Guidance & Control Networks for Time-Optimal Quadcopter Flight

TL;DR

This paper develops Guidance & Control Networks for quadcopters that approximate optimal control policies, enabling fast, robust, and adaptable autonomous flight with near-benchmark lap times.

Contribution

It introduces adaptive Guidance & Control Networks that account for rotor velocity limits and incorporate waypoint navigation, advancing autonomous quadcopter control.

Findings

Networks approach lap times of benchmark controllers.

Adaptive algorithms improve robustness to rotor velocity estimation errors.

Learning to navigate multiple waypoints enhances flight flexibility.

Abstract

Reaching fast and autonomous flight requires computationally efficient and robust algorithms. To this end, we train Guidance & Control Networks to approximate optimal control policies ranging from energy-optimal to time-optimal flight. We show that the policies become more difficult to learn the closer we get to the time-optimal 'bang-bang' control profile. We also assess the importance of knowing the maximum angular rotor velocity of the quadcopter and show that over- or underestimating this limit leads to less robust flight. We propose an algorithm to identify the current maximum angular rotor velocity onboard and a network that adapts its policy based on the identified limit. Finally, we extend previous work on Guidance & Control Networks by learning to take consecutive waypoints into account. We fly a 4x3m track in similar lap times as the differential-flatness-based minimum snap benchmark controller while benefiting from the flexibility that Guidance & Control Networks offer.