������ý

Drones that can’t see the wood for the trees won’t last long in the forest. A method for getting a swarm of them to maintain formation as they pick their way through trees and shrubs means that teams of drones could soon map and take pictures of areas that were previously off limits.

With their ability to fly under the canopy, drones can create 3D maps of forest interiors that aren’t possible from high in the sky. Satellites can only make relatively coarse maps, and clouds can block their cameras. Planes provide better resolution, but are expensive to operate.

Getting multiple drones to carry out a coordinated task means that more ground can be covered in less time. It took a swarm of them just 6 hours to map the Matterhorn mountain in the Alps, for example.

Moving about as a swarm is tricky, however. Single drones can pick their way around obstacles with relative ease, but it is a challenge to do so while maintaining their formation.

To tackle the problem, at the Singapore University of Technology and Design, and at Oregon State University in Corvallis developed a method that enables a drone swarm to find its way through one of the most difficult environments they could think of – a forest.

Irregularly placed trees and shrubs make it hard to both navigate and keep tabs on your swarm mates. “You need to detect obstacles, and at the same time find out where your closest neighbour is,” says Brust.

Follow the leader

Their method is based on a follow-the-leader approach. A lead drone is given a GPS destination from a ground-based controller, and the rest of the pack maintains its position relative to the leader, while avoiding obstacles. If the leader runs low on battery or crashes, the swarm picks a new captain on the fly.

Because trees and leaves – or buildings in a city setting – can also block the swarm’s communication with the ground base, the lead drone collects data from its teammates and periodically rises above the canopy to send images back home.

The researchers tested their system by running a computer simulation in which eight drones approached a tree and took up positions to scan it. The swarm reached its destination and spread out almost as quickly as it took a single drone to arrive there. If the drones can surround a tree without crashing, flying past them shouldn’t be a problem, says Brust.

Combining close-up data from drones with less-detailed maps from satellites would allow ecologists to scale up biomass measurements for huge areas, says at the University of New Mexico in Albuquerque. “It really does change the equation for how we can measure the environment,” he says.

Reference: arXiv,