Qualcomm Technology Helps Drones Learn About Their Environments

Drones took up a lot of floor (and air) space at the Consumer Electronics Show in Las Vegas this year, but one of the most impressive displays of new drone tech came from Qualcomm.

The chipmaker showed off the latest iteration of its Snapdragon Flight Drone Platform [pictured above], which allows for flight control and machine learning in real time onboard a drone.

With machine learning, drones can fly autonomously without knowing beforehand what they might encounter. Other drones have artificial intelligence that can do things like follow a moving object or avoid obstacles too, but Qualcomm's tech shows more advanced computing that can actually understand what the drone encountered in mid-air and create a flight path.

Qualcomm put together a set at its booth at CES that resembled the inside of a cluttered warehouse, showing how its drone processing and decision-making technology is nimble enough to allow drones to operate indoors and in unpredictable settings without using any GPS. That's important because GPS doesn't work well indoors, an environment where drones can be useful for inspecting buildings after a storm or for security in a warehouse, for example.

Qualcomm is showing off its autonomous, synchronized flight drone tech at #CES2017. These drones are not using GPS. pic.twitter.com/GhCpDk9voO -- Recode (@Recode) January 6, 2017

The Snapdragon onboard navigation processor only weighs 12 grams, which is less than a AAA battery, according to Sarah Gibson, a senior engineer at Qualcomm who spoke at the company's keynote address at CES this year. All of the drone computing, like the machine learning and flight control, happens on that processor without any off-board computing.

To get a sense of this technology at work, watch the video below of a drone equipped with Qualcomm's Snapdragon processor at CES. The drone identifies the objects it encounters, like a ladder, pallets and barrels.


Comments are closed.