Distributed Source Seeking Using A Bi-Level Distributed Model Predictive Control Algorithm

This paper presents a novel distributed source seeking algorithm, called Bi-Level Distributed Model Predictive Control (BLMPC), to locate the source using a group of agents. During the process, the source emits a signal that the agents use to guide their movements, and the agents utilize the signal collaboratively with BLMPC to generate continuous estimation of the source location and move accordingly. BLMPC employs a bi-level structure involving an upper distributed optimization level to estimate the source location and a lower MPC level to control the agents with time-varying goals. This structure ensures that the formation center of agents converges to a small neighborhood around the signal source's location theoretically. The effectiveness of the proposed approach is illustrated by simulations.