This week during the International Symposium on Experimental Robotics conference, MIT researchers will present a paper in which they describe an autonomous system for a fleet of UAS that would enable UAS to search for lost hikers in forests.

GPS signals used to guide UAS can be unreliable or nonexistent in forest environments, so the MIT researchers propose a strategy where UAS only use onboard computation and wireless communication—effectively eliminating the need for GPS—as they collaboratively search under dense forest canopies.

Each autonomous quadrotor UAS is equipped with laser-range finders for position estimation, localization, and path planning. While flying, the UAS create an individual 3-D map of the terrain, and a UAS would know when it has fully mapped an area thanks to algorithms helping it recognize unexplored and already-searched spots.

Using an off-board ground station, individual maps from multiple UAS are fused into a global 3-D map that can be monitored by human rescuers.

In a real-world implementation—not in the current system, the researchers note—the UAS would be equipped with object detection to identify a missing hiker. When that person is located, the UAS would tag the hiker’s location on the global map, and this information can be used to plan a rescue mission.

“Essentially, we’re replacing humans with a fleet of drones to make the search part of the search-and-rescue process more efficient,” explains first author Yulun Tian, a graduate student in the Department of Aeronautics and Astronautics (AeroAstro).

Researchers tested multiple UAS in simulations of randomly generated forests. They also tested two UAS in a forested area within NASA’s Langley Research Center. During both experiments, each UAS mapped an approximately 20-square-meter area in about two to five minutes and collaboratively fused their maps together in real-time.

Researchers note that the UAS performed well across a number of different metrics, including overall speed and time to complete the mission, detection of forest features, and accurate merging of maps.

Researchers mounted a LIDAR system on each UAS, creating a 2-D scan of the surrounding obstacles by shooting laser beams and measuring the reflected pulses. While this method can be used to detect trees, to UAS, individual trees appear “remarkably similar,” researchers say. So, if a UAS can’t recognize a given tree, it can’t determine if it’s already explored an area.

With this in mind, researchers programmed their UAS to instead identify the orientation of multiple trees, which is a lot more distinctive. Using this method, when the LIDAR signal returns a cluster of trees, an algorithm calculates the angles and distances between trees to identify that cluster.

“Drones can use that as a unique signature to tell if they’ve visited this area before or if it’s a new area,” Tian says.

Researchers say that this feature-detection technique helps the ground station accurately merge maps.

The UAS generally explore an area in loops and produce scans as they go, while the ground station continuously monitors the scans. When two UAS loop around to the same cluster of trees, the ground station calculates the relative transformation between the UAS, and then fuses the individual maps to maintain consistent orientations, to ultimately merge the maps. 

“Calculating that relative transformation tells you how you should align the two maps so it corresponds to exactly how the forest looks,” Tian says.

In the ground station, robotic navigation software called simultaneous localization and mapping (SLAM) uses the LIDAR input to localize and capture the position of the drones, which helps it fuse the maps accurately. SLAM not only maps an unknown area, but also keeps track of an agent inside the area.

The end result is a map with 3-D terrain features, as trees appear as blocks of colored shades of blue to green, depending on height, and unexplored areas are dark but turn gray as they’re mapped by a UAS. On-board path-planning software tells a UAS to always explore these dark unexplored areas as it flies around.

Tian says that producing a 3-D map is more reliable than simply attaching a camera to a UAS and monitoring the video feed. For example, transmitting video to a central station requires a lot of bandwidth that may not be available in forested areas.

MIT researchers note that an important innovation is a novel search strategy that lets the UAS more efficiently explore an area. A more traditional approach says that a UAS would always search the closest possible unknown area, but this could be in any number of directions from the drone’s current position, as the UAS usually flies a short distance, and then stops to select a new direction.

“That doesn’t respect dynamics of drone [movement],” Tian says. “It has to stop and turn, so that means it’s very inefficient in terms of time and energy, and you can’t really pick up speed.”

Alternatively, the researchers’ UAS explore the closest possible area, while considering their current direction. Researchers believe this can help the UAS maintain a more consistent velocity.

This strategy where the UAS tends to travel in a spiral pattern covers a search area much faster.

“In search and rescue missions, time is very important,” Tian says.

Researchers compared their new search strategy with a traditional method in the paper, and compared to that baseline, their strategy helped the UAS cover “significantly more area, several minutes faster and with higher average speeds.”

One limitation for practical use, though, is that the UAS still have to communicate with an off-board ground station for map merging. During their outdoor experiment, researchers had to set up a wireless router that connected each UAS and the ground station.

Researchers hope that in the future, they can design UAS to communicate wirelessly when approaching one another, fuse their maps, and then cut communication when they separate. In that case, the ground station would only be used to monitor the updated global map.