A Hybrid Scheme for Distributed Control of Autonomous Swarms

In this paper, a hybrid scheme for distributed control of autonomous vehicles is presented by combining the deterministic gradient-flow method and the stochastic method based on the Gibbs sampler. The scheme has the advantages of both methods and can potentially provide fast, distributed maneuvers while avoiding getting trapped at local minima of the potential function. Preliminary analysis is performed for the optimal design of the parameters controlling the switching between the two methods. The performance of the hybrid scheme is further enhanced by the introduction of vehicle memory. Simulation results are provided to confirm the analysis and show the effectiveness of the proposed algorithm.