Online Reinforcement Learning Control of Nonlinear Vehicle Platoons with Unknown Disturbances and Constraints

In our paper, an innovative framework for online reinforcement learning in a discrete-time setting has been introduced to address the complex issue of tracking within a third-order nonlinear discrete-time vehicle platoon. The Actor and Critic neural networks are concurrently refined in an online manner to formulate the most effective control strategy that remains within defined limits. The proposed method is suitable for nonlinear discrete-time tracking systems with unknown disturbances and control constraints, it does not require complete dynamic information. The results of the simulation confirm the scheme’s efficacy.