Drones need extensive training to fly around. However, training drones in a surrounding full of obstacles is an accident-prone exercise, which can translate into the need for engineers repairing or replacing vehicles on frustratingly high pace. Now, engineers at MIT have developed a novel virtual-reality training system for drones, which enables a vehicle to visualize a rich, virtual environment during its flight in an empty physical space.
This system, which the team has named as ‘Flight Goggles,’ could decrease the number of crashes that drones have to witness significantly in actual training sessions. It can also act as a virtual testbed for numerous situations as well as any type of environment in which specialists may try to train drones to fly-fast. Sertac Karaman, an associate professor of aeronautics and astronautics at MIT, stated that they think that this is a distinct advantage in the advancement of drone technology for those drones, which fly really fast. “This is the only framework that can make autonomous vehicles faster more responsive, and more competent,” he added further.
The points of interest of their virtual training will be presented by Karaman and his associates at the IEEE International Conference on Robotics and Automation in the next week. Co-authors to this are Winter Guerra, Thomas Sayre-McCord, Amado Antonini, Austin Brown, Jasper Arneberg, Guilherme Cavalheiro, Sebastian Quilter, Dave McCoy, Fabian Riether, Yunus Terzioglu, Ezra Tal, and Luca Carlone of the Laboratory for Information and Decision Systems at MIT, together with, Yajun Fang of the Laboratory of Computer Science and Artificial Intelligence at MIT, and Alex Gorodetsky from Sandia National Laboratories.