SILO: Simulation-in-the-Loop Sim-to-Real Transfer for Multi-Stage Cable Routing

The paper introduces SILO, a sim-to-real reinforcement-learning framework for multi-stage cable routing. It combines GPU-parallelized simulation, localized RL policies, and robust cable-state estimation in a deployment strategy designed to reduce the simulation-to-reality gap. Training uses thousands of parallel simulations to expose policies to varied cable geometries and deformation patterns. On real-world routing tasks, the authors report higher success rates and a twofold reduction in cycle time compared with prior state-of-the-art learning methods. The work claims the first successful sim-to-real transfer of RL policies for multi-stage cable routing.