A Gradient-Free Source Seeking Strategy

This paper investigates the distributed source seeking problem using a group of mobile agents equipped with sensors for source scalar field measurement. Each controlled mobile agent is represented by a cascade connection of an integrator and a nonlinear uncertain actuation system. In specific, the integrator describes the kinematic of mobile agent, while the actuation system represents the dynamic response of the reference control input signal, which is characterized by input-to-output stability. The main objective is to drive average state of all agents to the optimal solution of a given scalar field function and maintain desired formation, based on field value measured by each agent. In order to achieve this objective, a formation-based source seeking algorithm is designed. Based on the actual measurements of the field, the algorithm generates a moving direction to guide the average state of agents collectively move toward the source without estimating gradient of the field. Furthermore, the desired formation is achieved and Lyapunov-based convergence and robustness results are presented. Finally, the effectiveness and superiority of the proposed method are demonstrated through a physical experiment.