Design and Development of Multi-DOF Robotic Arm with Wheeled Mobile Platform

Deployment of automated ground vehicle (AGV) has become one of the approach for automated delivery services. However, their implementation in indoor applications, such as mail delivery, has been limited by the high-variance tasks that require the robots to possess diverse abilities and make complex decisions. This study presents the design of an indoor mail delivery robot that integrates a mobile platform with a Selective Compliance Assembly Robot Arm (SCARA) to transport low-weight and small parcels autonomously within the faculty building, Block D. The mobile platform is equipped with an adaptive Proportional-Integral-Derivative (PID) controller and rotary encoders for odometry, an RPLiDAR A2M8 for Adaptive Monte Carlo Localization (AMCL), A* algorithm, Dynamic Window Approach (DWA) path planning, and Hector Simultaneous Localization and Mapping (SLAM) for autonomous navigation. The SCARA is fitted with a vacuum gripper, and its kinematics are analyzed for joint movement control. Parcel detection is achieved using image processing trained through transfer learning with TensorFlow and OpenCV. The evaluations in Building D demonstrated the system’s high flexibility and effectiveness in handling uncertainties in the environment. Results of the evaluations have shown a success rate of 70 % for completing the delivery. This indicates that the mobile manipulation system offers control strategies that can handle reactive tasks and make coherent decisions on a whim.