When hikers get lost in a forested area, finding them can be a complex and difficult process. This is because search and rescue teams normally deploy helicopters or drones to facilitate their searches. In forests, the tree canopies block visual access to the forest floor. As such, it is often impossible to spot a hiker beneath the trees from a drone or chopper.  Recently, researchers and search and rescue services have suggested the deployment of autonomous drones to search beneath the trees. These drones would fly through the trees, and search the forest floor. The only challenge with this approach is that drones require GPS signals to guide their flight. Beneath the trees, it’s possible for GPS signals to be extremely weak.

A group of researchers at MIT has proposed a unique approach to solving this problem. They have proposed having a fleet of autonomous drones collaborating together to search under a forest canopy. These drones can team up to search different sections of the forest and identify any lost hiker.  To solve the GPS problems, the drones only use their onboard computers and wireless communications to find their current locations. They do not require GPS. Each drone is be fitted with laser-based systems for scanning its environment. This laser enables it to identify objects in its path and estimate its current position. As the drone flies around, it creates a 3D map the area.

Each drone communicates wirelessly with the other drones in the fleet. It shares the map of the terrain in its environment. An algorithm helps the drones to identify the areas of the forest which have already been searched. This eliminates the duplication of search efforts.  All the drones in the fleet also communicate with an off-board ground station. They send their maps to the ground station. These individual maps are fused to create a combined 3D map of the area. Search and rescue teams can scan the maps to see any signs of a lost hiker.

In tests carried out in randomly selected forest areas, each drone was able to map an area of 20 square meters in 2 – 5 minutes. The system was capable of combining the maps in real-time. The drones excelled in a number of metrics including speed, identifying forest objects and the accuracy of the maps. The drones also excelled in reading and interpreting the maps.  In the real world, such a system would be combined with an algorithm which enables it to identify the lost hiker.

This wasn’t included in the tests, though. However, the 3D map produced by the ground station was clear enough for human observers to distinguish a human being from other forest features.  Ultimately, the new system opens a door for search and rescue teams to scour forested areas fast and efficiently without having to send out teams of humans.