Path planning and tracking of a network intelligent robot

Abstract

Artificial intelligence (AI) has seen major improvements in the past decade, with much more applications of AI in various sectors, from board games to cybersecurity to engineering. In this report, we explore the use of AI in controlling a mobile robot to follow a given path, specifically using Reinforcement Learning (RL). A model of the mobile robot with inputs linear velocity and angular velocity, [v, ω] will be used for this project. Deep Deterministic Policy Gradient (DDPG) will be used to control these 2 inputs to follow a path. Firstly, a simple trajectory will be given, which is in a straight positive 𝑥 direction. This can be achieved using a reward-based algorithm where a reward is given for moving in the positive direction while a penalty is given for any deviation in the y-direction or orientation.