Multi-agent pathfinding (MAPF) is critical for robotics and autonomous systems, but optimal solutions are often computationally expensive. Inflated M* addresses this by combining subdimensional expansion with heuristic inflation, allowing bounded suboptimal solutions with significantly reduced computation. The algorithm expands the search space only in relevant dimensions and inflates heuristics to prune less promising paths, ensuring solutions are within a user-defined factor of optimal. This makes it suitable for real-time applications where exact optimality is not required. The article provides a detailed explanation of the algorithm's mechanics, including its theoretical guarantees and practical implementation considerations. For researchers and engineers working on multi-robot coordination, Inflated M* offers a promising tool for scalable pathfinding.
A new algorithm for multi-agent pathfinding that balances optimality and computational cost.