Skip to main content

Researchers turn to cameras for improved drone landings

Researchers from the Shibaura Institute of Technology, Tokyo, have improved drone landings with a camera mounted on a servo-controlled platform. The system uses image recognition to detect a landing location from a preset landing symbol.

The two-man team has developed this solution to allow drones to autonomously land for something like a search and rescue mission when the drone isn’t the center of attention.

Drones on the market already use cameras to help with landings. The Skydio 2 can even detect the pattern on the hardshell case it comes with to land on it as it has no landing gear. So, what makes this solution different?

Rather than relying on a stationary camera, the researchers have essentially created a two-axis gimbal mounted underneath the drone to which the camera is then mounted. The gimbal is used to keep the camera looking at the ground rather than moving with the drone.

Using an algorithm and image detection, the camera can find a landing spot autonomously using a database of landing symbols. This means that once the drone is not required anymore, the pilot can tell it to land and find a safe landing spot on its own.

The system scans for a landing symbol at 20 Hertz, which isn’t the fastest rate but still more than enough for a drone looking to find a landing zone. As alternative solutions already exist for consumer drones, it seems that this is more aimed toward current commercial drones.

If you look at all the commercial drones on the market today, many aren’t equipped with any obstacle avoidance or landing detection systems, leaving the pilot to ensure the drone doesn’t hit anything and land safely. Having access to an easy-to-use and affordable system to allow you to tackle one of these problems is a big step forward from a safety aspect.

As the system looks to be made from off-the-shelf hardware and 3D-printed parts, it could also work if the software powering the system is sold or opened sourced by the developers. Either way, researchers creating solutions like this is perfect for the industry as it inspires other people to create a system, if nothing else.

Researchers are also doing more epic things with drones, everything from quantum networks to preventing drones falling out of the sky.

Photo: IEEE



Avatar for Josh Spires Josh Spires

Josh started in the drone community in 2012 with a drone news Twitter account. Over the years Josh has gained mass exposure from his aerial photography work and spends his days writing drone content for DroneDJ as well as pursuing his business.