Cooperative Coverage Path Planning Using Q-Learning and Sarsa in Two Environments
Alireza Nezamzadeh, Hamed Jalaly Bidgoly, Marzieh Kamali · 2024
This paper presents the coverage environment by multi agents using reinforcement learning with Q-learning and Sarsa algorithms. Two environments are explored to study cooperative coverage path planning with these algorithms. The goal is to achieve coverage in the environment by having agents collaborate to reduce energy consumption. Initially, an obstacle-free environment is examined, then an obstacle is considered in the environment. The proposed algorithm involves sharing experiences among agents to cover environments containing obstacles, and compares its convergence speed to scenarios without sharing experiences. The proposed algorithm's performance is assessed through different simulations.