Distributed Planning for Rigid Robot Formations with Probabilistic Collision Avoidance

In [11], an APF is applied both is often required for tasks such as inspection and reconnaissance to the VRB and the robots, where the potential acting on [1]. The goal of formation planning algorithms is to the VRB is calculated at a central node and the potentials compute movements of the robots such that a formation is acting on the robots are calculated locally on each robot. In maintained while they perform their respective tasks and avoid [12], convex optimisation is used to find the optimal similarity collisions with obstacles and each other. Formation planning transformation of a configuration that minimises the distance algorithms can be categorised as centralised or distributed. In travelled by the robots, with constraints on the velocities of the centralised methods, all information is gathered at a central robots and environmental constraints on the formation. In [13], location where the plans are computed, while in distributed [14], a method for navigating a formation toward a desired methods all robots participate in the computation of the plans state, by iteratively computing the optimal VRB within the and coordinate via communication, requiring inter-robot coordination largest convex polytope containing the current formation, is mechanisms [1]. Centralised methods can be simpler presented. In [15], the authors demonstrate that finding the to implement since they do not require inter-robot coordination optimal rotation, translation, and assignment of the robots mechanisms, but have the drawback of system dependency in a VRB can be solved separately. In [16], consensus is on a single computer representing a single point of failure.