It is no doubt that a drone manufactured by the University of Zurich is not only a technical marvel but also engineering as well. And the speed of the drone is slower than a Sunday morning jogger. Last October, at the International Conference on Intelligent Robots and Systems held in Madrid, the drone which operates autonomously, and used AI – moved through various complicated series of movements. The drone was covered with sensors, and the top speed was 5.6 MPH.
Paul Nurkkala in Jeddah, Saudi Arabia, also demonstrated how a drone saved itself from crashing into the ground, even though all of the competitor drones did, and reached the finish line successfully. This was demonstrated at the Drone Racing League’s world championship, the drone that won the race, which was a league standard Racer 3, had speeds over 90 miles per hour. Still, this type of drone, like Xiro Xplorer Mini Drone, will need a human to control it. AI or Artificial Intelligence is being applied to every field nowadays, be it AlphaGo or AlphaStar produced by Google’s AI department called DeepMind. Still, AI in the real world does struggle to excel and make their mark.
The same was demonstrated in 2017, via the three autonomous drones developed by NASA’s Jet Propulsion Laboratory or the Fast Lightweight Autonomy program by Darpa. Each of these drones could not stand against a drone piloted by a human being. Still, onboard computers will get powerful over time, and more algorithms are being developed. It will take time for the computer to learn the difference between milliseconds and microseconds. A drone flying faster than 5.6 MPH will be an achievement. Artificial Intelligence Robotic Racing Circuit and Lockheed Martin hosting the AlphaPilot Competition will help in a way to drive more research and interest into autonomous or self-driving. The prize is $250,000 for beating a drone piloted by a human. The Racer 3 drone uses an AI chip made by Nvidia.
Autonomous drones to be successful, need to be able to work in environments where there is less reliance on external guidance systems like GPS. Founder of KEF Robotics, Kerry Snyder said that current AI drones can make very less onboard decisions. Also, currently AI is in simulation – so going to the real world from simulation will be difficult. This is where AI drones fall short of human skills. Professor of robotics and perception at the University of Zurich, Davide Scaramuzza said that AI drones’ main challenge is its perception based on its cameras.
AI drones will not be able to drive at 90 miles per hour and process images fast like a human. These drones cannot take onboard strategic decisions in real life, where an experienced human pilot can. AI drones also need to be done trade-offs between being more faster and being more safer as well. AI drones will end up just like Deep Blue or AlphaGo – and if it outraces a human controlled drone, it will be great technical progress and innovation. Programmers and engineers will be celebrities and companies will try endorsing them.