Smooth cooperative lane change control method for multi-connected and autonomous vehicle (CAV)

Patent metadata

Field	Value
Publication number	US-12134386-B2
Application number	US-202218055160-A
Country	US
Kind code	B2
Filing date	Nov 14, 2022
Priority date	Nov 14, 2022
Publication date	Nov 5, 2024
Grant date	Nov 5, 2024

Abstract

A smooth cooperative lane change control method for multi-connected and autonomous vehicles (CAVs), including: acquiring vehicle information of a lane-changing vehicle M and four surrounding vehicles A, B, C and D; constructing uncontrolled-vehicle and controlled-vehicle motion state prediction models; according to the motion state prediction models, predicting motion states of the lane-changing vehicle M, and the vehicles A, B, C and D; constructing an upper-layer optimization model to calculate an optimal control value and an optimized motion state of the lane-changing vehicle M and an optimal control value of the vehicle A; constructing a lower-layer optimization model to calculate an optimal control value of the vehicle D; and controlling the lane-changing vehicle M, the vehicle A and the vehicle D to run according to a corresponding optimal control value.

First claim

What is claimed is:

1. A smooth cooperative lane change control method for a multi-connected and autonomous vehicle (CAV), comprising:
(a) acquiring vehicle information of a lane-changing vehicle M and four surrounding vehicles thereof, wherein the four surrounding vehicles are vehicle A, vehicle B, vehicle C and vehicle D, respectively; a lane where the lane-changing vehicle M is currently located is defined as a main lane, and the vehicle C and the vehicle D are located in the main lane; the vehicle C is located in front of the lane-changing vehicle M, and the vehicle D is located behind the lane-changing vehicle M; the vehicle A and the vehicle B are located in a target lane; the vehicle B is located in front of the lane-changing vehicle M, and the vehicle A is located behind the lane-changing vehicle M; the vehicle D and the vehicle A are configured as cooperative lane-changing vehicles of the lane-changing vehicle M; the vehicle D and the vehicle A are controlled to behave cooperatively with the lane-changing vehicle M during a lane changing process, and the vehicle B and the vehicle C are uncontrolled; the lane-changing vehicle M, the vehicle D and the vehicle A are each a connected and autonomous vehicle; and the vehicle B and the vehicle C are each a human-driven vehicle (HDV);
(b) according to vehicle information and parameters of the vehicle B and the vehicle C, constructing an uncontrolled-vehicle motion state prediction model; and according to vehicle information and parameters of the lane-changing vehicle M, the vehicle A and the vehicle D, constructing a controlled-vehicle motion state prediction model; and
according to the uncontrolled-vehicle motion state prediction model, predicting a motion state of the vehicle B and a motion state of the vehicle C; and according to the controlled-vehicle motion state prediction model, predicting a motion state of the lane-changing vehicle M, a motion state of the vehicle A, and a motion state of the vehicle D;
wherein the uncontrolled-vehicle motion state prediction model comprises a long short-term memory (LSTM)-based neural network model; and the controlled-vehicle motion state prediction model comprises a vehicle kinematic model, and the vehicle kinematic model comprises a steering kinematic model and a longitudinal kinematic model;
the LSTM-based neural network model is configured to predict the motion state of the vehicle B and the motion state of the vehicle C, so as to obtain a predicted motion state of the vehicle B and a predicted motion state of the vehicle C, wherein the predicted motion state of the vehicle B and the predicted motion state of the vehicle C each comprise longitudinal position, longitudinal acceleration and longitudinal speed;
the steering kinematic model is configured to predict the motion state of the lane-changing vehicle M, wherein a predicted motion state of the lane-changing vehicle M comprises longitudinal position increment {dot over (y)} M , transverse position increment {dot over (x)} M , yaw angle increment {dot over (φ)} M and axial speed increment {dot over (v)} M ; and
the longitudinal kinematic model is configured to predict the motion state of the vehicle A and the motion state of the vehicle D, wherein the motion state of the vehicle A and the motion state of the vehicle D each comprise longitudinal position increment {dot over (x)} i and longitudinal speed increment {dot over (v)} i , wherein x i is a longitudinal position of the vehicle A and the vehicle D, and v i is a longitudinal speed of the vehicle A and the vehicle D;
(c) based on vehicle information and predicted motion states of the lane-changing vehicle M, the vehicle A, the vehicle B and the vehicle C, constructing an upper-layer optimization model; and
calculating an optimal control value and an optimized motion state of the lane-changing vehicle M and an optimal control value of the vehicle A according to the upper-layer optimization model; and
(d) based on vehicle information of the lane-changing vehicle M, the vehicle C and the vehicle D, the optimized motion state of the lane-changing vehicle M and predicted motion states of the vehicle C and the vehicle D, constructing a lower-layer optimization model;
calculating an optimal control value of the vehicle D according to the lower-layer optimization model; and
controlling the lane-changing vehicle M, the vehicle A and vehicle D to run according to a corresponding optimal control value.

2. The smooth cooperative lane change control method of claim 1 , wherein the motion state of the vehicle B and the motion state of the vehicle C are predicted by using the LSTM-based neural network model through steps of:
acquiring a motion state of the vehicle B in previous m steps and a motion state of the vehicle C in previous m steps, wherein a sampling step interval is dt; and inputting the motion state of the vehicle B in previous m steps and the motion state of the vehicle C in previous m steps into the LSTM-based neural network model which has been trained according to HDV motion state, to obtain a motion state of the vehicle B in next N steps and a motion state of the vehicle C in next N steps.

3. The smooth cooperative lane change control method of claim 1 , wherein the steering kinematic model is expressed as:

{

x
M

.

=

v
M

⁢
cos
⁢

φ
M

y
M

.

=

v
M

⁢
sin
⁢

φ
M

φ
M

.

=

v
M

⁢

tan
⁢

δ
M

L

Assignees

• Changan Univ

Inventors

• Gong Siyuan
• Sun Kang
• Zhao Xiangmo
• Wu Xia
• Zhang Licheng
• Wang Wenjing
• Ding Yukun
• Zhang Congli
• Sun Yihan
• Chang Hongda
• Li Ze
• Zeng Lu
• Wang Xinxin
• Yin Jiakai
• Ma Yuqian
• Zheng Guohui

Classifications

• B60W2554/4041

  Position · CPC title

• B60W2520/105

  Longitudinal acceleration · CPC title

• B60W2520/06

  Direction of travel · CPC title

• B60W2520/10

  Longitudinal speed · CPC title

• B60W2520/14

  Yaw · CPC title