Reinforcement Learning Based, Staircase Negotiation Learning: Simulation and Transfer to Reality for Articulated Tracked Robots

Autonomous control of reconfigurable robots is crucial for their deployment in diverse environments. However, its development hampered by the diversity hardware and task constraints. We advocate use artificial intelligence-based approaches to improve scalability different conditions portability platforms comparable traversability skills. In particular, we succeed tackling problem staircase traversal via a reinforcement learning (RL)-based framework applicable articulated, tracked powerful enough generalize varying learned simulation transferred reality zero-shot setting. Our extensive experiments demonstrate robustness tasks with increased risk difficulty induced platform diversification dimensionality.