Abstract: Implementations are described herein for single iteration, multiple permutation robot simulation. In various implementations, one or more poses of a simulated object may be determined across one or more virtual environments. A plurality of simulated robots may be operated across the one or more virtual environments. For each simulated robot of the plurality of simulated robots, a camera transformation may be determined based on respective poses of the simulated robot and simulated object in the particular virtual environment. The camera transformation may be applied to the simulated object in the particular virtual environment of the one or more virtual environments in which the simulated robot operates. Based on the camera transformation, simulated vision data may be rendered that depicts the simulated object from a perspective of the simulated robot. Each of the plurality of simulated robots may be operated based on corresponding simulated vision data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer-storage media, for a lighting controller for sea lice detection. In some implementations, fish are contained within an elliptical tank filled with water. An imaging station located on the elliptical tank is used to capture an image of a fish from which image analysis can be performed to detect sea lice or other skin features, including lesions, on the fish. Pairs of imaging assemblies coordinate pulsing light of at least a first and a second color and capturing images of the fish while the fish is illuminated by at least the first and the second color. By using the captured images with different color light, computers can detect features on the body of a fish including sea lice, skin lesions, shortened operculum or other physical deformities and skin features. Detection results can aid in mitigation techniques or be stored for analytics.

Abstract: Generating edge-depth values for an object, utilizing the edge-depth values in generating a 3D point cloud for the object, and utilizing the generated 3D point cloud for generating a 3D bounding shape (e.g., 3D bounding box) for the object. Edge-depth values for an object are depth values that are determined from frame(s) of vision data (e.g., left/right images) that captures the object, and that are determined to correspond to an edge of the object (an edge from the perspective of frame(s) of vision data). Techniques that utilize edge-depth values for an object (exclusively, or in combination with other depth values for the object) in generating 3D bounding shapes can enable accurate 3D bounding shapes to be generated for partially or fully transparent objects. Such increased accuracy 3D bounding shapes directly improve performance of a robot that utilizes the 3D bounding shapes in performing various tasks.

Abstract: A method includes creating a publisher configured to send messages over a channel having a shared memory. The method includes creating at least one subscriber configured to receive the messages over the channel by sequentially referencing memory slots of the plurality of memory slots. The method includes determining that the next sequential memory slot is currently referenced by a subscriber. The method includes delaying sending the message by the publisher based on determining that the next sequential memory slot is currently referenced by the subscriber. The method includes receiving an event trigger indicative of message reading by the subscriber. The method includes, responsive to receiving the event trigger, determining that the next sequential memory slot is not currently referenced. The method includes sending the message to the next sequential memory slot based on determining that the next sequential memory slot is not currently referenced.

Abstract: In some embodiments, an archipelago model is provided for distributed execution of evolutionary computing techniques. In some embodiments, an archipelago manager computing device manages a centralized archipelago management queue, and provides population subsets to a plurality of island computing devices. The archipelago manager computing device receives candidate solutions from the island computing devices, stores the candidate solutions in the archipelago management queue, and transmits candidate solutions from the archipelago management queue to the island computing devices in order to exchange candidate solutions between the island computing devices. The use of an archipelago management queue allows transfer of candidate solutions between different island computing devices, is robust to failure of any given island computing device, and does not require homogeneity within the plurality of island computing devices.

Abstract: A sensor device includes a sensor housing defining a channel extending along a channel axis through the housing from a first side of the sensor housing to a second side of the sensor housing opposite the first side, at least one contact electrode extending from the first side of the housing, an electrically-conducting lead attached to the housing in electrical communication with the at least one contact electrode, and a locking mechanism located in the channel permitting one-way axial motion of a thread threaded through the channel from the first side to the second side.

Abstract: Methods, systems, and apparatus for receiving a reference to an object located in an environment of a robot, accessing mapping data that indicates, for each of a plurality of object instances, respective probabilities of the object instance being located at one or more locations in the environment, wherein the respective probabilities are based at least on an amount of time that has passed since a prior observation of the object instance was made, identifying one or more particular object instances that correspond to the referenced object, determining, based at least on the mapping data, the respective probabilities of the one or more particular object instances being located at the one or more locations in the environment, selecting, based at least on the respective probabilities, a particular location in the environment where the referenced object is most likely located, and directing the robot to navigate to the particular location.

Abstract: This specification describes trajectory planning for robotic devices. A robotic navigation system can obtain, for each of multiple time steps, data representing an environment of a robot at the time step. The system generates a series of occupancy maps for the multiple time steps, and uses the series of occupancy maps to determine occupancy predictions for one or more future time steps. Each occupancy prediction can identify predicted locations of obstacles in the environment of the robot at a different one of the future time steps. A planned trajectory can be determined for the robot using the occupancy predictions, and the robot initiates travel along the planned trajectory.

Abstract: Methods, systems, and apparatuses, including computer programs encoded on a computer storage medium, for machine learning classification based on separate processing of multiple views. In some implementations, a system obtains image data for multiple images showing different views of an object. A machine learning model is used to generate a separate output based on each the multiple images individually. The outputs for the respective images are combined to generate a combined output. A predicted characteristic of the object is determined based on the combined output. An indication of the predicted characteristic of the object is provided.

Abstract: A system includes a robotic device, a sensor disposed on the robotic device, and circuitry configured to perform operations. The operations include determining a map that represents stationary features of an environment and receiving, from the sensor, sensor data representing the environment. The operations also include determining, based on the sensor data, a representation of an actor within the environment, where the representation includes keypoints representing corresponding body locations of the actor. The operations also include determining that a portion of a particular stationary feature is positioned within a threshold distance of a particular keypoint and, based on thereon, updating the map to indicate that the portion is to be cleaned. The operations further include, based on the map as updated, causing the robotic device to clean the portion of the particular stationary feature.

Abstract: In general, innovative aspects of the subject matter described in this specification can be embodied in an electrode headset and methods of using an electrode headset. An exemplary electrode headset includes a head covering, an electrode assembly, and a plurality of electrodes. The head covering has a plurality of holes and the electrode assembly includes a plurality of first connectors arranged to align with the plurality of holes of the head covering. Each electrode has a second connector configured to releasably mate with one of the first connectors of the electrode assembly through one of the plurality of holes of the head covering.

Abstract: An energy harvesting tape comprising a plurality of flexible layers. The plurality of flexible layers includes a solar cell layer configured to capture solar energy, a thermoelectric layer configured to capture thermal energy, one or more piezoelectric layers configured to capture mechanical energy; and an electrode layer configured to capture radiofrequency energy and to transmit a radiofrequency signal. The energy harvesting tape also includes one or more processing units on at least one of the plurality of flexible layers. The one or more processing units are configured to use the captured energy from the plurality of flexible layers to transmit the radiofrequency signal. The energy harvesting tape has a length, a width, and a thickness, where the length is greater than the width, and the width is greater than the thickness.

Abstract: A method of operating a communication network includes training a neural network and implementing the neural network to determine link availability. Training the neural network includes receiving first images and signal visibility data for first locations for one or more first nodes in the communication network, generating training data based on the first images and the signal visibility data, and training the neural network using the training data to output an attenuation category related to attenuation rate of a link based on a training image and a timestamp for the training image. Implementing the neural network includes receiving second images for second locations for one or more second nodes in the communication network, determining link availability based on the second images and outputs from the neural network, and operating the communication network based on the link availability.

Abstract: Methods, apparatus, and computer-readable media for determining and utilizing corrections to robot actions. Some implementations are directed to updating a local features model of a robot in response to determining a human correction of an action performed by the robot. The local features model is used to determine, based on an embedding generated over a corresponding neural network model, one or more features that are most similar to the generated embedding. Updating the local features model in response to a human correction can include updating a feature embedding, of the local features model, that corresponds to the human correction. Adjustment(s) to the features model can immediately improve robot performance without necessitating retraining of the corresponding neural network model.

Abstract: A system, apparatus, and method for optimizing structural parameters of a physical device are described. The method includes receiving an initial description of the physical device describing the structural parameters within a simulated environment. The method further includes performing a simulation of the physical device in response to an excitation source to determine a performance metric of the physical device. The simulation environment includes one or more absorbing boundaries for attenuation of an output of the excitation source during the simulation. The method further includes recording attenuated field values of the simulated environment associated with the attenuation during the simulation.

Abstract: A robotic device may include a spine defining a yaw axis. The robotic device may also include an arm joint rotatably connected to the spine at a first position along the yaw axis and configured to rotate about the yaw axis. The robotic device may further include an actuator including a ring that defines a bore. The spine may be fixedly connected to the ring at a second position along the yaw axis and may extend through the bore. The actuator may be connected to the arm joint and configured to rotate the arm joint about the yaw axis without rotating the spine.

Abstract: A whole cell model may be constructed and used to simulate cell behavior. The whole cell model may have a baseline cell state that can be perturbed by a user in order to understand the behavior and importance of various molecules, processes and/or sub-models within the whole cell model. The simulation data is evaluated according to a variety of heuristics. The simulation data is ranked within each heuristic. The heuristic evaluation of the simulation data is then compared to an input perturbation to determine the relative importance of the heuristics. The output is a visualization of the simulation data according to each heuristic within a dynamic ranked display.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for identifying one or more regions of a brain of a biological organism that are predicted to be functionally-specialized for performing a task. In one aspect, a method comprises: obtaining data defining a synaptic connectivity graph representing synaptic connectivity between neurons in the brain of the biological organism; identifying a plurality of sub-graphs of the synaptic connectivity graph; determining, for each sub-graph of the plurality of sub-graphs, a performance measure characterizing a performance of a neural network having a neural network architecture that is specified by the sub-graph in accomplishing the task; and determining, based on the performance measures, that one or more sub-graphs of the plurality of sub-graphs correspond to regions of the brain of the biological organism that are predicted to be functionally-specialized for performing the task.

Abstract: The present disclosure relates to techniques for deformulating the spectra of arbitrary compound formulations such as polymer formulations into their chemical components. Particularly, aspects of the present disclosure are directed to obtaining an initial set of spectra for a plurality of samples comprising pure samples and composite samples, constructing a basis set of spectra for a plurality of pure samples based on the initial set of spectra, and providing or outputting the basis set of spectrum. The basis set of spectra is constructed in an iterative process that attempts to decompose, using a decomposition algorithm or model, the spectrum from the initial set of spectra in order to differentiate the pure samples from the composite samples. The basis set of spectra may then be used to deduce the composition of a material from a spectrogram.

Abstract: A method of creating a fabricable segmented design for a physical device is provided. A computing system receives a design specification. The computing system optimizes an initial segmented design based on the design specification to create an improved segmented design. The computing system perturbs the improved segmented design to create a perturbed segmented design. The computing system optimizes the perturbed segmented design to create a second improved segmented design.