Autonomous Coffee Delivery Robot Using PRM and A* Path Planning

In this paper, My present the design and implementation of an autonomous robot system, designated as the “Coffeebot,” for the purpose of delivering coffee to a specified room on the third floor of an engineering building. The project explores the application of autonomous delivery robots in environments such as hospitals and warehouses, emphasizing the potential for significant cost savings and efficiency improvements in internal logistics. Central to my approach is the robot's ability to effectively localize itself within a predefined map and navigate autonomously to designated locations. My employed the Husky unmanned ground vehicle by Clearpath Robotics, equipped with a Hokuyo UTM-30LX laser scanner and a Microsoft Kinect sensor, as the hardware platform. The Robot Operating System (ROS) framework was utilized for software development, incorporating existing packages like RTAB-Map for mapping and localization and the ROS navigation stack for path planning and execution. My primary contribution is the comparison between two custom global path planners: a Probabilistic Roadmap (PRM) and a Grid-Based A* algorithm. Through extensive testing and simulations, My evaluation of the performance of both planners in terms of path generation time and path length. Results indicate that while PRM offers faster path generation in single-query scenarios, Grid-Based A* consistently finds optimal paths if they exist. The paper concludes with insights into the limitations encountered, proposed improvements, and directions for future research, including the integration of aerial vehicles and advanced feature detection methods.