Scalable Mechanism Design for Multi-Agent Path Finding
This addresses incentive misalignment in MAPF for applications like autonomous vehicle coordination, but it is incremental as it builds on existing mechanism design and approximate algorithms.
The paper tackles the problem of Multi-Agent Path Finding (MAPF) with self-interested agents by introducing scalable mechanism design, proposing three strategyproof mechanisms that work with approximate algorithms, and testing them on domains with dozens to hundreds of agents to improve welfare beyond a baseline.
Multi-Agent Path Finding (MAPF) involves determining paths for multiple agents to travel simultaneously and collision-free through a shared area toward given goal locations. This problem is computationally complex, especially when dealing with large numbers of agents, as is common in realistic applications like autonomous vehicle coordination. Finding an optimal solution is often computationally infeasible, making the use of approximate, suboptimal algorithms essential. Adding to the complexity, agents might act in a self-interested and strategic way, possibly misrepresenting their goals to the MAPF algorithm if it benefits them. Although the field of mechanism design offers tools to align incentives, using these tools without careful consideration can fail when only having access to approximately optimal outcomes. In this work, we introduce the problem of scalable mechanism design for MAPF and propose three strategyproof mechanisms, two of which even use approximate MAPF algorithms. We test our mechanisms on realistic MAPF domains with problem sizes ranging from dozens to hundreds of agents. We find that they improve welfare beyond a simple baseline.