Distributed control of multiple vehicle systems using constraint forces

Scope and Method of Study. The problem of coordination of multiple autonomous vehicles with distributed information exchange architecture is investigated. Coordination of a group of vehicles to achieve specific tasks, such as achieving and maintaining a desired formation without collisions and navigation of the group along a desired trajectory, by controlling individual vehicles based on a communication graph is considered. The goal is to determine a distributed control algorithm for the group of vehicles based on a fixed or dynamic communication graph. The new approach relies on the notion of constraint forces which are used in the development of the dynamics of a system of constraint particles with inertia. The governing equations of the constrained system of particles not only contain the external forces but also the constraint forces which limit the motion to be consistent with the constraints. The distributed control algorithm for the group of vehicles is designed by setting up appropriate distance constraints between vehicles and using the constraint force approach.;Findings and Conclusions. A stable, distributed control algorithm is developed for coordination of multiple vehicles. The algorithm is applicable to all vehicle dynamic models that can be feedback linearized to double integrator dynamics in the position variables. The advantage of the proposed method is that the constraint forces cancel only those applied navigation forces which act against the constraints. Another advantageous feature of the proposed distributed algorithm is that it allows addition/removal of vehicles into/from the formation gracefully with simple modifications of the control input based on the new communication graph, that is, the algorithm is scalable. The approach ensures that a prescribed safe distance is always maintained between vehicles, thus, avoiding collisions. The distributed control algorithm is shown to be applicable to formation control of a class of dynamic, nonholonomic mobile robots and aircraft. For vehicle dynamic models that contain parametric uncertainties an adaptation mechanism can be included in the distributed control algorithm to estimate the uncertain parameters. The proposed approach can also be applied to a group of vehicles with limited sensing capability, that is, each vehicle can communicate with other vehicles in the group in a limited region around it. In this case, the communication graph is dynamic. Given the sensing region and the target point for each vehicle, a distributed control algorithm is developed that is capable of driving the vehicles to their target points without any collisions from any initial configuration. Simulation results on a number of examples are shown to verify the proposed algorithms.