Ride Comfort Enhancement using Deep Reinforcement Learning

Authors

Guru Bhargava Khandavalli, M.Sc., Univ.-Prof. Dr.-Ing. Lutz Eckstein, Institute for Automotive Engineering (ika), RWTH Aachen University, Aachen, Germany

Summary

Controllable suspension systems play an important role in improving driver and passenger ride experiences. In recent years, with significant advances in the fields of Computational Engineering and Artificial Intelligence (AI), control strategies more efficient than traditional approaches are emerging. These control strategies are potent not only because of their capability to deal with complex uncertain models but also because of their moderate requirements of on-board computational power for real-time implementation. With the help of Reinforcement Learning (RL) methods and Deep Neural Networks (DNNs), it is possible to develop controllers that can accomplish certain objectives when interacting with an uncertain environment. The current study focuses on a novel idea of applying the Deep Deterministic Policy Gradient (DDPG) RL algorithm to synthesize a controller that enhances ride comfort by minimizing the sprung mass acceleration. 
....

Members Login