Publication

Abstract : Mobile robots are widely used in day-to-day life. The main challenging task in the case of the mobile robot is finding an obstacle-free path with maximum rewards. To gain a collision-free path planning of the mobile robot, this paper uses a classical approach of reinforcement learning based Q-learning to obtain an optimal path, that can transverse from any starting state to the terminal state. We have created a grid world of 121 states, where an agent must start by exploring all states to reach the terminal state, which is fixed, by avoiding obstacles in hole states. The algorithm’s parameters were carefully set, and its performance was verified, through a series of well-planned trials. The reinforcement learning formulation that helps the robot to reach a terminal state with maximum rewards and comparative graphs are plotted accordingly. Our strategy incorporates environmental exploration, where an agent exhibits maximum reward points, and comparative graphs are plotted accordingly.