Adaptive Reinforcement Learning Tracking Control of Vehicle Based on Threshold Band Event-Triggered

In this paper, an adaptive neural network control algorithm based on event-triggered reinforcement learning is proposed for a four-wheel independent steering and four-wheel independent driving (4WS4WD) mobile robot. A kinematic model is established based on the kinematic relationship between the robot wheels and the body under the consideration of the effect of slip-turn perturbation. The dynamics model is established using the Lagrangian dynamics equations. An improved performance metric function is designed and approximated using the Critic neural network and the Actor neural network to approximate the unknown long-term performance metric function and controller respectively. A threshold band event triggering is proposed for reducing the consumption of communication and computational resources. It is rigorously demonstrated using Lyapunov analysis that both the neural network error and the system error are up to the final consistent bound. As well as proved that the proposed event-triggered mechanism can eliminate the Zeno phenomenon. Finally, comparative experiments demonstrated the effectiveness of the proposed algorithm. Note to Practitioners—4WS4WD mobile robots are able to perform inspection tasks in more scenarios due to their excellent characteristics. However, due to its complex structure, it is difficult to control it accurately when there are slipping, sliding and input disturbances. In this paper, the model of the robot is established by considering the effects of slip, slip and disturbance. And an adaptive neural network control algorithm based on threshold band event-triggered reinforcement learning is proposed. Simulation results show that the proposed algorithm enables the robot to effectively reduce computational resources while maintaining good tracking results.