Experimental Demonstrations of Real-Time MILP Control

This paper presents results from recent hardware experiments using two forms of on-line Mixed-integer Linear Programming (MILP). The demonstrations were performed on wheeled ground vehicles, with a view to future use on autonomous teams of UAVs. The first experiment uses MILP at a high-level to account for a limited detection horizon while maneuvering in the presence of obstacles. The second performs low-level control with MILP to reject disturbances while performing a rendezvous. The paper describes the design of the multi-vehicle testbed and discusses the architecture and implementation details of the experiments. In both cases, the online MILP compensated for the uncertainties present and successfully completed the maneuvers.