A new autonomous drone system capable of navigating complex environments at speeds up to 45 mph represents a significant advance in aerial robotics.
Developed by engineers at the University of Hong Kong, the Safety-Assured High-Speed Aerial Robot (SUPER) combines advanced 3D LiDAR technology with real-time trajectory planning to achieve high-speed flight while maintaining safety parameters.
The compact system measures just 11 inches across and incorporates a lightweight Livox MID360 LiDAR sensor, enabling unprecedented environmental mapping precision.
The drone’s detection capabilities extend to 230 feet and can identify obstacles as small as half an inch. This level of precision allows the system to detect and avoid obstacles as thin as a guitar string during high-speed flight.
“SUPER represents a milestone in autonomous MAV systems, bridging the gap from laboratory research to real-world applications,” the research team notes in their published findings.
The engineering team developed a novel approach to processing environmental data. Rather than relying on traditional occupancy grid maps, which require significant computational resources, SUPER processes LiDAR point clouds directly. This technique reduces mapping time to one to five milliseconds, enabling real-time navigation decisions.
Alongside this, the system generates dual flight trajectories simultaneously. One path prioritises safety within known free spaces, while the second optimises speed across all available areas. This approach has demonstrated a significant reduction in failure rates – by a factor of 35.9 – while maintaining the ability to operate at high speeds.
In field tests, the drone achieved sustained speeds of 20 meters per second while successfully navigating through dense forest environments. The system demonstrated consistent performance in various lighting conditions, including low-light scenarios where traditional vision-based systems often fail.
The drone’s thrust-to-weight ratio exceeds 5.0, providing the agility needed for rapid directional changes. This capability proves essential when navigating complex environments at high speeds.
SUPER operates independently in GPS-denied environments using LiDAR-inertial odometry, processing all calculations onboard without external computational support. The system employs gradient-based trajectory planning methods, improving traditional mixed-integer quadratic programming approaches.
The engineering team focused on minimising computational overhead while maintaining reliable obstacle avoidance. This balance allows the system to operate effectively in unmapped environments without compromising safety or speed.
Current technical constraints centre on LiDAR sensor dimensions and processing power requirements. The research team identifies several areas for potential advancement, including integrating more compact sensors and enhanced aerodynamic design.
As LiDAR technology continues to evolve, smaller and lighter sensors could further improve the system’s capabilities. Additional developments in motion prediction algorithms may enhance efficiency in complex environments.
While the technology shows immediate potential for implementation across various sectors, the team must consider the following steps to commercialise it. So, don’t expect to see it in off-the-shelf drones for some time.
We hope that search and rescue operations could benefit from the system’s ability to rapidly and safely navigate complex environments and improve infrastructure inspection applications. Even logistics operations could leverage high-speed autonomous navigation features.
TLDR:
- University of Hong Kong engineers develop autonomous drone with 45 mph capability
- Uses advanced 3D LiDAR for precise navigation and obstacle avoidance
- Achieves 35.9x reduction in failure rates while maintaining high speeds
- Operates independently in GPS-denied environments
- Potential applications in search and rescue, inspection, and logistics
