Unlocking aerobatic potential of quadcopters: Autonomous freestyle flight generation and execution

Area of Science:

• Robotics
• Aerospace Engineering
• Control Systems

Background:

• Autonomous drones are limited in aerobatic capabilities, especially in complex environments.
• Manual control by skilled pilots allows sophisticated maneuvers, highlighting a gap in current autonomous systems.

Purpose of the Study:

• To develop a comprehensive system for autonomous drones to generate and execute complex aerobatic maneuvers in environments with dense obstacles.
• To enable drones to achieve flight performance comparable to professional pilots.

Main Methods:

• A universal flight representation using discrete aerobatic intentions (topology and attitude changes).
• A spatial-temporal joint optimization trajectory planner for smooth, collision-free, dynamically feasible paths.
• Investigation and compensation for yaw sensitivity issues and differential flatness singularities.

Main Results:

• Ablation studies confirmed the effectiveness of the spatial-temporal optimization and yaw compensation strategies.
• Simulations and experiments validated the system's stability and feasibility for complex aerobatic flight.
• The system allows drones to autonomously perform sophisticated maneuvers previously requiring expert human pilots.

Conclusions:

• The proposed system significantly enhances autonomous drone aerobatic capabilities in complex environments.
• It bridges the gap between manual and autonomous drone flight performance.
• This advancement holds potential for the evolution of aerobatic flight in uncrewed aerial vehicles.