Vehicle control method and apparatus, device, storage medium and autonomous vehicle

Patent metadata

Field	Value
Publication number	US-12472985-B2
Application number	US-202318112475-A
Country	US
Kind code	B2
Filing date	Feb 21, 2023
Priority date	Mar 4, 2022
Publication date	Nov 18, 2025
Grant date	Nov 18, 2025

Abstract

A vehicle control technical solution is provided. The solution relates to the field of computer technologies, and particularly to the field of artificial intelligence technologies, such as autonomous driving technologies, intelligent transportation technologies. A vehicle control method includes: acquiring a state quantity error between a desired state quantity at a current moment and a real state quantity at a previous moment of a vehicle; determining a control quantity of the vehicle at the current moment based on the state quantity error and a mapping relationship between the control quantity of the vehicle and the state quantity error; and controlling a driving behavior of the vehicle based on the control quantity at the current moment to obtain a real state quantity at the current moment.

First claim

What is claimed is:

1 . A vehicle control method, comprising:
acquiring, by an adaptive control module, an acceleration error between a desired acceleration at a current moment and a real acceleration at a previous moment of a vehicle, wherein the real acceleration at the previous moment is output by a chassis system of the vehicle, the desired acceleration at the current moment is obtained by the vehicle controller according to a reference signal and the real acceleration at the previous moment and output by the vehicle controller; determining, by the adaptive control module, a control quantity of the vehicle at the current moment based on the acceleration error and a mapping relationship between the control quantity of the vehicle and the acceleration error, and outputting the control quantity of the vehicle at the current moment to the chassis system; and controlling, by the chassis system, the driving behavior of the vehicle based on the control quantity at the current moment to obtain the real acceleration at the current moment, wherein the mapping relationship is a neural network model, and the determining the control quantity of the vehicle at the current moment based on the acceleration error and the mapping relationship between the control quantity of the vehicle and the acceleration error comprises:
determining a filtering error and input data based on the acceleration error;
processing the input data using the neural network model at the current moment to obtain output data; and
determining the control quantity of the vehicle at the current moment based on the filtering error and the output data;
updating the neural network model at the previous moment based on the filtering error to determine the neural network model at the current moment, wherein model parameters of the neural network model comprise a first parameter and a second parameter, the first parameter is a fixed value, and the second parameter is an updatable parameter; and the updating the neural network model at the previous moment based on the filtering error to determine the neural network model at the current moment comprises:
updating the second parameter at the previous moment based on the filtering error to determine the second parameter at the current moment.

2 . The method according to claim 1 , further comprising:
obtaining the desired acceleration of the vehicle at the current moment based on model prediction control MPC, the desired acceleration at the current moment being used for determining the acceleration error.

3 . An electronic device, comprising:
at least one processor; and a memory connected with the at least one processor communicatively; wherein the memory stores instructions executable by the at least one processor to cause the at least one processor to perform a vehicle control method comprising: acquiring, by an adaptive control module, an acceleration error between a desired acceleration at a current moment and a real acceleration at a previous moment of a vehicle, wherein the real acceleration at the previous moment is output by a chassis system of the vehicle, the desired acceleration at the current moment is obtained by the vehicle controller according to a reference signal and the real acceleration at the previous moment and output by the vehicle controller; determining, by the adaptive control module, a control quantity of the vehicle at the current moment based on the acceleration error and a mapping relationship between the control quantity of the vehicle and the acceleration error, and outputting the control quantity of the vehicle at the current moment to the chassis system; and controlling, by the chassis system, the driving behavior of the vehicle based on the control quantity at the current moment to obtain the real acceleration at the current moment, wherein the mapping relationship is a neural network model, and the determining the control quantity of the vehicle at the current moment based on the acceleration error and the mapping relationship between the control quantity of the vehicle and the acceleration error comprises:
determining a filtering error and input data based on the acceleration error;
processing the input data using the neural network model at the current moment to obtain output data; and
determining the control quantity of the vehicle at the current moment based on the filtering error and the output data;
updating the neural network model at the previous moment based on the filtering error to determine the neural network model at the current moment, wherein model parameters of the neural network model comprise a first parameter and a second parameter, the first parameter is a fixed value, and the second parameter is an updatable parameter; and the updating the neural network model at the previous moment based on the filtering error to determine the neural network model at the current moment comprises:
updating the second parameter at the previous moment based on the filtering error to determine the second parameter at the current moment.

4 . The electronic device according to claim 3 , wherein the method further comprises:
obtaining the desired acceleration of the vehicle at the current moment based on model prediction control MPC, the desired acceleration at the current moment being used for determining the acceleration error.

5 . An autonomous vehicle, comprising:
a vehicle controller, configured for outputting a desired acceleration at a current moment; a chassis system, configured for outputting a real acceleration at a previous moment; wherein the desired acceleration at the current moment is obtained by the vehicle controller according to a reference signal and the real acceleration at the previous moment; an adaptive control module, configured for acquiring an acceleration error between the desired acceleration at the current moment and the real acceleration at the previous moment, and determining a control quantity of the vehicle at the current moment based on the acceleration error and a mapping relationship between the control quantity of the vehicle and the acceleration error, and outputting the control quantity of the vehicle to the chassis system, wherein the chassis system controls the driving behavior of the vehicle based on the control quantity at the current moment to obtain the real acceleration at the current moment, wherein the mapping relationship is a neural network model, and the adaptive control module is specifically configured for: determining a filtering error and input data based on the acceleration error; processing the input data using the neural network model at the current moment to obtain output data; and determining the control quantity of the vehicle at the current moment based on the filtering error and the output data, wherein the adaptive control module is further configured for: updating the neural network model at the previous moment based on the filtering error to determine the neural network model at the current moment, wherein model parameters of the neural network model comprise a first parameter and a second parameter, the first parameter is a fixed value, and the second parameter is an updatable parameter; and the updating the neural network model at the previous moment based on the filtering error to determine the neural network model at the current moment comprises: updating the second parameter at the previous moment based on the filtering error to determine the second parameter at the current moment.

Assignees

• Apollo Intelligent Driving Technology Beijing Co Ltd

Inventors

• Liang Qi
• Tan Yinong
• Liu Zhengyu
• Li Liutao

Classifications

• G06N3/08

  Learning methods · CPC title

• B60W2050/0075

  Automatic parameter input, automatic initialising or calibrating means · CPC title

• B60W50/0205

  Diagnosing or detecting failures; Failure detection models · CPC title

• B60T2260/06

  Active Suspension System · CPC title

• B60T17/22

  Devices for monitoring or checking brake systems; Signal devices · CPC title