Follow
Subscribe

NVIDIA Drone Flies Through the Woods Using AI Instead of GPS

Home > News > Content

【Summary】Researchers at Nvidia are working on a drone navigation system that relies on visual recognition and computer learning instead of GPS

Original Eric Walz    Jul 07, 2017 5:23 PM PT
NVIDIA Drone Flies Through the Woods Using AI Instead of GPS

GPS can be invaluable for driving and finding your way and a requirement for self-driving cars. Drones rely on GPS as well. A standard drone without GPS is essentially lost. However, researchers at Nvidia are working on a drone navigation system that relies on visual recognition and computer learning instead, to make sure drones do not get lost, even in the woods.

Nvidia's researchers outfitted an off-the-shelf drone with a navigation system using the company's Jetson TX1 machine learning module, which is fed visual data from two onboard cameras.

Although still at the experimental stage of development, Nvidia's drone was initially designed to make its way through forest trails on rescue missions – looking out for lost hikers or storm damage. However, the Nvidia team believes the low-flying drone could broaden its scope to anywhere GPS coverage is unreliable or not available at all, including canyons, crowded urban environments, or checking stock in an enclosed warehouse. The system could even be adapted to search for damaged cables underwater.

"This works when GPS doesn't," said the team's technical lead Nikolai Smolyanskiy. "All you need is a path the drone can recognize visually."

The drone's navigation system also boasts obstacle avoidance prowess, and has been taught to follow railroad tracks and has been installed in a wheeled robot for zipping around the halls of buildings.

In this case, the project's main proving ground has been forested paths, which can be more of a challenge than a relatively predictable urban setting where points of reference such as mailboxes, buildings and streets remain constant. Wooded areas, by comparison, offer little in the way of uniformity – varying light, lack of markings and trees of different shapes and sizes all help to push a camera-based navigation system to its limits.

"Our whole idea is to use cameras to understand and navigate the environment," Smolyanskiy said. "Jetson gives us the computing power to do advanced AI onboard the drone, which is a requirement for operating in remote environments."

The NVIDIA team isn't the first to pursue a drone that navigates without GPS, but the researchers achieved what they believe is the longest and most stable flight of its kind. Their fully autonomous drone flies along a forest trail for a kilometer (0.6 mile), avoiding obstacles such as trees and maintaining a steady position in the center of the trail.

Nvidia team member Alexey Kamenev developed the deep learning techniques that allowed the drone to smoothly fly along trails without sudden movements that would make it wobble. He also reduced the need for massive amounts of data typically needed to train such a deep learning system.

The drone machine learning training data was watching video captured by three GoPro cameras attached to a rig as Smolyanskiy walked through eight miles of forested trails in the Pacific Northwest. In addition to their own footage, researchers trained their neural network — called TrailNet — video of trails in the Swiss Alps recorded by the Istituto Dalle Molle di Studi sull'Intelligenza Artificiale (IDSIA) in Zurich.

rig.png

Smolyanskiy's camera rig used to collect training data for the drone

The long term goal of the project is to be able to program coordinates into a drone's camera-based navigation system and then have it make the trip on its own. In the shorter term, software is being developed that drone or robot builders can download and use to get their own creations to navigate using only visual data.

The team now plans to create downloadable software for Jetson TX1 and Jetson TX2 so others can build drones and robots that navigate based on visual information alone.

Prev                  Next
Writer's other posts
Comments:
    Related Content