Adaptive Path-Planning for Autonomous Robots: A UCH-Enhanced Q-Learning Approach