Modelling and analysis of active full vehicle suspension model optimized using the advanced reinforcement learning controller

Abstract

Enhancing driving safety and comfort relies heavily on the optimization of vehicle suspension systems. This research introduces an innovative method that leverages cutting-edge reinforcement learning techniques, such as Proximal Policy Optimization (PPO), Soft Actor-Critic (SAC), Twin Delayed Deep Deterministic Policy Gradient (TD3), and Deep Deterministic Policy Gradient (DDPG), to develop and assess an active suspension system. The study incorporates mathematical models of vehicle dynamics into a reward-driven control framework designed to reduce chassis acceleration and enhance stability under various road conditions. A robust simulation environment has been created to enable a detailed comparison of these algorithms using consistent vehicle and suspension parameters. The reward function is meticulously designed to optimize key performance indicators, such as minimizing angular accelerations and wheel bounce, thereby boosting passenger comfort and vehicle stability. Preliminary findings highlight the promise of reinforcement learning in significantly improving suspension performance, offering a pathway toward intelligent and adaptive automotive systems.