Drone Swarm Learns to Explore Unknown Environments

A swarm of four Crazyfly drones in action. (McGuire et al. / Science Robotics)

(CN) – A team of researchers studied insect behavior to develop pocket-size drones that can explore different environments without human intervention – but are so interdependent that one will swoop in and take a photo when another’s camera doesn’t work.

Researchers from researchers from Delft University of Technology, University of Liverpool and Radboud University of Nijmegen began with the challenge of having small drones – unmanned aircraft weighing just over an ounce – navigate in search-and-rescue simulations with limited computational and sensing functions.

Such capabilities in small devices could have applications for first responders in disaster situations where a larger drone would be less effective or where placing an aid worker would be too risky.

Observing the navigation patterns of insects, scientists theorized that a drone could maneuver through a space more effectively as part of a swarm, according to the study published in Science Robotics.

Tiny drones could comb through a building that is about to collapse, searching for life and relaying information to first responders without raising the threat level for humans.

In the study’s crisis simulation, a swarm of drones was released into an office with an order to quickly locate two “dummies” representing disaster victims.

The swarm of six drones explored about 80% of the open rooms in six minutes, likely an improbable outcome for any single drone, the study said.

The drones also proved to be efficient and cooperative: One drone located a victim in the simulation but due to a camera failure could not send back images to researchers, so another drone quickly swooped in to capture an image of the victim, the study said.

Kimberly McGuire of Delft University in the Netherlands said in a statement that researchers began by controlling drones’ velocity and teaching them to avoid obstacles and each other.

“The biggest challenge in achieving swarm exploration lies at the level of the individual intelligence of the drones,” said McGuire. “We solved this by having each drone carry a wireless communication chip and then making use of the signal strength between these chips – this is like the number of bars shown on your phone that decrease when you move away from your Wi-Fi router in your home. The main advantages of this method are that it does not require extra hardware on the drone and that it requires very few computations.”

McGuire added the scientists developed “a novel type of bug algorithm” to navigate, though it was limited to dealing with obstacles, not developing maps.

“However, the costs of making such a map on tiny robots is prohibitive,” McGuire said. “The proposed bug algorithm leads to less efficient paths but has the merit that it can even be implemented on tiny robots.”

Guido de Croon of Delft University said in the statement that the researchers built upon the knowledge that insects rely not on detailed maps but on landmarks, such as food sources and their nest, to navigate.

“The main idea underlying the new navigation method is to reduce our navigation expectations to the extreme: we only require the robots to be able to navigate back to the base station,” said de Croon. “The swarm of robots first spreads out into the environment by having each robot follow a different preferred direction. After exploring, the robots return to a wireless beacon located at the base station.”

Dutch national science foundation NWO Natural Artificial Intelligence program financed the researchers’ drone project.

In a separate study released this month, University of Utah scientists unveiled a new lens material that could have military applications and allow drones to fly farther and for longer than ever before.

The lens is 20 times thinner than a human hair, a hundred times lighter and a thousand times thinner than current lenses on the market. It has a thermal imaging function for dark environments, using “microstructures” to bend light into a format observable by the human eye.

%d bloggers like this: