SuPLE: Robot Learning with Lyapunov Rewards