HANGZHOU, April 17 (Xinhua) -- A team of scientists in China has unveiled an extraordinary drone system capable of autonomously executing aerial aerobatics with a level of precision that surpasses even the most skilled human pilots.

The breakthrough, published in the journal Science Robotics on Thursday, could revolutionize search and rescue missions, enhancing their efficiency and effectiveness in intricate environments such as forests, ruins, tunnels and sewers.

The algorithms developed by researchers from Zhejiang University enable drones to autonomously perform complex aerial maneuvers such as barrel rolls and power loops, all while skillfully navigating through obstacles. These are skills previously required human control of professional pilots.

NATURAL WISDOM

Bats roosting in caves, crows performing aerial stunts in cities, and falcons diving at 180 degrees to catch prey -- these are survival skills shaped by evolution.

"Such biological intelligence is exactly what current drones need to excel in complex aerial tasks," said Gao Fei, who led the research.

Similarly, top drone pilots can perform what can only be described as an "aerial ballet" in complex environments, using an intuitive movement language akin to a dancer's muscle memory. However, current drone algorithms rigidly convert movements into manually adjusted parameters.

"We've created what is essentially 'building blocks' for drone acrobatics," explained Gao, the paper's corresponding author. "The system automatically converts simple commands into smooth, complex aerial movements."

The framework integrates safety measures and condenses desired trajectories into solvable metrics for maneuvers. It also leverages the drone's onboard computing and localization to navigate using preloaded maps.

AEROBATICS

In outdoor tests, the autonomous drone successfully executed three consecutive acrobatic maneuvers over a 220-meter course without human intervention, all while using the propulsion levels of commercial drones.

The system also showed remarkable precision in confined spaces. In a narrow 3.5-meter-high obstacle course, the drone flew backward through 80-centimeter-diameter rings and performed snake-like maneuvers through tunnels.

It maintained tracking accuracy within 15 centimeters while flying at a speed up to 7 meters per second in repeated tests, according to the study.

When pitted against China's top drone pilots, the autonomous system achieved a 100 percent success rate navigating through six 1.2-meter-wide narrow gates, while human pilots succeeded only 12.5 percent of the time.

"Even the best human pilots need more safety space to recover after performing aerobatics," said Wang Mingyang, the study's lead author. "Our system can perform flips in smaller space."

The results have shown that smarter algorithms can optimize movement patterns, allowing quadrotor drones to perform more graceful and complex flight maneuvers without the need for hardware upgrades.

The technology could transform disaster response by enabling drones to navigate collapsed buildings or deploy monitoring equipment near volcanic eruptions, and even operate in near-Earth space filled with debris, all through dynamic flight adjustments, said Gao.

The researchers acknowledge the current system's reliance on preloaded maps rather than real-time obstacle detection in unknown environments as a limitation.

Their future work will focus on creating fully autonomous systems that can instantly adapt to unknown and changing environments, added Gao.