In a significant advancement for autonomous aerial robotics, engineers at the University of Hong Kong have created a drone that mimics birds' remarkable ability to navigate complex environments at high speeds.
The Safety-Assured High-Speed Aerial Robot (SUPER) can fly at speeds exceeding 20 meters per second (45 mph) while detecting and avoiding obstacles as thin as 2.5 millimeters – such as power lines or twigs – using only onboard sensors and computing power. Unlike conventional drones that rely on GPS or pre-mapped routes, SUPER operates with complete autonomy in unknown environments.
The compact system, measuring just 11 inches across with a takeoff weight of 1.5 kg, utilizes a lightweight 3D LiDAR sensor capable of detecting obstacles up to 70 meters away with pinpoint accuracy. What makes SUPER truly innovative is its sophisticated planning framework that generates two trajectories during flight: one optimizing speed by venturing into unknown spaces and another prioritizing safety by remaining within known, obstacle-free zones.
"Picture a 'Robot Bird' swiftly maneuvering through the forest, effortlessly dodging branches and obstacles at high speeds. It's like giving the drone the reflexes of a bird, enabling it to dodge obstacles in real-time while racing toward its goal," explains Professor Fu Zhang, who led the research team.
In real-world tests, SUPER achieved a nearly perfect success rate of 99.63% across various challenging scenarios, including flying at high speed, dodging electrical wires, navigating dense forests, and flying at night. The drone also demonstrated excellent object tracking capabilities, successfully following a jogger through dense forest where commercial drones failed.
The technology has wide-ranging applications, particularly in search and rescue missions where MAVs equipped with SUPER technology could swiftly navigate disaster zones – such as collapsed buildings or dense forests – day and night, locating survivors or assessing hazards more efficiently than current drones. The research, published in Science Robotics, represents a milestone in transitioning high-speed autonomous navigation from laboratory settings to real-world applications.