Distributed Path Consensus Algorithm

Path planning methods based on discrete graph searches are highly effective in dealing with large-scale environments, arbitrarily shaped obstacles, frequent map updates and complex cost functions. When it comes to planning for teams of coupled agents however, typical graph searchbased planning techniques quickly become infeasible due to the exponential growth of the dimensionality of the joint state-space. In this paper, we study the problem of path planning for teams of agents coupled with timeparameterized constraints on the distances from each other, the problem that often arises in planning for multi-agent robotic systems with limited communication range. We first pose this problem as searching N (the number of agents) graphs for cost-minimal paths that satisfy the distance constraints. We then show how the solution to this problem path consensus can be achieved in distributed fashion. Our theoretical analysis gives the conditions under which the algorithm converges to an optimal solution. Our experimental analysis shows that the algorithm converges fast for teams of up to six agents navigating cluttered environments.