Mobile Service Robot Path Planning using Deep Reinforcement Learning

A mobile service robot operates in a constantly changing environment with other robots and humans. The is usually vast unknown, the expected to operate continuously for long period. can be dynamic, leading generation of new routes or permanent blocking old routes. traditional path planner that relies on static maps will not suffice dynamic environment. This work focused developing reinforcement learning-based proposed system uses deep Q-Learning algorithm learn initial paths using topological map In an environmental change, β-decay transfer learning trains agent experience vs. exploration exploitation-based training depending similarity environments. implemented Robotic Operating System framework tested Turtlebot3 Gazebo simulator. experimental results show learns all based different environments accuracy over 98%. comparative analysis non-transfer agents performed various evaluation metrics. converges twice faster than agent.