Navigating Mobile Manipulator Robot for Restaurant Application Using Open-Source Software
Sattra Piyapunsutti, Eisen Lance De Guzman, Ronnapee Chaichaowarat
Abstract
In this paper, a mobile manipulator with a lifting mechanism for food trays and a liquid dispensing system for drinks is developed for restaurant applications. Multiple nodes responsible for controlling the hardware are run in parallel by using the Robot Operating System (ROS) installed on a laptop (master node) and the Raspberry Pi microprocessor (slave node). The graphic user interface is prepared for the users to input commands through a casual web browser. The HTTP requests are sent to the REST API server implemented in the backend where several routes are responsible for publishing commands to the running nodes. The RPLIDAR sensor is used for scanning the 2D layout of surroundings and the Hector SLAM algorithm is used for constructing the map when the robot is deployed in a new environment. The ODrive motor controller is used for the differential wheeled robot. The robot’s odometry relied on the wheel encoder data is used for the ROS navigation stack to generate a path for navigating the robot to a goal position. The performance was evaluated experimentally in straight and turning combined scenarios. The gripper and the 4-DOF serial manipulator using Dynamixel servos are designed for picking a cup from a customer’s table, putting it on the drink dispenser, and serving the drink back to its position. The displacement and torque of the manipulator’s joints were observed during the serve-to-refill task to study the effect of liquid load bearing on the endpoint trajectories for the future improvement of gravity compensation.