Abstract: Quantization is a technique that may be used to reduce precision while maintaining performance of a neural network model during training of the neural network model, or before the neural network model is deployed. During inference, a neural network model with a fixed quantization level is used to produce predictions. In online services, such as running neural network models continuously or frequently for long periods of time in data centers, the neural network model with a fixed quantization level may not always perform optimally due to changes and/or shifts in system load and input data during inference time. A flexible and adaptive approach to quantization and a system to support adaptive quantization during inference time can be employed to address such concerns.

Abstract: Estimating a head pose may include obtaining sensor data corresponding to a head and at least a portion of the body of a human subject and determining an estimate of a three-dimensional (3D) body pose using the obtained sensor data. The estimation can further include generating a first rendering of at least the human subject's head using the obtained sensor data and generating a plurality of head pose sample data sets by applying the estimated 3D body pose to a head-pose generative model. Further, the head pose estimation can include generating a plurality of second renderings respectively from each of the plurality of head pose sample data sets; determining which of the plurality of second renderings is closest to the first rendering; and selecting the second rendering determined to be closest to the first rendering.

Abstract: A pallet loading system may comprise a processor, and memory with instructions stored thereon that, when executed by the processor, cause the processor to receive package loading data, the package loading data including characteristics of a package to be placed on the pallet and characteristics of a loading operator. A placement location for the package on the pallet may be determined using an artificial intelligence (AI) or machine learning (ML) algorithm based on the characteristics of the package and the characteristics of the loading operator and a visual marking (such as a visual projection) may be displayed at the placement location. The system may output an instruction to the loading operator to place the package at the displayed placement location.

Abstract: Disclosed herein are systems and methods for privacy-preserving facemask-compliance-level measurement. In an embodiment, a mask-compliance measurement system includes a processor that is configured to generate, from an image of a person, a facial depth image of a region of a face of the person, and to generate facial wavelet descriptors from the facial depth image. The processor is also configured to determine spectral-density values of the wavelet descriptors, and to analyze the spectral-density values to generate a mask-wearing-compliance result for the person. In an embodiment, the analyzing includes using a classification model that is trained to classify sets of spectral-density values with respect to facemask wearing in images from which the spectral-density values were derived.

Abstract: A system for adapting at least one parameter of a controller, the system including: processor circuitry; and a non-transitory computer-readable storage medium storing instructions that, when executed by the processor circuitry, cause the processor circuitry to: receive a robot model specification including differentiable robot dynamics, a controller specification including the at least one parameter, and a network condition specification including a non-differentiable discrete communication event; generate a differentiable simulation by: simulating the robot model specification; and transforming the non-differentiable discrete communication event into a continuous differentiable function based on a membership function; and tuning the at least one parameter using gradient-based optimization on the differentiable simulation to improve robot performance under the network condition specification; and output the tuned at least one parameter to configure the controller.

Abstract: Various aspects of techniques, systems, and use cases may be used for operating, controlling, programming, or configuring an energy efficient robotic arm. An example technique may include identifying a planned trajectory for a robotic arm, selecting a candidate joint of the robotic arm to lock, and recomputing the planned trajectory to determine a joint-locked trajectory while the candidate joint is locked. The example technique may include outputting a control signal to lock the candidate joint and causing the robotic arm to use the joint-locked trajectory.

Abstract: An apparatus comprising a memory to store an observed trajectory of a pedestrian, the observed trajectory comprising a plurality of observed locations of the pedestrian over a first plurality of timesteps; and a processor to generate a predicted trajectory of the pedestrian, the predicted trajectory comprising a plurality of predicted locations of the pedestrian over the first plurality of timesteps and over a second plurality of timesteps occurring after the first plurality of timesteps; determine a likelihood of the predicted trajectory based on a comparison of the plurality of predicted locations of the pedestrian over the first plurality of timesteps and the plurality of observed locations of the pedestrian over the first plurality of timesteps; and responsive to the determined likelihood of the predicted trajectory, provide information associated with the predicted trajectory to a vehicle to warn the vehicle of a potential collision with the pedestrian.

Abstract: A component of a system, including: processor circuitry; and a non-transitory computer-readable storage medium including instructions that, when executed by the processing circuitry, cause the processor circuitry to: receive image data of an object captured by a camera; analyze a visual feature of the object based on the received image data; generate illumination patterns based on the analyzed visual feature; and control arrays of light sources integrated into a plurality of fingers of a robotic gripper to project the illumination patterns within a grasp volume defined by the plurality of fingers during object manipulation to enhance detection of the visual feature of the object, wherein each light source in the arrays of light sources is individually controllable.

Abstract: A bidirectional haptic feedback system, including: a flexible membrane configured to be mounted on a handheld controller; sensor-actuator units arranged on the flexible membrane, the sensor-actuator units respectively including a damping mechanism configured to mechanically isolate vibrations between adjacent sensor-actuator units; a control system configured to: generate vibration signals within selected frequency bands within a proximity to a natural resonant frequency range of the sensor-actuator units to drive the actuators of the sensor-actuator units to deliver haptic feedback to a user based on a state of the robot; simultaneously detect user grasp contact and pressure through analysis of back electromotive force (EMF) signals generated by the sensor-actuator units; and adjust robot control parameters dynamically in response to the detected grasp contact and pressure.

Abstract: A transport system, including: a plurality of self-lifting wheel units individually controllable and mounted to a transport platform; one or more sensors mounted to the transport platform and configured to detect a floor obstacle, floor elevation change, or floor surface irregularity; a control system operatively connected to the plurality of self-lifting wheel units and the one or more sensors, wherein the control system is configured to: receive floor obstacle, elevation change, or surface irregularity detection data from the one or more sensors; plan and control the plurality of self-lifting wheel units to selectively lift or lower to maintain stability of the transport platform when traversing the floor obstacle, the floor elevation change, or the floor surface irregularity; and regulate movement of the transport platform to traverse the floor obstacle, the floor elevation change, or the floor surface irregularity based the plan and control.

Abstract: Techniques are disclosed to implement a mathematical framework to model a mechanical actuator such as robotic arm and compute the differential kinematics of an end effector represented by a circle in a three-dimensional space, described as a bi-vector of conformal geometric algebra. Additionally, by using a circle to describe the grasping pose on the object, a differential kinematics-based control scheme is implemented to guide the actuator and minimize the error between the end effector circle and the target circle. The circle has 3 degrees of freedom for the center, two degrees for the orientation, and one more for the radius, which may be used to describe the end effector pose, with the differential kinematics-based control scheme law adjusting the position and the orientations simultaneously.

Abstract: Disclosed herein are systems and methods for vehicle-occupancy-based and user-preference-based smart routing, and autonomous volumetric-occupancy measurement. In an embodiment, a system is configured to receive from a user device associated with a user, a routing-options request for routing options between two locations, and to responsively identify one or more routing options between the two locations based at least in part on occupancy data for a vehicle that would be utilized for at least a portion of at least one of the identified routing options. The occupancy data is based on an output of an automated occupancy-measurement system onboard the vehicle. The system is also configured to provide the one or more identified routing options to the user device. In some embodiments, the occupancy data is obtained using volumetric-occupancy measurement. Some embodiments relate to volumetric-occupancy measurement conducted by autonomous mesh nodes.

Abstract: Various aspects of techniques, systems, and use cases for robot movement within human-robot collaboration (HRC) areas are disclosed. Convex free-space regions around points of interest in the HRC are detected and updated using one or more sensors. Collision-free motion plans for robots use a Hierarchical Convex Polytope (HCP) region in which the robot is removed from the occupied space in the scene using depth cameras.

Abstract: Techniques are disclosed for task error correction for robots, such as collaborative robots (cobots). A controller of a robot may include an error detector to detect an error in a performance of a human-robot collaborative task, and an error corrector to correct the detected error. The error corrector may include a correction planner and a facilitator. The correction planner may determine an error correction plan based on the detected error. The error correction plan may include corrective subtasks to control the cobot to correct the detected error. The facilitator may determine a facilitation plan based on the determined error correction plan. The facilitation plan including an assistance subtask configured to control the cobot to assist a human operator in correcting the detected error. The error corrector may generate a control signal to control the cobot based on the correction plan and the facilitation plan.

Abstract: Various aspects of techniques, systems, and use cases may be used for probabilistic automatic determination of an action for a robotic device. A technique may include identifying a current context of a robotic device, determining from the current context, a set of actions performable by the robotic device corresponding to at least one object, the set of actions including one or more affordances generated from a basic skills library for the robotic device, and automatically selecting an action of the set of actions based on an acyclic graph describing action paths. The technique may include outputting control signals that, when executed, cause the robotic device to perform the action to interact with the at least one object.

Abstract: A component of an edge server, 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: distribute, to a client device, tokens that enable its end device to execute respective asynchronous adaptive motion primitives (A2MPs) of a task graph of a task, wherein an A2MP is a motion primitive of encoded motion factoring in motion updates from the end device; receive A2MP task execution status messages during execution of the A2MPs; and dynamically update the distribution of the token or the task graph based on the A2MP task execution status messages to modify a trajectory of the end device.

Abstract: Various aspects of techniques, systems, and use cases for selecting grasping configurations for a robot are disclosed. Geometric primitives are generated to model the robot for grasping and manipulation by the robot. The geometric primitives are combined using various functions to determine which configuration to use. The instantaneous configuration is determined, as well as the forward kinematics and links to determine active geometric primitives of the gripper. The active geometric primitives are used to approximate an x, y, and z coordinate of each point of the primitives, a distance between the point and a grasping target, and an associated surface link. The configurations are ranked based on grasping metrics and one of the configurations selected to use accordingly.

Abstract: A robotic manipulation planning 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: process perception data to detect known, familiar, and unknown objects to generate manipulation candidates; filter manipulation candidates against constraints to reduce the manipulation candidates; and determine quality metrics for the reduced manipulation candidates using a soft-body simulation technique.

Abstract: A safety system for a vehicle may include one or more processors configured to determine, based on a friction prediction model, one or more predictive friction coefficients between the ground and one or more tires of the ground vehicle using first ground condition data and second ground condition data. The first ground condition data represent conditions of the ground at or near the position of the ground vehicle, and the second ground condition data represent conditions of the ground in front of the ground vehicle with respect to a driving direction of the ground vehicle. The one or more processors are further configured to determine driving conditions of the ground vehicle using the determined one or more predictive friction coefficients.

Abstract: Various aspects of techniques, systems, and use cases may be used for camera and end-effector planning for visual servoing for example in redundant manipulators. A technique may include generating a set of paths of an end effector of a robotic arm, and ranking the set of paths using an objective function that results in an improvement to a distance between the end effector and a target while maintaining a collision mitigation path for the end effector and minimizing occlusion of a camera affixed to a joint of the robotic arm. The technique may include converting a path of the set of paths into a trajectory based on the ranking, and outputting the trajectory for controlling the robotic arm to move the end effector.