Autonomous Aerial Tracking of a Herd of Ground Robots

The objective of this project is to design, develop, and test an autonomous surveillance system able to track objects without human intervention. The autonomous aerial tracking system developed by the team will be used in the International Aerial Robotics Competition in August 2016. Competition requirements stipulate that the unmanned aerial vehicle cannot use GPS, and that it must remain below 3 meters above the ground at all times. The competition arena is a 20-by-20-meter grid, which forces the detection device to have a large field of view. It must also be able to locate several small, mobile robots traversing the grid. The system uses four Pixy CMUcam5 cameras and an ELP 180 degree fisheye USB camera for detection. The system must provide the locations, velocities, and headings of the robots, in addition to its own location, velocity, and heading. All calculations are done onboard by a Raspberry Pi microcontroller and sent directly to the drone’s autopilot, which makes decisions on vehicle and robot movement based on information processed.