimensional array of physical voxels configurable to represent the photonic device operating in response to electromagnetic radiation. The physical voxels includes a field detector to measure a local field response and an impedance adjuster to adjust an impedance to the electromagnetic radiation. The controller is coupled to memory, which stores instructions that when executed by one or more processors included in the controller causes the volumetric testbed to perform operations including determining a global field response of the photonic device and adjusting the impedance of the physical voxels to refine a design of the photonic device.

Abstract: Described herein are three-dimensional (3D) printer systems and methods, which may provide for “continuous pull” 3D printing. An illustrative 3D printer includes: a resin container, a base plate, a light source arranged below the resin container and operable to cure resin in the resin container; and a control system operable to: (a) receive model data specifying a 3D structure; (b) determine 2D images corresponding to layers of the 3D object; and (c) generate control signals to operate the light source and the base plate to sequentially form the layers of the 3D object onto the base plate, wherein the base plate moves a formed portion of the 3D object upward after formation of each layer, and wherein at least a surface of a formed portion of the 3D object remains in contact with the resin in the resin container throughout the formation of the layers of the 3D object.

Abstract: The disclosure may provide for a communication method and system. A transmitter of the communication system may include an interleaver and a first encoder for determining parity bits. The transmitter also may include a multiplexer for joining the parity bits with the data. A second encoder may be positioned after the multiplexer for implementing an error correcting code. A receiver of the communication system may include a decoder followed by an interleaver. When errors are detected in received data at the decoder, one or more processors of the receiver may be configured to correct portions of the received data and combine the corrected portions with the received data.

Abstract: Methods, apparatus, and computer readable media applicable to robots, such as balancing robots. Some implementations are directed to determining multiple measures of a property of a robot for a given time and determining a final measure of the property of the robot for the given time based on the multiple measures. One or more control commands may be generated based on the final measure of the property and provided to one or more actuators of the robot.

Abstract: Aspects of the technology relate to temperature regulation for high altitude, long duration balloons, such as balloons that operate in the stratosphere for weeks, months or longer. A balloon covering overlays the balloon envelope. The material properties and configuration of the balloon covering are selected to enable it to automatically cover or expose portions of the envelope depending on environmental conditions. The covering has multiple layers of materials with different coefficients of thermal expansion (CTE). The layers are chosen to have significantly different CTEs, and flaps are created in the envelope such that depending on the temperature, the flaps will either lie flat on the envelope or curl away from the envelope. In the former case, the flaps help to retain heat in the envelope, such as during nighttime operation. Conversely, in the latter case, the flaps permit heat to escape from the envelope, such as during daytime operation.

Abstract: A free space optical communication system transmits and receives optical signals in a colorless manner using an optical circulator. The system installs the optical circulator with a single mode (SM) fiber at port 1, a double clad (DC) fiber at port 2, and a multimode (MM) fiber at port 3. The system injects a first optical signal into a core of the SM fiber. The system then routes the first optical signal at port 1, using the optical circulator, into a SM core of the DC fiber via Port 2. Further, the system injects a second optical signal into a first cladding of the DC fiber. The system then routes the second optical signal at port 2, using the optical circulator, into the MM fiber via Port 3.

Abstract: A robotic system includes end-effector(s) that combine a plurality of objects in a production process. The system includes sensor(s) that obtain measurement(s) relating to a combination of a first object and one or more other objects during the production process. The system includes a control system communicatively coupled to the sensor(s). The control system stores specifications relating to the combination of the plurality of objects. The control system receives the measurement(s) from the sensor(s), determines a difference based on the measurement(s) and the specifications, determines adjustment(s) to the production process based on the determined difference, and sends, for the end-effector(s), instruction(s) based on the specifications and the one or more adjustment(s). The end-effector(s) combine a second object with the first object and the one or more objects based on the specifications and the one or more adjustment(s).

Abstract: Methods and apparatus related to receiving a request that includes robot instructions and/or environmental parameters, operating each of a plurality of robots based on the robot instructions and/or in an environment configured based on the environmental parameters, and storing data generated by the robots during the operating. In some implementations, at least part of the stored data that is generated by the robots is provided in response to the request and/or additional data that is generated based on the stored data is provided in response to the request.

Abstract: Methods, systems, and apparatus, including computer-readable media, for robot grasp learning. In some implementations, grasp data describing grasp attempts by robots is received. A set of the grasp attempts that represent unsuccessful grasp attempts is identified. Based on the set of grasp attempts representing unsuccessful grasp attempts, a grasp model based on sensor data for the unsuccessful grasp attempts. After training the grasp model, a performance level of the trained grasp model is verified based on one or more simulations of grasp attempts. In response to verifying the performance level of the trained grasp model, the trained grasp model is provided to one or more robots.

Abstract: Methods, systems, and apparatus for an infrared and visible imaging system. In some implementations, Image data from a visible-light camera is obtained. A position of a device is determined based at least in part on the image data from the visible-light camera. An infrared camera is positioned so that the device is in a field of view of the infrared camera, with the field of view of the infrared camera being narrower than the field of view of the visible-light camera. Infrared image data from the infrared camera that includes regions representing the device is obtained. Infrared image data from the infrared camera that represents the device is recorded. Position data is also recorded that indicates the location and pose of the infrared camera when the infrared image data is acquired by the infrared camera.