Hello,

We noticed you're browsing in private or incognito mode.

To continue reading this article, please exit incognito mode or log in.

Not an Insider? Subscribe now for unlimited access to online articles.

Intelligent Machines

How Drones May Avoid Collisions by Sharing Knowledge

Instead of programming drones to know what to do in every single scenario, a Stanford lab is giving them the intelligence to understand their surroundings.

If the U.S. Federal Aviation Administration allows the widespread use of commercial drones, the skies could soon buzz with swarms of unmanned aerial vehicles–especially in dense urban cores. That means drones will be tasked with autonomously avoiding collisions, as their numbers will be too high to rely on human air-traffic controllers at all times.

Drones flying in densely built cities will need to be programmed to make quick decisions to avoid collisions.

The Stanford Intelligent Systems Laboratory is just one team of more than 130 working with NASA to solve how to manage drone traffic. The traffic-management system, which will be under development for the next few years, will help drones communicate with each other and avoid potential collisions.

“They’re going to be doing much more unusual missions that will require them to fly in flight paths that are curvy,” says Mykel Kochenderfer, director of the Stanford laboratory. “Being robust to that uncertainty is very, very important.”

A recent paper published by Kochenderfer and mechanical engineering graduate student Hao Yi Ong describes a quick decision process the traffic-management system can use to reroute drones and avoid a collision. Their team ran more than a million simulations for conflict situations for anywhere between two and 10 drones. Drones were given varying levels of information about the other drones in the system and then were tested on their response time and how often they ran into conflict.

The Stanford researchers found that drones could make the quickest decisions when they were paired with the closest other drone, and the two solely considered the other’s behavior. The slowest response occurred when drones considered their own surroundings and then fed their results into a central system that sent decisions back to the entire group. Decision time always increased as more drones entered the simulation, but the system was always able to make a decision on rerouting a drone within 50 milliseconds.

While drones feeding their data into a central decision-making system was the slowest, it was also the safest. Drones were the least likely to encounter conflict when they fed data into a central system. Drones that received location data about other drones and assumed they would stay on the same path were the most likely to encounter conflict.

The Stanford lab also works with autonomous cars and air-traffic control for conventional planes. One of its projects, which Kochenderfer developed in part with former colleagues at MIT, involved using a small amount of computing power to decide how a plane should avoid a collision. Traditionally, collision avoidance has been guided by nearly 2,000 pages of documents that detail every possible scenario and how to react. Stanford and MIT’s solution is currently being standardized for use on all large aircraft.

NASA plans to spend 2016 testing the drone-traffic-management systems it has developed thus far at the drone test sites set up across the U.S. by the FAA. Back in November, a NASA team flew a drone at Moffett Field in California while simulating conflicts with drones generated on a computer, triggering an early version of the traffic-management system to alert the drones about the potential collisions. The FAA also tested similar systems developed by drone software and services company Precision Hawk (see “FAA Will Test Drones’ Ability to Steer Themselves Out of Trouble”).

“To allow large-scale UAS [unmanned aircraft systems] with a mix of beyond visual line of sight and within visual line of sight, we need a system that consists of technologies to manage airspace and capabilities on the UAS itself, rules of the airspace, and procedures for managing contingencies and emergencies,” says Parimal Kopardekar, who leads NASA’s drone-traffic-control program.

Kochenderfer says the Stanford researchers have tested their work in simulations, but have yet to see it operate with real drones. Validating that it works in the air is one of the final steps.

“This is one of the most exciting areas of aerospace right now—the use of drones,” Kochenderfer says. “Many of the applications they enable can lead to new economic models, but the potential for saving lives and improving efficiency, I think that’s really quite interesting.”

Tech Obsessive?
Become an Insider to get the story behind the story — and before anyone else.

Subscribe today

Uh oh–you've read all of your free articles for this month.

Insider Premium
$179.95/yr US PRICE

More from Intelligent Machines

Artificial intelligence and robots are transforming how we work and live.

Want more award-winning journalism? Subscribe to Insider Plus.
  • Insider Plus {! insider.prices.plus !}*

    {! insider.display.menuOptionsLabel !}

    Everything included in Insider Basic, plus the digital magazine, extensive archive, ad-free web experience, and discounts to partner offerings and MIT Technology Review events.

    See details+

    What's Included

    Unlimited 24/7 access to MIT Technology Review’s website

    The Download: our daily newsletter of what's important in technology and innovation

    Bimonthly print magazine (6 issues per year)

    Bimonthly digital/PDF edition

    Access to the magazine PDF archive—thousands of articles going back to 1899 at your fingertips

    Special interest publications

    Discount to MIT Technology Review events

    Special discounts to select partner offerings

    Ad-free web experience

/
You've read all of your free articles this month. This is your last free article this month. You've read of free articles this month. or  for unlimited online access.