一种循迹控制参数调节器及其训练集构建方法

doi:10.19562/j.chinasae.qcgc.2025.02.005

Abstract

Abstract:

To improve the control accuracy of intelligent vehicle tracking controllers in variable operating conditions， controllers generally use multidimensional control parameter tables based on operating condition characteristics. When engineers manually adjust multidimensional control parameter tables， the workload is large and the tuning effect is not satisfactory. In order to enable the tracking controller of dynamic parameter adjustment capability， in this paper a vehicle speed and curvature adaptive parameter tuner is proposed based on radial basis function （RBF） neural network. Besides， a training set construction method based on Monte Carlo Probabilistic Inference for Learning Control （MC-PILCO） algorithm is proposed to address the problems of excessive real vehicle testing interactions and heavy tuning workload encountered during the training of tuner. By grouping typical operating conditions based on vehicle speed in the construction process of the training set， all different curvature working conditions within each vehicle speed working condition group are trained using the dynamic model trained on the data collected from tracking the straight-line scene at that vehicle speed for parameter tuning. By sharing the model， the number of real vehicle interactions is reduced. Real vehicle experiments show that the parameter adaptive tracking controller proposed in this paper has better lateral trajectory-tracking performance compared to controllers with fixed parameters under medium and low speed conditions.

Key words: trajectory tracking control, RBF neural network, multidimensional control parameters, training set construction, MC-PILCO

Kegang Zhao,Weilin Ou,Zheng Zhang,Zhihao Liang. A Tuner of Trajectory Control Parameters and the Construction Method of its Training Set[J].Automotive Engineering, 2025, 47(2): 248-258.

Figures/Tables 22

速度工况	曲率工况	LQR横向控制器参数					双PI控制器参数
速度工况	曲率工况	$Q 1,1$	$Q 2,2$	$Q 3,3$	$Q 4,4$	$R 1,1$	$K p s$	$K p v$	$K i v$
10 km·h^-1	0	$e 0.72$	$e - 0.61$	$e 3.31$	$e 0.95$	$e 2.30$	1.70	-4.50	-0.03
	1/25	$e 1.85$	$e - 1.86$	$e 3.77$	$e - 0.08$	$e 0.83$	1.70	-4.50	-0.03
	1/15	$e 2.07$	$e - 2.15$	$e 3.97$	$e - 0.21$	$e 0.61$	1.70	-4.50	-0.03
20 km·h^-1	0	$e 0.89$	$e - 0.32$	$e 2.72$	$e - 3.90$	$e 3.67$	0.37	-4.31	-7.53
	1/65	$e 2.08$	$e - 0.08$	$e 3.77$	$e - 5.16$	$e 2.09$	0.37	-4.31	-7.53
	1/55	$e 1.84$	$e - 0.35$	$e 3.26$	$e - 4.77$	$e 2.27$	0.37	-4.31	-7.53
30 km·h^-1	0	$e 1.22$	$e - 0.50$	$e 2.45$	$e - 4.39$	$e 4.49$	1.74	-4.37	-1.35
	1/65	$e 2.11$	$e - 0.16$	$e 3.10$	$e - 2.24$	$e 3.14$	1.74	-4.37	-1.35
	1/55	$e 2.11$	$e - 0.41$	$e 2.88$	$e - 1.99$	$e 3.19$	1.74	-4.37	-1.35

LQR横向控制器参数					双PI控制器参数
$Q 1,1$	$Q 2,2$	$Q 3,3$	$Q 4,4$	$R 1,1$	$K p s$	$K p v$	$K i v$
$e 1.24$	$e - 0.48$	$e 2.53$	$e - 4.88$	$e 4.49$	1.45	-5.40	-0.08

误差名称	单位	固定参数	车速自适应调节	车速与曲率自适应调节
$e d m a x$	m	0.15	0.11	0.05
$e θ m a x$	rad	0.01	0.01	0.01
$e s m a x$	m	0.10	0.22	0.19
$e v m a x$	m·s^-1	0.19	0.18	0.17
$m e a n (e d)$	m	0.06	0.03	0.01
$m e a n (e θ)$	rad	0.00	0.00	0.00
$m e a n (e s)$	m	0.03	0.05	0.05
$m e a n (e v)$	m·s^-1	0.05	0.05	0.05

References 18

1	高建平，余佳衡，孟垚，等. 电动车辆控制参数自动优化标定系统的研究及验证［J］. 中国机械工程， 2022， 33（1）： 118-125.
	GAO J P， YU J H， MENG Y， et al. Research and verification of control parameter automatic optimization calibration simulation system for electric vehicles［J］. China Mechanical Engineering， 2022， 33（1）： 118-125.
2	KOTHARI M L， MADNANI S， SEGAL R. Orthogonal least squares learning algorithm based radial basis function （RBF） network adaptive power system stabilizer［J］. IEEE， 1997： 542-547.
3	杨建伟，刘海波，孙守光. 基于RBF网络自适应控制的车辆道路友好性与平顺性研究［J］. 汽车工程， 2010， 32（12）： 1077-1082.
	YANG J W， LIU H B， SUN S G. A study on the road-friendliness and ride comfort of vehicle based on RBF network adaptive control［J］. Automotive Engineering， 2010， 32（12）： 1077-1082.
4	陈刚，张为公. 基于模糊自适应PID的汽车驾驶机器人的车速控制［J］. 汽车工程， 2012， 34（6）： 511-516.
	CHEN G， ZHANG W G. Speed control of vehicle robot driver based on adaptive fuzzy PID control［J］. Automotive Engineering， 2012， 34（6）： 511-516.
5	ZHANG M W， TIAN F J， HE Y Q， et al. Adaptive path tracking for unmanned ground vehicle［J］. IEEE International Conference on Unmanned Systems （ICUS）， 2017： 139-145.
6	MOOREN N， WITVOET G， OOMEN T. Feedforward motion control： from batch-to-batch learning to online parameter estimation［J］. American Automatic Control Council， 2019： 947-952.
7	AMADIO F， DALLA-LIBERA A， ANTONELLO R， et al. Model-based policy search using monte carlo gradient estimation with real systems application［J］. IEEE Transactions on Robotics， 2022.
8	梁军，朱方博，蔡英凤，等. 面向复杂曲率变化的智能车路径跟踪控制［J］. 汽车工程， 2021， 43（12）： 1771-1779.
	LIANG J， ZHU F B， CAI Y F， et al. Intelligent vehicle path tracking control based on complex curvature variation［J］. Automotive Engineering， 2021， 43（12）： 1771-1779.
9	赵健，杜金朋，朱冰，等. 基于自适应动态滑模控制的智能汽车纵向巡航控制［J］. 汽车工程， 2022， 44（1）： 8-16.
	ZHAO J， DU J P， ZHU B， et al. Longitudinal cruise control of intelligent vehicles based on adaptive dynamic silding mode control［J］. Automotive Engineering， 2022， 44（1）： 8-16.
10	DEISENROTH M， RASMUSSEN C E. PILCO： a model-based and data-efficient approach to policy search［C］. Proceedings of the 28th International Conference on Machine Learning， 2011： 465-472.
11	梁俊朗，高健. 基于PILCO架构的自适应内模控制器优化［J］. 组合机床与自动化加工技术， 2023（9）： 59-62.
	LIANG J L， GAO J. An adaptive internal model optimization based on PILCO architecture［J］. Modular Machine Tool &amp； Automatic Manufacturing Technique， 2023（9）： 59-62.
12	SRIVASTAVA N， HINTON G， KRIZHEVSKY A， et al. Dropout： a simple way to prevent neural networks from overfitting［J］. Journal of Machine Learning Research， 2014， 15（1）： 1929-1958.
13	王萍，彭香园，程泽. 基于DTV-IGPR模型的锂离子电池SOH估计方法［J］. 汽车工程， 2021， 43（11）： 1710-1719.
	WANG P， PENG X Y， CHENG Z. SOH estimation method for lithium-ion batteries based on DTV-IGPR model［J］. Automotive Engineering， 2021， 43（11）： 1710-1719.
14	王艳辉. 复杂非线性动态系统的滤波与平滑方法研究［D］. 成都：电子科技大学， 2019.
	WANG Y H. Research on filtering and smoothing methods for complex nonlinear dynamic systems［D］. Chengdou： University of Electronic Science and Technology of China， 2019.
15	RAJAMANI R. Vehicle dynamics and control［M］. Springer Science &amp； Business Media， 2011.
16	王耀健，顾洁，温洪林，等. 基于在线高斯过程回归的短期风电功率概率预测［J］. 电力系统自动化， 2024.
	WANG Y J， GU J， WEN H L， et al. Short-term wind power probability prediction based on online gaussian process regression［J］. Automation of Electric Power Systems， 2024.
17	XU S B， PENG H， TANG Y F. Preview path tracking control with delay compensation for autonomous vehicles［J］. IEEE transactions on Intelligent Transportation Systems， 2021， 22（5）： 2979-2989.
18	公路路线设计规范： JTG D20—2017［S］.2017
	Design specification for highway alignment： JTG D20—2017［S］.2017

车辆参数/车载设备	参数大小/设备型号
车长	4 337 mm
车宽	1 825 mm
车高	1 658 mm
车轴距	2 560 mm
车载激光雷达	RoboSense RS-LIDAR-128
车载定位设备	广州导远INS570D
车载工控机	Ubuntu 16.04
	Intel Core i9-10900F
	32 GB RAM
	Nvidia GeForce RTX 2060

[1]	Qin Shi,Xin Liu,Helie Ying,Mingwei Wang,Zejia He,Lin He. Study on the Backstepping Control Algorithm for the Hydraulic Pressure in Electro-hydraulic Brake-by-wire System [J]. Automotive Engineering, 2022, 44(5): 747-755.
[2]	Yunfei Zha,Lü Xiaolong,Xinye Liu,Fangwu Ma. Fusion Estimation of Vehicle State Parameters Based on Dichotomy [J]. Automotive Engineering, 2022, 44(12): 1910-1918.
[3]	Jian Zhao,Jinpeng Du,Bing Zhu,Zhiwei Wang,Zhicheng Chen,Xiaowen Tao. Longitudinal Cruise Control of Intelligent Vehicles Based on Adaptive Dynamic Sliding Mode Control [J]. Automotive Engineering, 2022, 44(1): 8-16.
[4]	Yue Ren,Jie Ji,Ying Zhao,Yixiao Liang,Ling Zheng. Path Tracking Control of Intelligent Vehicle Based on Minimal Model Error Estimation [J]. Automotive Engineering, 2021, 43(4): 580-587.
[5]	Shi Qiujun, Li Jing. Anti-Rollover Control of Bus Based on Nonlinear Disturbance Estimation [J]. Automotive Engineering, 2020, 42(9): 1224-1231.
[6]	Chen Long, Xie Yunpeng, Cai Yingfeng, Sun Xiaoqiang, Teng Chenglong, Zou Kai. Stable Tracking Control of Autonomous Vehicles at Extreme Conditions [J]. Automotive Engineering, 2020, 42(8): 1016-1026.
[7]	Cai Yingfeng, Qin Shunqi, Zang Yong, Sun Xiaoqiang & Chen Long. Lateral Trajectory Tracking Control Scheme for Intelligent Vehicle Based on Extension Goodness Evaluation [J]. Automotive Engineering, 2019, 41(10): 1189-1196.
[8]	Hu Yuhui, Yue Xinyu, Wu Hongzhen & Xi Junqiang. Gear-shifting Control with Speed / Torque Double-synchronization for Hybrid Electric Vehicles [J]. , 2018, 40(9): 1076-1082.

A Tuner of Trajectory Control Parameters and the Construction Method of its Training Set

RichHTML

PDF (PC)

Like

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 22

References 18

Related Articles 8

Metrics

Comments

Recommended 0