Enhancing Autonomous Door Traversal for Mobile Manipulators Using Behavior Trees

Door-traversal research has been intensively studied. The challenge of the door traversal problem is derived from the requirement of the mobile manipulator robot to regulate the position of its end-effector while moving along a path. This research presents the door traversal of a mobile manipulator robot by simulating the work environment in the Gazebo program, using the Behavior Trees (BT) architecture to decide on various tasks with complex behaviors to manipulate the open and closed-door conditions. This study developed a program using Robot Operating System (ROS2) as middleware to connect different parts of the program. The results of this research show that classical BT still have limitations in the case of robot positioning errors, resulting in the possibility of BT making incorrect decisions. To overcome these limitations, the robot must be able to learn and comply with the errors. However, experiments have shown many advantages of BT, including flexibility in creating complex behaviors, easy understanding of work steps, decision-making processes, and maintenance capabilities that make it easy to modify robot behavior. As a result, we successfully created robot behaviors using BT to address the door traversal problem, encompassing both door opening and closing scenarios. Furthermore, we identified the limitations of position errors affecting the output of the BT.