Watch: MIT Drone Autonomously Avoids Obstacles at 30 MPH
The software runs 20 times faster than current solutions and maps the drone's environment in real time.
“Our current approach results in occasional incorrect estimates known as ‘drift,’” he says. “As hardware advances allow for more complex computation, we will be able to search at multiple depths and therefore check and correct our estimates. This lets us make our algorithms more aggressive, even in environments with larger numbers of obstacles.”
So what’s the significance of all this? Well, if this technology was built into drones, they wouldn’t be able to crash into things, such as powerlines in West Hollywood that caused 650 people to be without power for three hours. But if Amazon, Google, Walmart and others want to drones to deliver products to customers, the drones can’t be crashing into things. It could also help hobbyists from destroying their drones.
MIT also recently tested a new system that combines simple control programs to enable fleets of drones to collaborate in unprecedented ways. MIT figured out how to sync the actions of multiple robots working in the same space, testing the concept on a small group of delivery drones. The drones need to cross paths to make the deliveries, obviously without crashing. Before they fly, the drones map out the optimal flight path and the Dec-POMDPs take over to aid the collaboration.