Improving Online Railway Deadlock Detection using a Partial Order Reduction

Although railway dispatching on large national networks is gradually becoming more computerized, there are still major obstacles to retrofitting (semi-)autonomous control systems. In addition requiring extensive and detailed digitalization of infrastructure models information systems, exact optimization for computationally hard. Heuristic algorithms manual overrides likely be required semi-autonomous operations the foreseeable future. this context, being able detect problems such as deadlocks can a valuable part runtime verification system. If bound-for-deadlock situations correctly recognized early possible, human operators will have time better plan recovery operations. Deadlock detection may also useful in feedback loop with heuristic or algorithm if cannot itself guarantee deadlock-free plan. We describe SAT-based planning online situations. The exploits parallel updates train positions partial order reduction technique significantly reduce number state transitions (and correspondingly, sizes formulas) SAT instances needed prove whether deadlock situation bound happen Implementation source code benchmark supplied, direct comparison against another recent study demonstrates significant performance gains.