Comparison of Temporal Difference Learning Algorithm and Dijkstra's Algorithm for Robotic Path Planning

Abstract

Robotic Navigation is a crucial issue for any robotic based automation. In order to implement a seamless robotic navigation, path planning is a key challenge in robotic navigation. Several algorithms exist in literature on robotic path planning of static and dynamic obstacles. In this work, an approach called Modified Temporal Difference Learning for path planning and obstacle avoidance is proposed for static obstacles. The algorithm is developed in MATLAB software and path planning is implemented in a 4 × 4 grid environment. A GUI for the same is created, which access the user inputs like obstacle number, positions and type. The developed algorithm is compared with the conventional path planning Dijkstra's algorithm in the same environment. It is observed that computational complexity is less in the proposed approach compared to the conventional method.