Abstract: A component of a robotic system, including: processor circuitry; and a non-transitory computer-readable storage medium including instructions that, when executed by the processor circuitry, cause the processor circuitry to: train a neuro-capability map plugin, which is a continuous or semi-continuous resolution neural network component encoded with kinematic capability attributes with respect to an action to be performed by a robot in a workspace; and publish the neuro-capability map plugin to a robotic skills repository where it is obtainable by robotic controller circuitry for embedding within a neural network usable perform one or more inferences to control the robot to perform the action.

Abstract: A system can map occupancy of objects. The system may generate an occupancy representation of an object based on a point cloud of the object. The system may generate first geometric entities based on the point cloud. Each first geometric entity contains one or more points in the point cloud. The system may also generate one or more second geometric entities, each of which contains one or more first geometric entities. The occupancy representation of the object includes the one or more second geometric entities and the plurality of first geometric entities. The occupancy representation may have a hierarchical structure where the first geometric entities may be on a lower level than the one or more second geometric entities. The system can also detect collision of the object with another object by using the occupancy representation of the object and an occupancy representation of the other object.

Abstract: System and techniques for tool position determination in a robotic appendage are described herein. A robotic appendage is put through a rotational movement to induce acceleration in a tool mounted to the appendage. A model for acceleration is created from positional kinematics of the appendage. A measurement of acceleration is taken at the tool and fit to the model to determine distance from the axis of rotation to the tool. The distance is provided for use in control or modeling of the robotic appendage.

Abstract: Example wireless feedback control systems disclosed herein include a receiver to receive a first measurement of a target system via a first wireless link. Disclosed example systems also include a neural network to predict a value of a state of the target system at a future time relative to a prior time associated with the first measurement, the neural network to predict the value of the state of the target system based on the first measurement and a prior sequence of values of a control signal previously generated to control the target system during a time interval between the prior time and the future time, and the neural network to output the predicted value of the state of the target system to a controller. Disclosed example systems further include a transmitter to transmit a new value of the control signal to the target system via a second wireless link.

Abstract: Techniques are disclosed to facilitate, in autonomous vehicles, the robust detection and classification of objects in a scene using a static sensors in conjunction with event-based sensors. A trained system architecture may be implemented, and the fusion of both sensors thus allows for the consideration of scenes with overexposure, scenes with underexposure, as well as scenes in which there is no movement. In doing so, the autonomous vehicle may detect and classify objects in conditions in which each sensor, if operating separately, would not otherwise be able to classify (or classify with high uncertainty) due to the sensing environment.

Abstract: A robot configured to be operable within a multi-robot system. The robot includes an input configured to receive global coordinate state information of the robot and of any neighboring robots or obstacles; and processing circuitry configured to: transform the global coordinate state information into a relative coordinate system that is with respect to the robot and is based on a type of desired formation of the robot and any neighboring robots or obstacles around a point; generate a reference formation algorithm which is based on the desired formation; and controlling, based on the reference formation algorithm and tracking errors between the desired formation and a current state of the robot, a trajectory of the robot to converge towards the desired formation while avoiding collisions with any neighboring robots or obstacles.

Abstract: Techniques are disclosed for a decentralized path and motion planning of autonomous agents within an environment. The planning may include determining if an active neighboring autonomous agent is present and selectively controlling the autonomous agent to operation in in an independent path planning operation mode and in a coordinating path planning operation mode, based on the detection of the neighboring agent(s).

Abstract: Methods and apparatus for dynamically routing robots based on exploratory on-board mapping are disclosed. A control system of a robot includes an image manager to command a depth camera to capture depth images of an environment. The depth camera has a field of view. The control system further includes a map generator to generate a map of the environment based on the depth images. The map includes a representation of unoccupied space within the environment, and a path extending through the unoccupied space from a reference location of the robot to a target location of the robot. The control system further includes a field of view evaluator to determine whether the field of view associated with the reference location satisfies a threshold. The control system further includes a route generator to generate, in response to the field of view associated with the reference location satisfying the threshold, a route to be followed by the robot within the environment.

Abstract: Methods and apparatus to facilitate autonomous navigation of robotic devices. An example autonomous robot includes a region model analyzer to: analyze a first image of an environment based on a first neural network model, the first image captured by an image sensor of the robot when the robot is in a first region of the environment; and analyze a second image of the environment based on a second neural network model, the second image captured by the image sensor when the robot is in a second region of the environment, the second neural network associated with the second region. The example robot further includes a movement controller to: autonomously control movement of the robot within the first region toward the second region based on the analysis of the first image; and autonomously control movement of the robot within the second region based on the analysis of the second image.

Abstract: Systems and techniques for an architecture for contextual memories in map representation for 3D reconstruction and navigation are described herein. In an example, a system for contextual memory mapping is adapted to receive a data set of physical world sensor readings. The system may be further adapted to generate voxel data from the data set, the voxel data includes voxel coordinates and a physical world occupancy indicator. The system may be further adapted to select a block of addresses in the memory to store the voxel data. The system may be further adapted to generate a hash map to map voxel coordinates to memory locations in the block of addresses, the voxel coordinates having a contextual relationship that is maintained by the hash map. The system may be further adapted to store the voxel data at memory addresses based on the hash map.

Abstract: Various systems and methods for implementing an enhanced occupant collision safety system in a vehicle are described herein. A system for controlling deployment of vehicle safety mechanisms of a host vehicle, includes a sensor array interface to receive acceleration data from an impact sensor; an image processor to receive events from an event camera, the event camera configured to monitor a scene of an interior of the host vehicle that includes an occupant; and an occupant safety system controller to: cluster the events to discriminate between events associated with a background of the scene and events associated with a body part of the occupant; detect a collision based on the acceleration data; verify the collision based on the events associated with the background of the scene; and in response to verifying the collision, cause a collision safety mechanism of the host vehicle to deploy.

Abstract: A human-robot collaboration system, including at least one processor; and a non-transitory computer-readable storage medium including instructions that, when executed by the at least one processor, cause the at least one processor to: predict a human atomic action based on a probability density function of possible human atomic actions for performing a predefined task; and plan a motion of the robot based on the predicted human atomic action.

Abstract: An apparatus of an autonomous device comprises one or more state estimators to estimate one or more states of the autonomous device, wherein the one or more state estimators are to generate one or more derivatives of translational measurements, orientation measurements, reference translational values, and reference orientation values, and one or more controllers to receive an output from the one or more state estimators to provide control signals to control the autonomous device. The one or more state estimators include a hardware differentiator to generate the one or more derivatives.

Abstract: Techniques are disclosed for the acceleration of the operation of a probabilistic sampling device for applications including homography estimation and wireless signal detection, which may include reducing the number of iterations associated with probabilistic sampling. Techniques are also disclosed for corner feature extraction from event-based cameras, which may be implemented via an Advanced Driver Assistance System (ADAS) or Autonomous Driving (AD) system.

Abstract: Techniques are disclosed to facilitate cooperative mapping for safe and efficient trajectory planning and collision avoidance by allowing nearby agents to share contextual information. The described techniques also function to extend the mapping range of a single agent by leveraging observations made by multiple agents. Furthermore, the techniques as described herein function to reduce uncertainty in trajectory planning by allowing agents to “see” behind occlusions, thus taking advantage of observations made by neighboring agents from different points of view. An efficient hardware implementation of the system is also presented that leverages the methodologies as discussed herein.

Abstract: Methods and apparatus for preventing collisions between drones based on drone-to-drone acoustic communications are disclosed. A control system of a first drone includes a route manager to command the first drone to track a first route. The control system further includes a plurality of acoustic sensors arranged on the first drone to detect acoustic signals. The control system further includes a collision prevention engine to determine a second route based on acoustic signals detected by the acoustic sensors, the acoustic signals being received from a second drone while the first drone is tracking the first route. The second route differs from the first route to prevent a drone-to-drone collision between the first drone and the second drone. The collision prevention engine is to cause the route manager to command the first drone to track the second route instead of the first route.

Abstract: A drone controller comprises processing circuitry to, at a first time when the drone is in flight, to determine a drone state comprising a position and a velocity of the drone and determine a relative obstacle state comprising a relative position and a relative velocity of the drone with respect to an obstacle. The processor then determines a reaction to avoid the obstacle based on the relative obstacle state and applies a signal related to the reaction to one or more actuator control inputs of the drone that modifies a drone path existing at the first time to avoid the obstacle.

Abstract: System and techniques for drone obstacle avoidance using real-time wind estimation are described herein. A wind metric is measures at a first drone and communicated to a second drone. In response to receiving the wind metric, a flight plan of the second drone is modified based on the wind metric.

Abstract: In one example, notice of a priority vehicle is obtained. A first protective field is generated around a current vehicle and a second protective field is generated around the priority vehicle. A priority lane is created by moving the first protective field of the current vehicle away from the second protective field of the priority vehicle.

Abstract: Various systems and methods for drone path planning are described herein. A system for drone path planning for a drone, the system to perform the operations: storing a current location of the drone as a current waypoint in a set of waypoints, the set of waypoints to include the current waypoint and a previous waypoint; detecting a current set of frontiers from the current location; storing the current set of frontiers in a list of active goals, the current set of frontiers associated with the current location; determining whether a target is discovered in the current set of frontiers; exploring the current set of frontiers to attempt to find the target; and exploring a previous set of frontiers from the list of active goals, when exploring the current set of frontiers does not, discover the target, the previous set of frontiers associated with the previous waypoint.