Data-driven prediction-based system and method for trajectory planning of autonomous vehicles

Patent metadata

Field	Value
Publication number	US-11853071-B2
Application number	US-202016848809-A
Country	US
Kind code	B2
Filing date	Apr 14, 2020
Priority date	Sep 7, 2017
Publication date	Dec 26, 2023
Grant date	Dec 26, 2023

Abstract

A data-driven prediction-based system and method for trajectory planning of autonomous vehicles are disclosed. A particular embodiment includes: generating a first suggested trajectory for an autonomous vehicle; generating predicted resulting trajectories of proximate agents using a prediction module; scoring the first suggested trajectory based on the predicted resulting trajectories of the proximate agents; generating a second suggested trajectory for the autonomous vehicle and generating corresponding predicted resulting trajectories of proximate agents, if the score of the first suggested trajectory is below a minimum acceptable threshold; and outputting a suggested trajectory for the autonomous vehicle wherein the score corresponding to the suggested trajectory is at or above the minimum acceptable threshold.

First claim

What is claimed is:

1. A system comprising:
a data processor; and
a trajectory planning module, executable by the data processor, configured to:
generate a first suggested trajectory for an autonomous vehicle, the first suggested trajectory configured to comply with pre-defined goals for the autonomous vehicle;
generate a first distribution of predicted resulting trajectories for each of proximate agents using a prediction module configured to generate a first output in response to receiving the first suggested trajectory as an input, the first distributions of predicted resulting trajectories being reactions to the first suggested trajectory of the autonomous vehicle and generated based on the output of the prediction module fed back to the trajectory planning module, wherein the proximate agents are vehicles in the vicinity of the autonomous vehicle, wherein the first distribution of predicted resulting trajectories includes a first set of predicted trajectories with first probability values obtained based on the first suggested trajectory, wherein each of predicted trajectories in the first distribution has a corresponding first probability value obtained based on the first suggested trajectory;
score the first suggested trajectory based on each corresponding first probability value and determine whether a score corresponding to a highest first probability value of the first probability values complies with the pre-defined goals of the autonomous vehicle;
generate a second suggested trajectory for the autonomous vehicle upon determination that the score for the first suggested trajectory corresponding to the highest first probability value is below a minimum acceptable threshold, or
cause a vehicle control subsystem of the autonomous vehicle to manipulate the autonomous vehicle to follow the first suggested trajectory, upon determination that the score for the first suggested trajectory corresponding to the highest first probability value is at or above the minimum acceptable threshold;
generate, upon determination that the score for the first suggested trajectory corresponding to the highest first probability value is below the minimum acceptable threshold, an updated first distribution of the predicted resulting trajectories for each of proximate agents using the prediction module configured to generate a second output in response to receiving the second suggested trajectory as an input, the updated first distribution of the predicted resulting trajectories being reactions to the second suggested trajectory of the autonomous vehicle and generated based on the second output of the prediction module fed back to the trajectory planning module, wherein the updated first distribution of the predicted resulting trajectories includes the first set of predicted trajectories with second probability values obtained based on the second suggested trajectory, wherein each of the predicted trajectories in the updated first distribution has a corresponding second probability value obtained based on the second suggested trajectory; and
cause the vehicle control subsystem of the autonomous vehicle to manipulate the autonomous vehicle to follow the second suggested trajectory, upon determination that any of the second probability values is higher than the highest first probability value.

2. The system of claim 1 wherein the first distribution of predicted resulting trajectories is generated to predict behaviors and trajectories of proximate vehicles.

3. The system of claim 2 wherein the prediction module is trained with machine learning using a human driving behavior data.

4. The system of claim 1 wherein the first distributions of predicted resulting trajectories are based on an input object data corresponding to the proximate agents and the first suggested trajectory, wherein the predicted resulting trajectories of each of the first distributions has an associated confidence level.

5. The system of claim 4 wherein the trajectory planning module is configured to generate the updated first distribution of predicted resulting trajectories for each of the proximate agents using the prediction module, wherein the updated first distribution of predicted resulting trajectories is based on the input object data corresponding to the proximate agents and the second suggested trajectory, wherein the predicted resulting trajectories of the updated first distribution has an associated confidence level.

6. The system of claim 5 wherein the trajectory planning module is configured to repeatedly generate distributions of predicted resulting trajectories and associated confidence levels, until the score corresponding to the suggested trajectory is at or above the minimum acceptable threshold or until a time period or iteration count is exceeded.

7. The system of claim 1 wherein the trajectory planning module is configured to output a suggested trajectory for the autonomous vehicle when the score corresponding to the suggested trajectory is at or above the minimum acceptable threshold.

8. The system of claim 1 wherein the trajectory planning module is configured to output a suggested trajectory for the autonomous vehicle according to the predicted resulting trajectories of the proximate agents.

9. The system of claim 1 wherein the trajectory planning module and the prediction module form a feedback loop.

10. A method comprising:
generating, by use of a data processor, a first suggested trajectory for an autonomous vehicle, the first suggested trajectory configured to comply with pre-defined goals for the autonomous vehicle;
generating, by use of the data processor, a first distribution of predicted resulting trajectories for each of proximate agents using a prediction module configured to generate a first output in response to receiving the first suggested trajectory as an input, the first distributions of predicted resulting trajectories being reactions to the first suggested trajectory of the autonomous vehicle and generated based on the output of the prediction module fed back to a trajectory planning module, wherein the proximate agents are vehicles in the vicinity of the autonomous vehicle, wherein the first distribution of predicted resulting trajectories includes a first set of predicted trajectories with first probability values obtained based on the first suggested trajectory, wherein each of predicted trajectories in the first distribution has a corresponding first probability value obtained based on the first suggested trajectory;
scoring, by use of the data processor, the first suggested trajectory based on each corresponding first probability value and determine whether a score corresponding to a highest first probability value of the first probability values complies with the pre-defined goals of the autonomous vehicle;
generating, by use of the data processor, a second suggested trajectory for the autonomous vehicle upon determination that the score for the first suggested trajectory corresponding to the highest first probability value is below a minimum acceptable threshold, or
causing a vehicle control subsystem of the autonomous vehicle to manipulate the autonomous vehicle to follow the first suggested trajectory, upon determination that the score for the first suggested trajectory corresponding to the highest first probability value is at or above the minimum acceptable threshold;
generating, upon determination that the score for the first suggested trajectory corresponding to the highest first probability value is below the minimum acceptable threshold, an updated first distribution of the predicted resulting trajectories for each of proximate agents using the prediction module configured to generate a second output in re

Assignees

• Tusimple Inc

Inventors

• Sun Xing
• Lin Wutu
• Liu Liu
• Ma Kai-Chieh
• Xuan Zijie
• Zhao Yufei

Classifications

• B60W60/00276 Primary

  for two or more other traffic participants · CPC title

• G05D1/0221 Primary

  involving a learning process · CPC title

• B62D15/026

  combined with automatic distance control, i.e. electronic tow bar · CPC title

• B62D15/0255

  Automatic changing of lane, e.g. for passing another vehicle · CPC title

• B62D15/0265

  Automatic obstacle avoidance by steering · CPC title