Researchers Use Artificial Intelligence to Prevent Drone Collisions

In the next few years, it is expected that the number of autonomous drones in unmanned controlled airspace below 400 feet will significantly increase. Experts predict that by 2027, there will be nearly one million commercial unmanned aircraft systems (UAS) in the United States, performing tasks such as package delivery, traffic monitoring, and emergency assistance.

A research team led by Lanier Watkins and Louis Whitcomb from the Institute for Assured Autonomy has used artificial intelligence to model a system that coordinates drone traffic more safely by replacing certain human decision-making processes. Their research results have been published in the journal "Computer".

Watkins said, "We wanted to see if we could use different artificial intelligence methods to handle these operations safely, and the answer is yes." "Our simulation system enhances the safety and scalability of UAS operations below 400 feet using autonomous algorithms."

To address the growing challenges of UAS traffic, the Hopkins team evaluated the impact of autonomous algorithms in a simulated 3D airspace. Based on previous research, they found that using collision avoidance algorithms can significantly reduce accidents. They discovered that adding a strategic resolution algorithm that controls traffic time to avoid collisions increases safety and almost eliminates airspace accidents.

The researchers also equipped their simulator with two tools that simulate real-world conditions. The "noise sensor" mimics the unpredictability of real-world conditions and makes the system more adaptable, while the "fuzzy interference system" calculates the risk level of each drone based on various factors, including the proximity of obstacles and the degree of adherence to planned flight routes. Watkins and Whitcomb stated that these methods enable the system to make autonomous decisions to prevent collisions.

Whitcomb said, "Our research considers various variables, including the case of 'rogue drones' deviating from planned flight routes. The results are very promising."

The team plans to further improve their simulation by introducing dynamic factors such as weather and other real-world factors to obtain more accurate results.

Watkins explained, "This work has been studied through simulations of environments and performance in systems that are being considered for deployment in future airspace by third parties, as well as in academic and foundational research communities such as IEEE and ACM."

"This work helps researchers understand the performance of autonomous algorithms for protecting airspace in a 3D simulated airspace, considering noise and uncertainty, and emphasizes the need for continuous monitoring of the results of these autonomous algorithms to ensure they do not reach potential failure states."