Abstract:
An efficient online planning or replanning methodology is a critical requirement for scalable and responsive real world multi-robot deployments. The need to replan typically stems from the invalidation of existing plans due to incomplete knowledge of the environment, or, from scenarios that necessitate changing goal locations in response to evolving application requirements.

In this work, we consider the problem of online kinodynamic planning for teams of aerial robots in response to changing goal locations in known and static three-dimensional workspaces. Offline reachability analysis on position-invariant geometric trees is leveraged for online composition of dynamically-feasible and collision-free motion plans from potentially non-stationary initial states. The key contribution of the approach is that a collision-free lower dimensional geometric solution implies a dynamically feasible and collision-free solution in the full state space, without additional refinement, yielding low online planning times.

In this talk, I will describe the procedure for offline reachability analysis on position-invariant geometric trees and discuss its application in a decoupled multi-robot planning algorithm. I will demonstrate the viability of the proposed approach as a framework for online planning for multi-robot teams via simulation and hardware experiments in cluttered three dimensional workspaces.

Committee:
Nathan Michael
Maxim Likhachev
George Kantor
Fahad Islam