Echodyne, a radar array startup with investors including Bill Gates, his Microsoft cofounder Paul Allen’s Vulcan Capital, and Seattle’s own Madrona Ventures, just made what the company considers a landmark announcement for the future of self-driving vehicles.
Basically, Echodyne’s big breakthrough product is the Metamaterial Electronically Scanning Array (MESA) – a radar arrays that’s small enough to mount on a drone – like the ones Amazon or Google might use for their drone delivery programs – but that the company says is orders of magnitude more powerful than the kind you’d find on existing self-driving cars and drones.
On Tuesday, Echodyne announces the successful test of a system that lets drones mounted with MESA automatically detect and evade oncoming aircraft and other obstacles… even when it would otherwise be too far away, too cloudy, or too dark for a human operator to see.
Drones mounted with Echodyne’s system flew several missions over a few days. You can watch one test here:
It’s a little dry, but stay with it. The important part is that the radar can pick up a decoy made to appear on radar like an oncoming small aircraft or other small drones, even when the camera can’t see them on visual, and steer around it.
“I can see that drone, and I can see another aircraft,” says Echodyne founder and CEO Eben Frankenberg.
This is a big deal, Frankenberg says, because it opens the door for drones to operate further away from human operators. Other startups are working on this problem, but Echodyne says theirs is the first to successfully complete real-world tests.
Right now, the FAA requires a drone pilot essentially be within eyeshot of the vehicle they’re flying, which is no good if you’re Amazon and you’re trying to send drones all over town. But with the ability to do automatic collision detection, a drone doesn’t need a human pilot to be paying attention in order to dodge hazards.
Frankenberg says Echodyne is leaving it up to its customers how to use this so-called “detect and avoid” system. There are two options, either leaving it to the vehicle’s artificial brain to make the call, or having it suggest a course correction to a human who can approve it. With issues of liabiltiy and safety still up in the air, Echodyne doesn’t want to wade into the debate.
“It’s still unclear what the FAA will be comfortable with, with algorithms,” Frankenberg says.
Right now, Echodyne is focused on drones, because that’s the biggest market with the most need for this kind of always-aware radar scanner.
But eventually, he says, it could come back down to earth: Self-driving cars from the likes of Uber and Google use a combination of radar and lasers to scan the road, but they lack a certain ability to “see” elevation – it can see a tree that’s fallen on the road, but if there’s a too-low overpass or a falling hazard, Echodyne’s MESA would be able to see it before the relatively weaker systems on the car and drive around.
“Lasers are not going to be a solution,” Frankenberg says.