Vehicle path planning

Patent metadata

Field	Value
Publication number	US-2022063651-A1
Application number	US-202017004342-A
Country	US
Kind code	A1
Filing date	Aug 27, 2020
Priority date	Aug 27, 2020
Publication date	Mar 3, 2022
Grant date	—

Abstract

A computer, including a processor and a memory, the memory including instructions to be executed by the processor to calibrate utility functions that determine optimal vehicle actions based on an approximate Nash equilibrium solution for multiple agents by determining a difference between model-predicted future states for the multiple agents to observed states for the multiple agents. The instructions can include further instructions to determine a vehicle path for a vehicle based on the optimal vehicle actions.

First claim

1 . A computer, comprising:
a processor; and
a memory, the memory including instructions executable by the processor to:
calibrate utility functions that determine optimal vehicle actions based on an approximate Nash equilibrium solution for multiple agents by determining a difference between possible model-predicted states for the multiple agents and observed states for the multiple agents; and
determine a vehicle path for a vehicle based on the optimal vehicle actions.

2 . The computer of claim 1 , wherein the difference between the possible future states for the agents and the observed states includes a noise term that is normally distributed with a constant covariance.

3 . The computer of claim 1 , the instructions including further instructions to calibrate the utility functions by minimizing a cost function determined by the difference between the model-predicted states and the observed states.

4 . The computer of claim 1 , the instructions including further instructions to determine the observed states based on sensor data acquired from sensors included in a traffic infrastructure system arranged to observe the multiple agents.

5 . The computer of claim 1 , wherein the utility functions simulate behavior of the multiple agents to determine the possible future states for the multiple agents based on determining one or more of each of agents' locations, agents' velocities, where velocity includes speed and heading, and one or more possible paths for each agent.

6 . The computer of claim 5 , wherein the utility functions includes parameters that determine rewards and penalties for actions of each of the multiple agents based on estimated states of the multiple agents at future time steps t included within a time horizon h.

7 . The computer of claim 1 , wherein the approximate Nash equilibrium solution performs an adaptive grid search optimization technique to determine the optimal vehicle actions based on estimating the possible future states of the multiple agents, wherein the multiple agents include one or more of autonomous vehicles, non-autonomous vehicles, stationary objects, and non-stationary objects including pedestrians and the possible future states are estimated by simulating behavior of the multiple agents based on the utility functions to determine the possible future states for the multiple agents.

8 . The computer of claim 1 , wherein the utility functions include one or more of moving forward at a desired speed and deviating from smooth vehicle operation, wherein the smooth vehicle operation includes limits on agent acceleration, agent steering and agent braking.

9 . The computer of claim 1 , wherein the utility functions include one or more of lane departure, out of roadway departure, collisions with stationary objects, and collisions with non-stationary objects.

10 . The computer of claim 1 , wherein the vehicle path based on the optimal vehicle actions is determined based on polynomial functions.

11 . The computer of claim 1 , the instructions including further instructions to determine the vehicle path for the vehicle based on the optimal vehicle actions and to download the vehicle path to a second computer including a second processor and second memory included in the vehicle.

12 . The computer of claim 11 , wherein the second computer includes instructions to operate the vehicle along the vehicle path by controlling vehicle powertrain, vehicle steering and vehicle brakes.

13 . A method, comprising:
calibrating utility functions that determine optimal vehicle actions based on an approximate Nash equilibrium solution for multiple agents by determining a difference between model-predicted states for the multiple agents and observed states for the multiple agents; and determining a vehicle path for a vehicle based on the optimal vehicle actions.

14 . The method of claim 13 , wherein the difference between the model-predicted states for the agents and the observed states includes a noise term that is normally distributed with a constant covariance.

15 . The method of claim 13 , further comprising calibrating the utility functions by minimizing a cost function determined by the difference between the possible model-predicted states to the observed states.

16 . The method of claim 13 , further comprising determining the observed states based on sensor data acquired from sensors included in a traffic infrastructure system arranged to observe the multiple agents.

17 . The method of claim 13 , wherein the utility functions simulate behavior of the multiple agents to determine the possible future states for the multiple agents based on determining one or more of each of agents' locations, agents' velocities, where velocity includes speed and heading, and one or more possible paths for each agent.

18 . The method of claim 17 , wherein the utility functions includes parameters that determine rewards and penalties for actions of each of the multiple agents based on estimated states of the multiple agents at future time steps t included within a time horizon h.

19 . The method of claim 13 , wherein the approximate Nash equilibrium solution performs an adaptive grid search optimization technique to determine the optimal vehicle actions based on estimating the possible future states of the multiple agents, wherein the multiple agents include one or more of autonomous vehicles, non-autonomous vehicles, stationary objects, and non-stationary objects including pedestrians and the possible future states are estimated by simulating behavior of the multiple agents based on the utility functions to determine the possible future states for the multiple agents.

20 . The method of claim 13 , wherein the utility functions include one or more of moving forward at a desired speed and deviating from smooth vehicle operation, wherein the smooth vehicle operation includes limits on agent acceleration, agent steering and agent braking.

Assignees

• Ford Global Tech Llc

Inventors

• Dai Qi
• Wang Jinhong
• Guo Wen
• Xu Xunnong
• Huang Suzhou
• Filev Dimitar Petrov

Classifications

• G08G1/096816

  where the complete route is transmitted to the vehicle at once · CPC title

• G08G1/056

  with provision for distinguishing direction of travel · CPC title

• G08G1/0133

  for classifying traffic situation · CPC title

• G08G1/0116

  from roadside infrastructure, e.g. beacons · CPC title

• B60W2050/0083

  Setting, resetting, calibration · CPC title