Join or Sign in

Register for your free asmag.com membership or if you are already a member,
sign in using your preferred method below.

To check your latest product inquiries, manage newsletter preference, update personal / company profile, or download member-exclusive reports, log in to your account now!
Login asmag.comMember Registration
https://secutechindia.tw.messefrankfurt.com/mumbai/en/planning-preparation/exhibitors.html
INSIGHTS

Now, AI enables self-flying drones to understand what they see

Now, AI enables self-flying drones to understand what they see
Iris Automation’s solution, based on AI, allows drones to actually see the world the way a pilot does and make real-time decisions to help the mission or to actively keep the drone safe from any possible collisions.
With the advancements in technology and reducing costs, autonomous drones are becoming increasingly popular in various industries. The fact that you can assign a drone to accomplish a task without having a pilot navigating it has made them attractive to customers who wish to automate their processes, saving labor and operational costs.

However, as the number of self-flying drones increase, there are also growing concerns on technologies to ensure these machines operate without colliding against each other. In fact, this fear has been one of the major factors hindering the development and market adoption of autonomous drones in general.

Fortunately, there are now some innovative companies trying to deal with this issue and offer solutions that would enable drones to navigate through the skies without problems. One of them is the San Fransisco-based Iris Automation. According to Alexander Harmsen, CEO of the company, Iris Automation has created a solution that allows the drones to see what is around them.

“We’ve built a collision avoidance ‘Sense and Avoid’ system for industrial drones so that they can see the world the way a pilot can,” Harmsen said. “It’s 3D reconstruction and dynamic tracking of moving objects at long ranges, in order to keep the aircraft safe while they fly in national airspace, potentially beyond-visual-line-of-sight.”

The solution is completely based on computer vision, harnessing the recent breakthroughs in embedded computing and cameras to its advantage. The company has built proprietary algorithms that allow their system to operate in outdoor, unstructured environments, onboard in real-time ranges that apparently no other company has ever attempted. Harmsen believes that fitting all this into a tiny module that weighs 300 grams and is actually affordable will be a game changer for the entire industry.

At the core of Iris Automation’s solution is a software that runs on an embedded GPU. The software is capable of understanding, segmenting and classifying objects in the surrounding, as well as being able to completely, geometrically, recreate the environment into an internal world model.

“Propagating forward in time allows the system to assess whether we will actually collide with anything or not, allowing the system to take an avoidance action,” Harmsen added. “Of course a lot of this is not possible without the underlying deep learning and artificial intelligence onboard.”

The arrival of AI has clearly revolutionized the drone industry. Before this year, drones were blindly following GPS waypoints or following the direct instructions from a pilot’s joystick or controller. Iris Automation’s solution, based on AI, allows drones to actually see the world the way a pilot does and make real-time decisions to help the mission or to actively keep the drone safe from any possible collisions.

“There has been a huge artificial constraint in the commercial market that is really holding back progress: lack of trust in autonomous vehicles,” Harmsen said. “We can bring that trust and safety to drones and actually allow them to get permissions and insurance to fly beyond-line-of-sight, on scale and finally allow these drones to be flown completely autonomously – the way they were always meant to be flown.”


Share to: