Distributed Aggregative Optimization on Nonholonomic Multirobot Systems With Predefined-Time Estimation

Abstract

This letter investigates the distributed aggregative optimization problem in nonholonomic multirobot systems. The local objective function of each robot depends not only on its own state but also on aggregative information that incorporates the states of all robots in the system. In nonholonomic multirobot systems, the motion of robot is subject to nonholonomic constraints, where the movement direction is determined by robot’s current orientation. A fully distributed control protocol is proposed to solve distributed aggregative optimization on nonholonomic multirobot systems by leveraging predefined-time estimation and projection techniques. The protocol integrates a predefined-time estimator and a posture controller. The estimator enables each robot to obtain global information within a predefined-time, facilitating accurate gradient computation and system simplification. The posture controller generates control inputs by projecting the gradient onto the robot’s heading. Convergence properties are analyzed using Lyapunov method, and numerical simulations verify the effectiveness of the proposed approach.