A Vision-based Unmanned Aircraft System for Autonomous Grasp & Transport

The progress in sensor technologies, computer capabilities and artificial intelligence has endowed the unmanned aircraft system (UAS) with more autonomous abilities. Motivated by the 6th International Unmanned Aerial Vehicle Innovation Grand Prix (UAVGP), a UAS with high degree of autonomy was developed to perform the mission of building a simulated tower using prefabricated components. According to the requirement of the competition, the UAS was designed and implemented from the following four parts: 1) navigation and control, 2) recognition and location, 3) grasp and construction, and 4) task planning and scheduling. Different levels of autonomy have been given to the UAS based on these parts. The system hardware was developed on a quadrotor platform by integrating various components, including sensors, computers, power and grasp mechanism. Software which included precise navigation, mission planning, real-time perception and control was implemented and integrated with the developed UAS hardware. The performance in the test environment and actual competition showed that the UAS could perform the mission without human intervention with high autonomy and reliability. This paper addresses the major components and development process of the UAS and describes its application to the practical mission.