A team of engineers at Massachusetts Institute of Technology (MIT) have created a clever way of testing new drones: put them in an empty room and use virtual reality to make them ‘think’ they are traveling through rich environments.
The typical process of testing a drone involves flying them out in a natural setting such as around a field or carpark.
Problem is, when drones are first being trialled they often crash. Depending on the severity of the accident and the extent of the damage, a drone crash may delay research and project take longer and cost more money.
Credit: MIT news.
How does their system work?
The MIT team developed “Flight Goggles” – a method of stimulating the drone’s sensors in a controlled setting. The drone still flies, but only around an empty room. This method saves the drone from unexpected gusts of wind or low-hanging powerlines that pilots frequently encounter in an open-world setting.
Because it’s virtual, the team can simulate whatever type of flying conditions or obstacles they like; whether it be a calm day in open space or extremely windy in a built-up area.
Drones are capable of seeing images at approximately three times the rate a human can. Whereas we can see reality at around 30 frames per second (fps), drone sensors can register around 90 fps.
Associate professor of aeronautics and astronautics at MIT, Sertac Karaman told MIT news: “We think this is a game-changer in the development of drone technology, for drones that go fast.”
“If anything, [our] system can make autonomous vehicles more responsive, faster, and more efficient.”
Inspired by drone racing
Karaman’s interest in developing this tool was partially sparked by a hope to engineer an automated drone that can more quickly and effectively navigate its way around a racing course than a physical pilot. The team hopes to enter a drone race in the next few years and be victorious with an autonomous drone.
A group of FPV drone racing competitors.
Drone racing is becoming big business and stakeholders involved in this project include funding from the U.S. Office of Naval Research, MIT Lincoln Laboratory and NVIDIA Corporation.
The team behind the technology are set to present their research to the IEEE International Conference on Robotics and Automation next week.
We look forward to bringing you future developments with this project.
