Distributed Robust Receding Horizon Control for Multivehicle Guidance

Abstract

This paper presents a new distributed robust model predictive control algorithm for multivehicle trajectory optimization and demonstrates the approach with numerical simulations and multivehicle experiments. The technique builds on the robust-safe-but-knowledgeable (RSBK) algorithm, which is developed in this paper for the multivehicle case. RSBK uses constraint tightening to achieve robustness to external disturbances, an invariant set to ensure safety in the presence of changes to the environment, and a cost-to-go function to generate an intelligent trajectory around known obstacles. The key advantage of this RSBK algorithm is that it enables the use of much shorter planning horizons while still preserving the robust feasibility guarantees of previously proposed approaches. The second contribution of this paper is a distributed version of the RSBK algorithm, which is more suitable for real-time execution. In the distributed RSBK (DRSBK) algorithm, each vehicle only optimizes for its own decisions by solving a subproblem of reduced size, which results in shorter computation times. Furthermore, the algorithm retains the robust feasibility guarantees of the centralized approach while requiring that each agent only have local knowledge of the environment and neighbor vehicles' plans. This new approach also facilitates the use of a significantly more general implementation architecture for the distributed trajectory optimization, which further decreases the delay due to computation time.