EN
Search
Photo: Interesting Engineering

HKU team builds single-propeller drone that spins to scan its surroundings

Researchers at the University of Hong Kong (HKU) have developed an autonomous drone that flies on a single motor and scans its surroundings in every direction. The aircraft, named PULSAR II, relies on one LiDAR sensor, a laser-based distance tool, which picks up obstacles overhead, underfoot and to the sides.

Standard quadrotor drones use four motors. Widening their sensing coverage typically means adding cameras, LiDAR units or rotating mechanisms, which increase weight, power use and system complexity. PULSAR II takes a different route by letting the airframe's own spin do much of the scanning.

The propulsion design does not use a swashplate and drives two propeller blades with one motor. According to the researchers, changing the motor's speed within each revolution alters the angle of attack of the blades, which produces the torque needed to steer the drone.

The PULSAR II drone uses its body spin to combine flight with sensing in all directions. Photo: Interesting Engineering

The PULSAR II drone uses its body spin to combine flight with sensing in all directions. Photo: Interesting Engineering

As the body turns, the vertically mounted LiDAR sweeps around the aircraft. The sensor fires laser pulses and measures their reflections to work out distances, and the readings are combined into three-dimensional (3D) maps of the environment. The spin lets it capture ceilings, floors and nearby obstacles that would otherwise fall outside its view.

In indoor experiments, the drone first produced only partial maps while still on the ground. After takeoff, the rotating sensor collected more measurements, producing fuller 3D point-cloud maps with fewer blind spots.

Operating on one motor complicates precise control, because the propulsion does not respond the same way at different rotational speeds, which makes accurate path tracking difficult. To deal with this, the researchers built an experimental propulsion model and a compensation method, and they applied model predictive control on manifolds to improve flight accuracy.

In tests, PULSAR II flew high-speed figure-eight paths and performed in-air flips. Its trajectory-tracking error was substantially lower than that of a conventional control method the researchers used for comparison. The team also built a collision-resistant control system and fitted a lightweight carbon-fiber protective frame.

In one test, a soft ball was dropped onto the hovering aircraft. In others, the drone was set up to hit obstacles and push them aside while moving forward, and it remained stable throughout. The researchers summed up their aim in this way:

“The key is to coordinate the mechanical structure, environmental sensing, and flight control so that this simple design can also handle complex environments,”

Fu Zhang, associate professor at HKU

Researchers tested the drone in dense forests and in underground parking garages with pillars, doorways and overhead pipes. It threaded narrow passages and avoided obstacles at more than 6 meters per second, or about 20 feet per second.

All mapping, obstacle avoidance, path planning and flight control ran on board. The tests did not depend on GPS, external motion-capture equipment or a ground computer. The drone weighs about 1.12 kilograms (2.5 pounds), and in measurements it hovered for 15 minutes and 50 seconds.

The researchers see possible uses in search and rescue, disaster response and 3D terrain mapping, especially where satellite navigation signals are weak or missing. They plan to improve localization in open areas that lack enough laser-reflecting features.

Source