An output of a first one of a plurality of layers within a neural network is identified. A bitmap is determined from the output, the bitmap including a binary matrix. A particular subset of operations for a second one of the plurality of layers is determined to be skipped based on the bitmap. Operations are performed for the second layer other than the particular subset of operations, while the particular subset of operations are skipped.

Systems and methods for performing visual odometry more rapidly. Pairs of representations from sensor data (such as images from one or more cameras) are selected, and features common to both representations of the pair are identified. Portions of bundle adjustment matrices that correspond to the pair are updated using the common features. These updates are maintained in register memory until all portions of the matrices that correspond to the pair are updated. By selecting only common features of one particular pair of representations, updated matrix values may be kept in registers. Accordingly, matrix updates for each common feature may be collectively saved with a single write of the registers to other memory. In this manner, fewer write operations are performed from register memory to other memory, thus reducing the time required to update bundle adjustment matrices and thus speeding the bundle adjustment process.

A system for creating synthetic data for testing an autonomous system, comprising at least one hardware processor adapted to execute a code for: using a machine learning model to compute a plurality of depth maps based on a plurality of real signals captured simultaneously from a common physical scene, each of the plurality of real signals are captured by one of a plurality of sensors, each of the plurality of computed depth maps qualifies one of the plurality of real signals; applying a point of view transformation to the plurality of real signals and the plurality of depth maps, to produce synthetic data simulating a possible signal captured from the common physical scene by a target sensor in an identified position relative to the plurality of sensors; and providing the synthetic data to at least one testing engine to test an autonomous system comprising the target sensor.

An interactive autonomous driving system for an autonomous driving vehicle may include: a target mapping device that determines whether an obstacle is present in a predetermined range of a target selected by a passenger and outputting obstacle information; a target attribute determination device that determines a target attribute based on the obstacle information and outputs target controllable item information; and a processor that generates control mode recommendation information selectable by the passenger based on the target controllable item information and outputs target attribute information and a selected control mode when control mode selection information is received from the passenger.

Systems and methods described herein pertain to maintaining a virtual representation of a workspace in a material handling system and updating the virtual representation without downtime. Methods described include maintaining an initial virtual representation of a material handling grid, receiving an updated virtual representation, and generating and implementing an intermediate virtual representation that does not conflict with the initial virtual representation. Methods further include, upon determining that the intermediate virtual representation is performing without conflicts, deploying the updated virtual representation to replace the intermediate virtual representation without halting operations in the workspace. Multiple intermediate virtual representations can be generated to allow for complex changes, and the deployments performed in series.

A vehicle control device includes a detection unit detecting an object around the vehicle, a storage unit storing static object information acquired in advance and reliability of the static object information in association with each other, an acquisition unit acquiring calculation load information of the vehicle control device, a decision unit deciding a reliability threshold to be small with respect to an increase in a calculation load based on the calculation load information, a selection unit selecting the static object information associated with reliability equal to or greater than the reliability threshold from the static object information, a target decision unit deciding a tracking target by comparing a detection result and the static object information selected by the selection unit to each other, a tracking unit tracking the tracking target, and a control unit performing the traveling control based on a tracking result.

High definition maps for autonomous vehicles are very high resolution and detailed, and hence require storage of a great deal of data. A vehicle computing system provides multi-layered caching makes this data usable in a system that requires very low latency on every operation. The system determines which routes are most likely to be driven in the near future by the car, and ensures that the route is cached on the vehicle before beginning the route. The system provides efficient formats for moving map data from server to car and for managing the on-car disk. The system further provides real-time accessibility of nearby map data as the car moves, while providing data access at optimal speeds.

A method for automatic determination and/or monitoring of a target trajectory for a vehicle by which a starting point corresponding to the current position of the vehicle is connected to a target point. The method includes determining different trajectories of the vehicle that connect the starting point to the target point, detecting a further target trajectory for each road user, wherein each of the further target trajectories connects the starting point of the respective road user to a target point corresponding to the respective road user, determining the trajectories of the vehicle as collision-free trajectories that do not result in a collision with one of the further road users if the respective road user is moving on the target trajectory, and determining and/or monitoring the target trajectory of the vehicle depending on the collision-free trajectories of the vehicle.

Provided is a robot-implemented, real-time, process to plan a coverage path, the process including: obtaining environment-sensor data indicating distances from the robot to surfaces in a portion of a working environment; obtaining odometry-sensor data; based on the environment-sensor data and the odometry-sensor data, determining at least a part of a coverage path of the robot through the working environment; and commanding an electric-motor driver to move the robot along the at least part of the path.

A system to reduce vehicle resource depletion risk which includes a memory, controller, efficiency module, mobile computing device, and fleet vehicle. The memory includes executable instructions. The controller executes the instructions. The controller communicates with an efficiency module. The efficiency module causes a fleet vehicle to optimally perform a rideshare task. The mobile computing device generates first location data and communicates the first location data to the controller. The fleet vehicle includes a vehicle system and a vehicle controls device and can communicate with the controller. The vehicle system generates second location data. The vehicle controls device commands the fleet vehicle to perform a rideshare task. The instructions enable the controller to: receive the first and second location data; perform the efficiency module to produce an output being partially based on the first and second location data and instructs the vehicle to perform a rideshare task; and communicate the output.