An agile new drone has demonstrated the ability to follow a person jogging through the forest and avoid thin obstacles like power lines, all while moving at speeds of 20 meters per second or more.
According to its Hong Kong developers, the SUPER drone, meaning “safety-assured high-speed aerial robot,” will be used for search and rescue or disaster relief applications. The new micro air vehicle (MAV) demonstrates capabilities far outpacing earlier attempts to construct similar tiny autonomous drones.
Avian Agility Unmatched By Any Drone
In nature, birds are capable of incredible feats of agility, which remain difficult to replicate in MAVs. Advances in autonomous drones haven’t matched the ability of creatures like birds to intuitively navigate their surroundings at a glance, weaving around tree branches and other obstacles.
Short camera ranges, slow speeds, and limited safety precautions to mitigate cluttered environments have all plagued attempts to develop a drone capable of matching its natural rival.
Programming choices have also limited MAVs’ effectiveness due to tradeoffs between speed and safety. Other autonomous drones have reached similar speeds to SUPER, but they only operate on human-constructed race tracks or in large open areas. They focus on a safe flight that sacrifices quickness, resulting in performance that reaches only about half the speed of SUPER.
Designing a ‘Super’ Drone
A team led by Yunfan Ren at the University of Hong Kong developed SUPER, a compact, lightweight aerial robot with onboard sensing and computing systems designed to bring MAV technology closer to matching avian capabilities.
The drone maps its environment in a 70-meter radius through a light detection and ranging (LiDAR) sensor, which it then uses to plot routes using a two-trajectory strategy. The drone’s computer analyses the LiDAR data, from which it calculates both the safest route regarding obstacle avoidance and a faster route while traversing unknown spaces.
The drone then oscillates between the two routes to optimize its path for speed and collision avoidance, which it successfully demonstrated in real-world tests. Those real-world tests took place during daylight and nighttime over densely wooded terrain, through which the drone successfully tracked and followed a jogger.
In comparison tests, a commercial drone equipped with optical sensors eventually lost track of the target SUPER could follow for the duration.
Finding The Best Path
Beyond tracking a moving object, SUPER also significantly outperformed the commercial drone in successful object detection and avoidance throughout its flight path.
“SUPER effectively addresses the challenges of navigation in complex real-world environments, offering valuable insights for advancing autonomous robots from laboratory settings to real-world applications,” Ren and his colleagues write in a recent paper detailing their findings.
Its most intriguing advancement is SUPER’s ability to easily navigate tight spaces and avoid narrow objects, including wires and small trees. As the jogger the MAV followed moved from lighter to more densely wooded terrain, the human struggled to successfully navigate the foliage, bending and stooping to avoid branches.
Because the commercial drone utilized in the tests was physically larger than SUPER, the Hong Kong University team attached sticks to their MAV to increase its width in the interest of a fair contest. Despite the handicap, SUPER’s performance remained impressive during the team’s test flights.
The Drones of Tomorrow
LiDAR is key to SUPER’s performance, but the Hong Kong University team says this remains a limiting factor for developing even more impressive MAVs. SUPER’s onboard LiDAR is much lighter than earlier iterations of the technology, though it still significantly outweighs less effective visual sensors.
Additionally, denser and longer-range sensing capabilities would substantially improve the raw data engineers must use to design flightpath algorithms.
According to the team, the final step will be developing the flight path algorithm to dynamically track other moving objects and predict their trajectory to aid collision avoidance.
The paper “Safety-Assured High-Speed Navigation for MAVs” appeared on January 29, 2025, in Nature.
Ryan Whalen covers science and technology for The Debrief. He holds an MA in History and a Master of Library and Information Science with a certificate in Data Science. He can be contacted at ryan@thedebrief.org, and follow him on Twitter @mdntwvlf.