International Aerial Robotics Competition
During MISSION 7, teams demonstrated interaction between fully autonomous aerial robots and moving objects (specifically, autonomous ground robots), navigation in a sterile environment with no external navigation aids such as GPS or large stationary points of reference such as walls, and interaction between moving obstacles.
The 7th mission of the AUVSI Foundation’s (now "RoboNation") International Aerial Robotics Competition (IARC) was initially held at both of its venues on the campuses of the Georgia Institute of Technology (American Venue) and Beihang University (Asia/Pacific Venue) in August of 2014. That began the 21st year of continuous operation for the International Aerial Robotics Competition.
Without doubt, MISSION 7 has been the most challenging IARC mission to date. Over the past six missions, supportive technologies have been successively removed (first no GPS, then no SLAM) to the point that the fully autonomous aerial robots had to know where they were using solely "eye sight" just like a biological entity. No aerial robot existed which could perform the behaviors required by this mission statement, however as with all past missions, the Organizer, Judges, and Staff fully expected a university team to meet this challenge within the space of several years, thereby again advancing the state-of-the-art in aerial robotics.
Another distinctive of MISSION 7 was the requirement for the autonomous aerial robot to interact physically with numerous autonomous ground robots by touching them on top (commanding a 45° CW turn) or blocking their forward motion (commanding a 180° turn) intelligently so that the ground robots would change direction and ultimately be "herded" out one side of a rectangular arena. One problem for the aerial robots was the fact that the ground robots were programed to reverse direction periodically and also experience small random changes in direction (15° or less) so that they never went in a straight line for more than twenty seconds. To further increase the workload of the aerial robot, a series of four obstacles which had to be avoided, circled within the arena and caused collisions with the autonomous ground robots, further disrupting any predictability in their trajectories.
A prize award was set at $30,000 for the best performance which equaled or exceeded inducing four ground robots to leave the designated side of the arena. The 7th mission continued to adhere to the Competition’s long standing practice of posing tasks that cannot be completed with current technology and skills. As with previous missions, nothing within the world military or industrial arsenal of robots was able to complete the proposed mission at the time the guidelines were released.
After four years of development, a number of the over 50 teams entered in MISSION 7 were able to perform the autonomous behaviors required to win the competition, however due to its difficulty, the 10-minute time limit, and only four chances to demonstrate the requirement, the maximum number of ground robots that were able to be intentionally herded out of the arena was only 3. Analysis clearly showed that given extra attempts, the goal of 4 was certainly possible and that the desired technologies were amply demonstrated, so in 2018, the top prize was awarded to the top three performers: Zhejiang University ($20,000); Xiamen University ($6,000); and the Harbin Institute of Technology ($4,000).
In the top performances, which were replicated multiple times, the Zhejiang University team showed that its autonomous aerial robot could track individual ground robots, redirect them in either 45° increments by touching them on the top or causing a direction reversal of 180° by landing in front of their path to block them while at the same time staying within the arena boundaries and avoiding the mobile obstacles circulating within the arena. The Zhejiang aerial robot would survey the arena and then choose likely candidate ground robots to track. Once locked onto a ground robot, it would reorient it, follow it, and when the ground robot was just about to change direction, the aerial robot would block the last few seconds of progress toward the goal line causing a reversal in direction away from the line. This seems counterintuitive, however the aerial robot was using the known timing of the ground robot reversals as specified in the Official Rules, to reverse its direction immediately before it would have reversed direction on its own. This resulted in the ground robot reversing 180° under the influence of the collision with the Zhejiang aerial robot, and then immediately reversing 180° again of its own accord -- the result being that the ground robot stopped, did a 360° rotation, and then continued along its path to the goal line. The Zhejiang aerial robot would then follow the ground robot until the next reversal was about to occur, whereupon the process was repeated until the ground robot crossed the line. This intelligent strategy was successful and entirely what the IARC organizers had hoped to see. The stochastic geometry of the ground robot field factored into the time required to complete the task as did the relatively slow speed of the ground robots. As such, the maximum number of ground robots that could be intentionally herded in the 10-minute run interval was only 3 (not including those that found their own way to an boundary edge and were removed from play).