A Leader-Follower Control System Based on Dynamic Error with Nonlinear Model Predictive Control

This paper addresses the leader-follower control problem of differential wheeled mobile robots based on observations and proposes a following control system using a nonlinear model predictive control (NMPC) algorithm. Firstly, the follower estimates the state of the leader using an extended Kalman filter (EKF) and simultaneously estimates its own state using a localization module. Under a predefined virtual spatial structure, the state error and dynamic error equations are computed, followed by the establishment of the NMPC model. The error model considers not only position error but also heading error to reduce overshoot caused by the position error term. Additionally, penalty terms for velocity, angular rate, acceleration, and angular acceleration are included. The effectiveness of the algorithm is validated through simulation in Gazebo using a ROS-based development package.