Improving Controllers for Formations and Deconfliction among Non-holonomic Vehicles
AdvisorBekris, Kostas E
Computer Science and Engineering
StatisticsView Usage Statistics
The first contribution of this work corresponds to a computationally efficient geometric method for simulating formations of systems with non-holonomic motion constraints inspired by solutions in robotics. The geometric reasoning takes place in curvilinear coordinates, which are defined by the curvature of the leaders reference trajectory, in order to directly satisfy the constraints, instead of the typical rectilinear coordinates. The approach directly provides the feasible controls that each individual robot has to execute for the team to maintain the formation based on the controls of a reference agent, either a real leader or a virtual one. The methods generality lies on the ability to define dynamic formations so as to smoothly switch between different configurations , where the robots can change both of their relative coordinates as they move. It is also possible to acquire a desired formation given an initial random configurations.The second contribution corresponds to a motion coordination algorithm, where multiple non-holonomic vehicles are steered in a decentralized manner between assigned start and goal configurations without collisions. The approach builds on top of a hybrid control law, known as Generalized Roundabout Policy, which ensures safety. The focus of this work is on improving the performance and liveness features for such problems. Towards this objective, a new hybrid policy that updates the desired direction for each vehicle based on a dynamic priority scheme is proposed. Minimal communication between the various vehicles is employed for the dynamic priority scheme, where vehicles occasionally exchange information. Each agent has communication only with its local neighborhood in the proposed scheme. This work can solve the problems both faster and with less assumptions.