Air Races Asimov-Style
From its inception as a literary form, Science Fiction writers have plumbed the subtleties of the human condition by pitting man against machine in conflicts that have raged from Earth to Arrakis to Tatooine to Caprica.
The contests of intellect, cunning, strength, and speed imagined by Asimov, Clarke, and P.K. Dick were recently actualized at the University of Zurich (Earth), where a trio of champion (human) drone-racers squared-off against a number of autonomous racing quadrotor drones developed by Professor David Scaramuzza’s Robotics and Perception Group.
Professor Scaramuzza is a gifted boffin whose research lies at the intersection of robotics, computer vision, and machine learning. His contributions to autonomous, vision-based, drone navigation has won Scaramuzza prestigious prizes and promoted his divinization among nerds.
Sophisticated sensors and fast computing enable powerful and agile robots—like a drone—to autonomously execute high-speed maneuvers beyond the abilities of all but the most accomplished human unmanned aerial system (UAS) operators—read Drone Pilots.
Mechanical advantages notwithstanding, some exceptionally capable humans are, in fact, able to pilot drones with an adroitness that rivals that of powerful A.I guidance systems. It comes as no surprise, therefore, that the sport of FPV (first-person view) drone racing has risen to prominence in the modern era.
Across the world, in competitions held for grant money, school pride, or old-fashioned, human one-upmanship, individuals and teams congregate to tests what’s possible when absurdly powerful drones are relegated to the deft hands of highly practiced humans who guide them through complex courses relying solely on a video feed sent from a camera on the drone’s snout to the pilot’s VR headset.
One year ago, autonomous racing quad-rotors from Scaramuzza’s Robotics and Perception Group—an assemblage of robotics researchers at the University of Zurich—prevailed in a contest against the world’s best drone pilots.
This year, a rematch will pit drone speedsters Thomas Bitmatta, Alex Vanover, and Marvin Schäpper against similar machines—albeit with a few key rule-changes enacted to level the playing-field, as it were.
In the first contest, the drones enjoyed the significant advantage of being guided by a motion-capture system that provided the contraptions very high resolution, real-time, position information. What’s more, the machines’ control information was transmitted from off-board computer arrays.
The 2022 race disallows such techno-tom-foolery. The competing drones will benefit from neither motion-capture systems nor off-board guidance computers. Truly, the rematch will afford the world an opportunity to witness drones and humans using their own vision systems and their own brains—organic and digital alike—to fly around a racing track as fast as possible.
How Isaac Asimov would have enjoyed such a spectacle!