Agilicious: Open-source and open-hardware agile quadrotor for vision-based flight.

Area of Science:

• Robotics
• Autonomous Systems
• Control Theory

Background:

• Autonomous, agile quadrotor flight presents significant challenges in perception, planning, learning, and control.
• Existing research platforms often lack the versatility and standardization needed to accelerate progress.

Purpose of the Study:

• To introduce Agilicious, a novel, open-source, codesigned hardware and software framework specifically for autonomous, agile quadrotor flight.
• To provide a standardized platform that enables researchers to focus on core challenges in quadrotor robotics.

Main Methods:

• Developed a codesigned hardware and software framework featuring high thrust-to-weight and torque-to-inertia ratios.
• Integrated onboard vision sensors, GPU-accelerated compute hardware, and a real-time flight controller.
• Supported both model-based and neural network-based controllers within a versatile software stack.

Main Results:

• Demonstrated Agilicious on various agile tasks, including trajectory tracking at high speeds (70 km/h) and accelerations (up to 5g).
• Showcased vision-based acrobatic flight and obstacle avoidance using only onboard perception in diverse environments.
• Validated its utility for hardware-in-the-loop simulation in virtual reality.

Conclusions:

• Agilicious offers a unique combination of high-performance hardware and flexible software, accelerating research in autonomous quadrotor flight.
• The framework's versatility supports the next generation of scientific and industrial quadrotor applications.