Giving drones a sense of 'pain' could help them predict instability before it happens

Researchers at Delft University of Technology and Wageningen University & Research have shown that ecological early-warning indicators can identify when controlled drones are nearing instability. The approach draws on critical slowing down, a phenomenon in which a system loses resilience and takes longer to recover after disturbances as it approaches a tipping point. The researchers compare the effect to pain in people: feedback after an injury can indicate which actions remain safe and prompt behavioral adjustment.

The team examined whether this concept, previously used in ecology and climate science, would work in systems governed continuously by flight controllers. At Delft's CyberZoo drone facility, they combined simulations, flight-data analysis, and experiments in which drones were deliberately damaged and flown near loss of control. The work identified combinations of damage, flight conditions, and maneuvers that were most likely to cause loss of control. Experiments included quadrotors with progressively damaged propeller blades.

The indicators use real-time measurements from inexpensive onboard sensors rather than requiring detailed physical models of a drone. According to the researchers, this enables systems not only to recognize approaching instability but potentially to alter their behavior in real time to preserve operation despite damage, analogous to a person limping after an ankle injury.

The study was published in Proceedings of the National Academy of Sciences. Its authors say the method could provide a general way to monitor resilience across engineered systems. Potential applications named in the report include aircraft and vehicle predictive maintenance, critical-infrastructure monitoring, manufacturing quality control, and more reliable autonomous systems such as self-driving cars. The researchers expect its earliest practical impact in the expanding drone sector, where it could help prevent accidents.