AI Software Reduces the Risk of Drone Collisions

ai-reduces-the-risk-of-drone-collisions

Boston based Neurala, Inc. creates software that makes drones, toys, consumer electronics, self-driving cars, and Internet of Things (IoT) devices more autonomous through a combination of deep learning, machine learning, computer vision, and machine reasoning. Founded in 2006 by Max Versace, Anatoly Gorshechnikov, and Heather Ames, the company initially developed software for NASA and for the U.S. Air Force Research Laboratory and is a 2013 graduate of the TechStars startup accelerator program. Teal Drones, which bills its product as the world’s fastest production drone, recently announced it will incorporate Neurala’s follow-me software in its products. This supports Tractica’s forecast that spending on artificial intelligence (AI) software in the aerospace industry will grow significantly over the next 10 years.

Why There Is a Need

Drones used for industrial, research, security, and military purposes are proliferating rapidly. Loss, destruction, or damage to drones due to collisions with trees and other objects such as buildings is very expensive and frustrating for their owners and operators. The Federal Aviation Administration (FAA) even considers the danger of drones colliding with aircraft to be a major threat to safety. In the same way that a flock of birds can lead to mid-flight catastrophe for a large or small airplane, collision with a drone can be a tragic, although rare, event. Although it has never happened, even the potential for such an event makes it more likely that drones will be subject to severe government restrictions.

How it Works

With the help of deep learning and machine vision technology, Neurala, which specializes in AI, is tackling the problem of drone collisions. The company trained its software by feeding it video images of potential collisions from Microsoft Flight Simulator. Neurala’s software notifies drone users and operators whenever it recognizes similar, real-time images from a single camera mounted on the drone.

Benefits

Collision avoidance mechanisms raise drones’ profiles and broaden their acceptance as safe tools that can be used in a variety of benign and useful ways, including for search and rescue, mapping, agriculture, weather prediction, and package delivery. The technology will also allow more drones to fly more closely together, which will increase their commercial appeal.

Potential Problems

Although collision avoidance is already central to the automated piloting systems that keep the vast majority of airplanes and their passengers safe, the public is often skeptical about automated safety systems. Instead of recognizing the serious nature of many drone activities, people may also dismiss them as nuisances and toys that do not need collision avoidance mechanisms as much as they need to be grounded. Also, hacking and spoofing the computer vision that is key to collision avoidance programs can result in serious accidents that may delay how long it will take drones to assume their place as trustworthy citizens of the skies.

Market Forecast

Trendsetters like Neurala are part of the reason we feel the aerospace sector is well positioned to leverage AI capabilities. In Tractica’s upcoming Artificial Intelligence for Enterprise Applications report, we forecast that spending on AI software in the aerospace industry will grow from $17 million in 2016 to $2.9 billion by 2025.

AIE-16 chart Aerospace

Comments are closed.