Multiple robots path planning based on reinforcement learning for object transportation | |
| Author | Sarawat Ruamngern |
| Call Number | AIT Thesis no.ISE-22-05 |
| Subject(s) | Robots--Control systems Reinforcement learning (Machine learning) Mobile robots--Control |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering in Mechatronics |
| Publisher | Asian Institute of Technology |
| Abstract | This thesis proposed reinforcement learning method to perform the object transportation task for multiple robots. This task consists of two main processes, path planning and motion control task. Double deep Q-learning (DDQN) model is selected to achieve path planning for any random environment. To increase the capability of reinforcement learning model, semi-supervised method by A* algorithm is applied during the training process in order to find the optimal path. For motion control task, reinforcement learning model must control a motion of differential wheeled mobile robot by providing actions composed of speeds of left wheel and right wheel. The models are separately trained for two different purposes. The first agent is trained to deal with path following task and another agent is trained to handle with point following task. The agent of point following task is used to control the group of robots to move with a fixed group shape. The agent which is trained in the environment without disturbance is used in simulation. And the agent which is trained in the environment included disturbance is applied in practice with three differential wheeled mobile robots. Proximal policy optimization (PPO) is selected to achieve path following task in simulation and point following task in practice. Deep deterministic policy gradient (DDPG) is selected to complete path following task in practice. And soft actor-critic is selected to complete point following task in simulation. Finally, the integration of proposed reinforcement learning models can accomplish the object transportation task for multi-robot system appropriately both in simulation and practice. |
| Year | 2022 |
| Type | Thesis |
| School | School of Engineering and Technology (SET) |
| Department | Department of Industrial Systems Engineering (DISE) |
| Academic Program/FoS | Industrial Systems Engineering (ISE) |
| Chairperson(s) | Manukid Parnichkun |
| Examination Committee(s) | Dailey, Matthew N.;Mongkol Ekpanyapong |
| Scholarship Donor(s) | His Majesty the King's Scholarships (Thailand) |
| Degree | Thesis (M. Eng.) - Asian Institute of Technology, 2022 |