Stanford engineers are working on software that can predict and prevent drone collisions in urban airspace. With use of drone technology on the rise in real estate, journalism, emergency situations and soon package delivery, the need for a system to monitor these unmanned aircraft has increased. Researchers from the Stanford Intelligent Systems Laboratory (SISL), partnered with NASA Ames, are working on an unmanned aerial system traffic management system (UTM) to manage UAV traffic.
"UTM is meant to fulfill a lot of the functions of air traffic control, but it will be in the cloud and largely automated," says Mykel Kochenderfer, an assistant professor of aeronautics and astronautics and SISL Director. According to Stanford, NASA wants the UTM system to support a large number of drone operations without air traffic control operators monitoring each vehicle in the air, so the system will need to include automated conflict avoidance. The software would be able to alert multiple drones when a collision is possible, and then figure out how each can avoid it.
Since it would not be feasible to have an air-traffic control system similar to the one in place for commercial air traffic, NASA is leaning more toward a cloud-based, autonomous UTM system. The first phase of the system was released over the summer and focused on GPS-based corridors for drone flights that can help keep drones safe and still efficient. "That works for farming applications," says Hao Yi Ong, mechanical engineering graduate student. "But once you want to start moving transport drones around urban areas, you can't really do that, because you're not going to block out the airspace over entire residential areas just for when your aircraft is flying through."
The Stanford team is now working on new algorithms to predict and avoid potential collisions. Ong is developing a system that separates multi-aircraft conflicts into paired problems and then picks the best action for each pair of drones from a table predicting each drone's flight path. The server coordinates the solutions and issues an avoidance order to the drones in danger.
To test the system, the researchers conducted over 1 million simulations of encounters between 2 to 10 aircraft. They compared their solution to other solutions, such as a less-coordinated strategy in which each drone only reacts to its closest threat, but the researchers said that the software’s solution proved to be safer and produced faster decision times.
The team will continue to work on the system, looking into handling disruptions such as weather, but hopes their software will be used for NASA’s final system, which Stanford says is estimated to be complete by 2019.