MIT Researchers Create Virtual Proving Ground for Drones
【Summary】The innovative system, called Fight Goggles, utilizes motion-capturing components and image-rendering software to facilitate virtual testing.
Flying drones in open spaces, such as fields and rooftops, is a relatively simple and straightforward task, with minimal obstructions around the location that can damage the unit.
On the other hand, navigating around a manufacturing plant is exponentially harder, as one wrong move can send the quadcopter flying through a window or crashing into heavy-duty machines.
In order to prepare drone operators and autonomous UAVs for such environments, researchers from Massachusetts Institute of Technology (MIT) are taking training programs out of the real world and into virtual platforms, opening up a plethora of new, cost-effective testing methods.
The innovative system, called Fight Goggles, utilizes motion-capturing components and image-rendering software to facilitate virtual testing. In application, the trials are conducted inside a large, empty building. Motion-sensing cameras are installed throughout the area, which are used to track the quadcopters, as the units fly around the location.
Next, the image-rendering software creates various obstacles and structures for the drone operator to navigate around. The testing method is applicable to honing the skills of autonomous drones as well. The image-rendering system can feed virtual components to a quadcopter in real-time, at a rate of 90 frames per second.
"The moment you want to do high-throughput computing and go fast, even the slightest changes you make to its environment will cause the drone to crash," said Karaman in a statement.
"You can't learn in that environment. If you want to push boundaries on how fast you can go and compute, you need some sort of virtual reality environment."
Autonomous Drone Applications
MIT's virtual reality testbed for drones is highly efficient. The team was able to teach an autonomous UAV to fly through a tiny (virtual) window during a demonstration. The drone completed 10 virtual flights and was able to pass through the virtual window 361 times.
As a final test, the researchers setup a real window, mimicking the virtual opening the autonomous drone was previously trained to pass through. With help from an onboard camera and cutting-edge algorithms that were honed during the virtual trial, the UAV successfully flew through the window 119 times (over eight flights). Human intervention was needed six times to assist the unit during the final test.
"We think this is a game-changer in the development of drone technology, for drones that go fast," highlighted Sertac Karaman, associate professor of aeronautics and astronautics at MIT. "If anything, the system can make autonomous vehicles more responsive, faster, and more efficient."
This technology has numerous applications and will likely play an important role in drone adoption. At the moment, companies are interested in quadcopters capable of flying at blisteringly fast speeds with precise navigational capabilities. In virtual proving grounds, businesses could easily create a replica of specific work environments, allowing autonomous drones to master the area in preparation for real flights and tasks.
Additionally, for public locations, autonomous drones must ‘learn' how to avoid humans, cars and buildings. A virtual environment can facilitate such trials safely without harming people or damaging expensive drones.
Michael Cheng is a legal editor and technical writer with publications for Blackberry ISHN Magazine Houzz and Payment Week. He specializes in technology business and digesting hard data. Outside of work Michael likes to train for marathons spend time with his daughter and explore new places.
Waymo Receives Permit to Participate in California’s Autonomous Vehicle Pilot
How Do Autonomous Cars Deal with Double-parked Vehicles?
Kitty Hawk and Boeing Form Partnership to Make Flying Cars Safer
Waymo to Bring Driverless Cars to France and Japan via Nissan-Renault Partnership
Porsche Forecasts EVs Going Mainstream
Zomato Tests Drones for Food Deliveries in India
Alibaba Brings Tmall Genie to Audi, Honda and Renault Vehicles
Driverless Sensor Startup Sense Photonics Raises $26 Million in Series A Funding
- Michigan Looking to Have Autonomous-Car Only Lanes
- Amazon Signs Agreement to Buy Self-Driving Startup Zoox for More Than $1 Billion
- Uber Acquires Software Company Routematch in its Push to Support City Public Transit Services
- Here’s What We Know About BMW's Battery-powered 3 Series Sedan
- BMW Introduces the 330e Touring PHEV with All-Wheel-Drive Option
- Electric Vehicle Startup Bollinger Motors Unveils its Battery-powered DELIVER-E Van
- Mercedes Benz Launches the Next-Generation ‘Mercedes me’ Apps to Keep Drivers Connected to Their Vehicles
- Chinese Electric Vehicle Startup NIO is Offering a Battery Subscription Plan for its Vehicles
- Ford Motor Co. is Working on Automated Valet Parking for its Future Vehicles
- Ford’s New F-150 Pickup Goes on Sale in November, a Fully-Electric Version Will Enter Production in 2022