Autonomous navigation in drone racecourse

摘要

This paper reports the challenges and solutions to autonomously navigate in a drone racecourse with adjustable course gates and reconfigurable obstacles. The work fulfills the educational goals of Science, Technology, Engineering, and Math (STEM) through the Autonomous Unmanned Air Vehicle Racing program by MIT Beaver Works, a joint venture between Lincoln Laboratory and School of Engineering. Rising high-school seniors are selected nationwide. After completing an online course in the spring, offered by multidisciplinary experts from all over the world, the students develop team projects in the summer through a four-week residential program. The tasks include the coding of a quadrotor with sensors and controllers, aided by a ground station, as well as the development of a sensing, control, and planning system with efficient algorithms. The experiments show the quadrotor's capability of autonomous navigation in unknown racecourses while avoiding obstacles. The final race day leverages the lessons learned and challenges the students to test their new skills against the clock. Future work will incorporate research skills into this world-class STEM program.