New drone safety supervisor prevents crashes in real time

A novel safety monitoring system for quadrotor drones has been developed by an engineer at the University of Houston, designed to maintain course and avoid hazards during operation. The system acts as a real-time "safety supervisor," intervening when unexpected events, such as strong wind gusts, could push the drone off its intended path.

Quadcopters, known for their agility and hovering capabilities, are widely used for tasks like structural inspections and photography due to their ease of flight and maneuverability in confined spaces. The new supervisor module, integrated onboard the drone, is engineered to provide run-time assurance, a safety mechanism that continuously oversees the drone's flight.

Marzia Cescon, a professor of Mechanical & Aerospace Engineering at the UH Cullen College of Engineering, explained that the module monitors the drone's tilt and position in real time. This creates an "invisible fence" defining safe operational boundaries, with the algorithm designed to steer the drone away from these limits if it predicts proximity to danger.

Technically, the supervisor is implemented using Control Barrier Functions (CBFs), a mathematical framework that determines if a drone is approaching unsafe conditions. If so, the CBF takes control of the flight to ensure safety. This work demonstrates the practical integration of CBF-based run-time assurance schemes with standard optimal controllers on actual hardware, bridging the gap between theoretical concepts and deployable real-world control systems.