Prior-Guided Residual Reinforcement Learning for Active Suspension Control

Yang, Jiansen; Wang, Shengkun; Bai, Fan; Wei, Min; Sun, Xuan; Wang, Yan

doi:10.3390/machines13110983

This is an early access version, the complete PDF, HTML, and XML versions will be available soon.

Open AccessArticle

Prior-Guided Residual Reinforcement Learning for Active Suspension Control

by

Jiansen Yang

¹,

Shengkun Wang

¹,

Fan Bai

²,

Min Wei

²,

Xuan Sun

^3,*

and

Yan Wang

^4,*

¹

CATARC (Tianjin) Automotive Engineering Research Institute Co., Ltd., Tianjin 300300, China

²

Technical Development Center, Shanghai Automotive Industry Corporation-General Motors-Wuling Automobile Co., Ltd., Liuzhou 545007, China

³

State Key Laboratory of Advanced Rail Autonomous Operation, Beijing Jiaotong University, Beijing 100044, China

⁴

The Department of Aeronautical and Aviation Engineering, The Hong Kong Polytechnic University, Hong Kong

^*

Authors to whom correspondence should be addressed.

Machines 2025, 13(11), 983; https://doi.org/10.3390/machines13110983 (registering DOI)

Submission received: 16 September 2025 / Revised: 15 October 2025 / Accepted: 22 October 2025 / Published: 24 October 2025

(This article belongs to the Special Issue Active and Passive Safety and Noise, Vibration, and Harshness (NVH) of Intelligent Vehicles)

Download Versions Notes

Abstract

Active suspension systems have gained significant attention for their capability to improve vehicle dynamics and energy efficiency. However, achieving consistent control performance under diverse and uncertain road conditions remains challenging. This paper proposes a prior-guided residual reinforcement learning framework for active suspension control. The approach integrates a Linear Quadratic Regulator (LQR) as a prior controller to ensure baseline stability, while an enhanced Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm learns the residual control policy to improve adaptability and robustness. Moreover, residual connections and Long Short-Term Memory (LSTM) layers are incorporated into the TD3 structure to enhance dynamic modeling and training stability. The simulation results demonstrate that the proposed method achieves better control performance than passive suspension, a standalone LQR, and conventional TD3 algorithms.

Keywords: active suspension; suspension control; residual reinforcement learning; prior knowledge guided

Share and Cite

MDPI and ACS Style

Yang, J.; Wang, S.; Bai, F.; Wei, M.; Sun, X.; Wang, Y. Prior-Guided Residual Reinforcement Learning for Active Suspension Control. Machines 2025, 13, 983. https://doi.org/10.3390/machines13110983

AMA Style

Yang J, Wang S, Bai F, Wei M, Sun X, Wang Y. Prior-Guided Residual Reinforcement Learning for Active Suspension Control. Machines. 2025; 13(11):983. https://doi.org/10.3390/machines13110983

Chicago/Turabian Style

Yang, Jiansen, Shengkun Wang, Fan Bai, Min Wei, Xuan Sun, and Yan Wang. 2025. "Prior-Guided Residual Reinforcement Learning for Active Suspension Control" Machines 13, no. 11: 983. https://doi.org/10.3390/machines13110983

APA Style

Yang, J., Wang, S., Bai, F., Wei, M., Sun, X., & Wang, Y. (2025). Prior-Guided Residual Reinforcement Learning for Active Suspension Control. Machines, 13(11), 983. https://doi.org/10.3390/machines13110983

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Menu

Prior-Guided Residual Reinforcement Learning for Active Suspension Control

Abstract

Share and Cite

Article Metrics

Article Access Statistics

Further Information

Guidelines

MDPI Initiatives

Follow MDPI