Scientists from Zhejiang University in China have developed a drone swarm that can navigate through a thick bamboo forest without the assistance of humans.

To stay in formation, the ten palm-sized drones interact with one another, exchanging data acquired by onboard depth-sensing sensors to scan their surroundings. This strategy indicates that if a drone’s path in front of it is obstructed, it may calculate an alternative route using data acquired by its neighbors. According to the researchers, the swarm may utilize this technology to track a human passing through the same location.

Drone swarms like these might be employed for disaster assistance and ecological assessments in the future, according to the researchers in a report published in the journal Science Robotics.

“During natural disasters like earthquakes and floods, a swarm of drones may search out, direct, and deliver essential supplies to trapped people,” they write. “In wildfires, agile multicopters, for example, may quickly collect data from a close perspective of the front line without endangering humans.”

Experts, on the other hand, believe the project has obvious military applications. Drone swarms are currently being developed by several countries, including the United States, China, Russia, Israel, and the United Kingdom. The most typical uses for this work are surveillance and reconnaissance, although the same technology might be used to monitor and target both soldiers and civilians.

This research, according to Elke Schwarz, a senior lecturer at the Queen Mary University of London who specializes in the use of drones in battle, has definite military potential.

The current conflict between Russia and Ukraine showed how rapidly drone technology can be modified for use on the battlefield and the destructive impact it may have. Both sides in the battle are employing low-cost consumer drones for reconnaissance and, on occasion, offensive purposes. Drones can be used to dump grenades on opposing forces, for example.

Drone swarms are possibly more deadly than lone robots not because of their numbers, but because of their autonomy. A single human cannot control a swarm of ten drones at the same time, but if this duty can be delegated to algorithms, military strategists are more inclined to accept the employment of autonomous systems in combat.

Drone swarms are currently restricted in their use. Making extravagant light shows is the most popular real-world use. Drones, on the other hand, in these cases follow pre-programmed courses in open regions, utilizing tracking technologies such as GPS to find themselves.

Zhejiang University’s study improves on this by controlling the drones’ flight using solely on-board sensors and algorithms, rather than mapping their surroundings beforehand. “This is the first time a swarm of drones has successfully flown outside in an unstructured environment, in the wild,” according to Enrica Soria, a drone swarm researcher at the Swiss Federal Institute of Technology Lausanne. The work was also “impressive,” according to Soria.

The authors point out in their research that methods for drone swarms typically follow one of two programming paradigms: “bird” or “insect.” A “bird” swarm aims to steer drones along with lengthy, flowing courses (the latter being the researchers’ strategy), whereas an “insect” swarm focuses on quick, reactive movements that need less advanced preparation. Thinking like an insect takes less computational power, while planning like a bird saves energy. Programming bird-like behavior has grown increasingly achievable as hardware computer capability advances.

Although the focus of drone swarm research is generally on technological advances, Schwarz points out that this might hide the more difficult problems of how much work should be applied. Norbert Wiener, a 20th-century American mathematician whose work lay the groundwork for AI research, is quoted.