# US 2021/0276598 A1 · Machine-learning based system for path and/or motion planning and method of training the same > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2021/0276598 A1 - Document ID: US20210276598A1 - Status: Published application - Kind: A1 - Application number: 16/810,552 - Filed: 2020-03-05 - Published: 2021-09-09 - Earliest claimed: 2020-03-05 - Assignees: Huawei Technologies Co., Ltd. - Inventors: Elmira AMIRLOO ABOLFATHI; Mohsen ROHANI; Jason Philip KU; Jun LUO - Web page: https://patentgrep.com/patent/US20210276598A1 ## CPC - B60W 30/09: Taking automatic action to avoid collision, e.g. braking and steering - B60W 60/00272: {relying on extrapolation of current movement} - B60W 60/00276: {for two or more other traffic participants} - G06V 10/764: using classification, e.g. of video objects - G06V 10/774: Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting - G06V 10/82: using neural networks - G06V 20/58: Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads - G06V 20/588: {Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road} - B60W 2420/403: Image sensing, e.g. optical camera - B60W: CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT - G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING ## Abstract A system and method for path and/or motion planning and for training such a system are described. In one aspect, the method comprises generating a sequence of predicted occupancy grid maps (OGMs) for T-T1 time steps based on a sequence of OGMs for 0-T1 time steps, a reference map of an environment in which an autonomous vehicle is operating, and a trajectory. A cost volume is generated for the sequence of predicted OGMs. The cost volume comprises a plurality of cost maps for T-T1 time steps. Each cost map corresponds to a predicted OGM in the sequence of predicted OGMs and has the same dimensions as the corresponding predicted OGM. Each cost map comprises a plurality of cells. Each cell in the cost map represents a cost of the cell in corresponding predicted OGM being occupied in accordance with a policy defined by a policy function. ## Claims ### Claim 1 A processor system, comprising: a processing device; a memory storing instructions which when executed by the processing device cause the processing system to: generate a sequence of predicted occupancy grid maps (OGMs) for T-T1 time steps based on a sequence of OGMs for 0-T1 time steps, a reference map of an environment in which an autonomous vehicle is operating, and a trajectory; and generate a cost volume for the sequence of predicted OGMs, the cost volume comprising a plurality of cost maps for T time steps, each cost map corresponding to a predicted OGM in the sequence of predicted OGMs and having the same dimensions as the corresponding predicted OGM, each cost map comprising a plurality of cells, each cell in the cost map representing a cost of the cell in corresponding predicted OGM being occupied in accordance with a policy defined by a policy function. ### Claim 2 (depends on 1) The processor system of claim 1, wherein the memory stores further instructions which when executed by the processing device cause the processing device to: generate a set of trajectories; generate a sequence of predicted OGMs and a cost volume for each trajectory in the set of trajectories; determine a cost of each trajectory in the set of the trajectories based on the corresponding cost volume; and select the trajectory with the lowest cost. ### Claim 3 (depends on 2) The computing device of claim 2, wherein the memory stores further instructions which when executed by the processing device cause the processing device to generate the set of trajectories based on a current position and a current velocity of the autonomous vehicle. ### Claim 4 (depends on 2) The processor system of claim 2, wherein memory stores further instructions which when executed by the processing device cause the processing device to determine the cost of each trajectory in the set of the trajectories by determining a cost for each time step in each trajectory in the set of the trajectories in accordance with the corresponding cost map for the time step and summing the cost of each time step for each trajectory in the set of the trajectories. ### Claim 5 (depends on 1) The processor system of claim 1, memory stores further instructions which when executed by the processing device cause the processing device to generate control signals in accordance with the selected trajectory, wherein the control signals define one or more of a steering angle for a steering unit, a throttle value for a throttle unit and braking value for a braking unit. ### Claim 6 (depends on 5) The processor system of claim 5, wherein the memory stores further instructions which when executed by the processing device cause the processing device to send the control signals to an electromechanical system, wherein the control signals are configured to cause the electromechanical system to control movement of a host vehicle in accordance with the control signals. ### Claim 7 (depends on 1) The processor system of claim 1, wherein memory stores further instructions which when executed by the processing device cause the processing device to generate the sequence of predicted OGMs from a sequence of observed OGMs. ### Claim 8 (depends on 1) The processor system of claim 1, wherein the memory stores further instructions which when executed by the processing device cause the processing device to generate the sequence of predicted OGMs based on features extracted from the observed OGMs. ### Claim 9 (depends on 8) The processor system of claim 8, further comprising: a sensor system configured to: receive sensor data about the sensed environment from one or more sensors; and generate the sequence of observed OGMs based on the sensor data. ### Claim 10 (depends on 1) The processor system of claim 1, wherein the memory stores further instructions which when executed by the processing device cause the processing device to generate the sequence of predicted OGMs based on features extracted from the reference map. ### Claim 11 (depends on 1) The processor system e of claim 1, wherein the memory stores further instructions which when executed by the processing device cause the processing device to generate the sequence of predicted OGMs based on determined motion-flow between the observed OGMs and previous OGMs. ### Claim 12 (depends on 11) The processor system of claim 11, wherein the memory stores further instructions which when executed by the processing device cause the processing device to generate the sequence of predicted OGMs based on features extracted from the determined motion-flow between the observed OGMs and previous OGMs. ### Claim 13 (depends on 1) The processor system of claim 1, wherein the cost volume has a size W×H×Tpath, wherein W×H is the dimension of the predicted OGMs, wherein W is the width and H is the height of the predicted OGMs, wherein Tpath=T-T1. ### Claim 14 (depends on 1) The processor system of claim 1, wherein the policy function is based on one or more predefined rules. ### Claim 15 (depends on 14) The processor system of claim 14, wherein the one or more predefined rules comprise one or more driving rules and/or one or more environmental rules. ### Claim 16 (depends on 15) The processor system of claim 15, wherein the driving rules comprise any one of more of vehicle collision avoidance, lane following, speed limit following, maintaining a threshold distance from other vehicles, or traffic law/rule following. ### Claim 17 (depends on 15) The processor system of claim 15, wherein the environmental rules comprise any one of more of remaining within a drivable area, remaining with lanes, object collision avoidance, maintaining a threshold distance from a perimeter of the drivable area, maintaining a threshold distance from lane markings, or maintaining a threshold distance from environmental objects. ### Claim 18 A method, comprising: generating a sequence of predicted occupancy grid maps (OGMs) for T time steps based on a sequence of OGMs for 0-T1 time steps, a reference map of an environment in which an autonomous vehicle is operating, and a trajectory; and generating a cost volume for the sequence of predicted OGMs, the cost volume comprising a plurality of cost maps for T-T1 time steps, each cost map corresponding to a predicted OGM in the sequence of predicted OGMs and having the same dimensions as the corresponding predicted OGM, each cost map comprising a plurality of cells, each cell in the cost map representing a cost of the cell in corresponding predicted OGM being occupied in accordance with a policy defined by a policy function. ### Claim 19 A non-transitory machine readable medium having tangibly stored thereon executable instructions that, in response to execution by a processing system, cause the processing system to: generating a sequence of predicted occupancy grid maps (OGMs) for T-T1 time steps based on a sequence of OGMs for 0-T1 time steps, a reference map of an environment in which an autonomous vehicle is operating, and a trajectory; and generate a cost volume for the sequence of predicted OGMs, the cost volume comprising a plurality of cost maps for T T1 time steps, each cost map corresponding to a predicted OGM in the sequence of predicted OGMs and having the same dimensions as the corresponding predicted OGM, each cost map comprising a plurality of cells, each cell in the cost map representing a cost of the cell in corresponding predicted OGM being occupied in accordance with a policy defined by a policy function. ### Claim 20 A method, comprising: generating a sequence of predicted occupancy grid maps (OGMs) for a set of trajectories and an expert trajectory for T-T1 time steps based on a sequence of OGMs for 0-T1 time steps, a reference map of an environment in which an autonomous vehicle is operating, the set of trajectories and the expert trajectory; generating by a neural network a cost volume for the sequence of predicted OGMs for each trajectory in the set of trajectories and the expert trajectory, each cost volume comprise a plurality of cost maps for T-T1 time steps, each cost map corresponding to a predicted OGM in the sequence of predicted OGMs and having the same dimensions as the corresponding predicted OGM, each cost map comprising a plurality of cells, each cell in the cost map representing a cost of the cell in corresponding predicted OGM being occupied in accordance with a policy defined by a policy function; determining an imitation cost for each trajectory in the set of trajectories by summing an imitation cost of each point for each time step; determining an imitation loss for the trajectory with a lowest imitation cost in accordance with a loss function which determines the imitation loss based on whether the state of the cells in the predicted OGM for the proposed trajectory differs from the state of the cells in the predicted OGM for the expert trajectory; determining an environmental cost for each trajectory in the set of trajectories by summing an environmental cost of each point for each time step; determining an environmental loss for the trajectory with a lowest environmental cost in accordance with a loss function which determines the imitation loss based on whether the state of the cells in the predicted OGM for the proposed trajectory differs from the reference map; determining whether a combination of the loss functions for the imitation loss and the environmental loss are below a threshold; and updating parameters of the neural network in response to a determination that the imitation loss and environmental loss is not below a threshold.