Swarm of micro flying robots in the wild

Area of Science:

• Robotics
• Artificial Intelligence
• Computer Science

Background:

• Dense, cluttered environments like forests pose significant challenges for current aerial robot deployment.
• Swarm coordination in unknown, confined spaces requires advanced navigation and planning capabilities.

Purpose of the Study:

• To develop miniature, autonomous drones with an extensible trajectory planner for swarm navigation in challenging environments.
• To enable timely and accurate drone navigation using limited onboard sensor data.

Main Methods:

• Developed a trajectory planner optimizing for flight efficiency, obstacle avoidance, collision avoidance, and swarm coordination.
• Implemented spatial-temporal joint optimization for trajectory deformation and time allocation.
• Integrated the planner into a palm-sized swarm platform with onboard perception, localization, and control.

Main Results:

• Achieved high-quality trajectory generation in milliseconds, even in highly constrained environments.
• Demonstrated superior performance in trajectory quality and computation time compared to benchmarks.
• Validated system extensibility through real-world field experiments.

Conclusions:

• The developed system enhances aerial robotics by enabling navigation in cluttered environments.
• The approach offers extensibility for diverse task requirements and facilitates swarm coordination without external infrastructure.
• This work significantly advances the capability of aerial robot swarms in wild, unstructured settings.