Purpose-based Delivery Robot Hybrid Control System Design Using Pixhawk and Edge Computing

Delivery robots have emerged as a technology that can revolutionize the delivery process across various industries. Utilization of autonomous robots within urban environments holds great promise for transforming purpose-based delivery systems. In this study, practical system design and control techniques of purpose-based delivery robots are investigated in a real-world outdoor environment, and a lab-fabricated Bravo delivery robot is reported. The robot follows predefined waypoints while seamlessly transitioning into obstacle avoidance mode when encountering obstacles in its path by using GPS signals and a Pixhawk-based navigation module. However, when GPS is unavailable, pre-embedded visual-inertial localization, mapping techniques, and deep-learning-based object detection are used by the robot to autonomously navigate and avoid obstacles through its onboard sensors and an NVIDIA Edge computer. A model predictive control technique is programmed inside the computer for the precise path tracking and obstacle avoidance of the delivery robot. Experimental results show the adaptability and responsiveness of the Bravo robot in dynamic settings, providing valuable insights into its real-world deployment for last-mile deliveries. The proposed approach offers an architectural design and algorithms to control a delivery robot. In addition, various design aspects, including mobility systems, sensing technologies, communication systems, path planning, obstacle avoidance, and control, are discussed. Thus, an autonomous delivery robot can be implemented in urban environments.