Q-Learning of Bee-Like Robots through Obstacle Avoidance

Abstract

Modern robots are frequently utilized for search and rescue operations, with a significant focus on navigating complex environments. A key approach for robots in these scenarios is reinforcement learning. Through reinforcement learning, robots learn to navigate towards their goal while avoiding obstacles. Q-learning, a sophisticated reinforcement learning technique, is employed to teach robots the optimal path. By interacting with their surroundings, robots gradually learn to achieve their objectives. This paper presents a simulation model of bee-like robots implemented in NETLOGO. Initially, the learning rate was low but increased over time. Using the Q-learning technique, the bees successfully learned to reach their goal while avoiding obstacles.