Efficient sampling-based bottleneck pathfinding over cost maps
This work addresses pathfinding for multiple agents with motion constraints, but it appears incremental as it builds on existing sampling-based methods.
The paper tackles the bottleneck pathfinding problem over cost maps by introducing a sampling-based planner called bottleneck tree (BTT), which outperforms the state-of-the-art T-RRT* on challenging multi-agent instances, though specific performance numbers are not provided.
We introduce a simple yet effective sampling-based planner that is tailored for bottleneck pathfinding: Given an implicitly-defined cost map $\mathcal{M}:\mathbb{R}^d\rightarrow \mathbb{R}$, which assigns to every point in space a real value, we wish to find a path connecting two given points, that minimizes the maximal value with respect to~$\mathcal{M}$. We demonstrate the capabilities of our algorithm, which we call bottleneck tree (BTT), on several challenging instances of the problem involving multiple agents, where it outperforms the state-of-the-art cost-map planning technique T-RRT*. On the theoretical side, we study the asymptotic properties of our method and consider the special setting where the computed trajectories must be monotone in all coordinates. This constraint arises in cases where the problem involves the coordination of multiple agents that are restricted to forward motions along predefined paths.