RA*pex algorithm makes rulebook-based robot planning 100x faster

Robotic planning often requires juggling multiple objectives—safety, efficiency, regulatory compliance—that have complex priority relationships. Rulebooks formalize these relationships, providing a partial ordering that generalizes Pareto and lexicographic dominance. However, computing the full set of rulebook-optimal solutions is computationally expensive, limiting real-time use. Researchers Omar Muhammetkulyyev, Oren Salzman, and Tichakorn Wongpiromsarn tackle this by introducing epsilon-rule-dominance, a principled notion of approximate dominance under rulebooks. They then built RA*pex, a best-first search algorithm that efficiently returns a compact set of epsilon-approximate rulebook-optimal solutions.

RA*pex uses dimensionality reduction, a common speedup technique for multi-objective search, while respecting rule hierarchies by keeping separate closed sets and performing dominance checks over truncated and residual rule sets. The team proves that every rulebook-optimal solution is epsilon-rule-dominated by at least one solution in the returned set, ensuring no truly optimal plan is missed. Empirically, RA*pex achieves computation times over two orders of magnitude faster than existing methods—roughly a 100x speedup—making it practical for autonomous vehicles, drones, and other robots that must balance safety and mission goals in real time. The paper is available on arXiv (2608.04398).