Learning Agile Flight Maneuvers: Deep SE(3) Motion Planning and Control for Quadrotors

TL;DR

This paper introduces a deep reinforcement learning approach for agile quadrotor flight through dynamic narrow gates, combining model predictive control with neural network-based adaptive references for robust and efficient navigation.

Contribution

It presents a novel reinforcement learning framework with adaptive SE(3) references and a binary search algorithm for real-time dynamic environment adaptation.

Findings

Robust flight through dynamic gates demonstrated in high-fidelity simulations.

Adaptive references improve safety margins and navigation accuracy.

The method generalizes well across various initial conditions and gate trajectories.

Abstract

Agile flights of autonomous quadrotors in cluttered environments require constrained motion planning and control subject to translational and rotational dynamics. Traditional model-based methods typically demand complicated design and heavy computation. In this paper, we develop a novel deep reinforcement learning-based method that tackles the challenging task of flying through a dynamic narrow gate. We design a model predictive controller with its adaptive tracking references parameterized by a deep neural network (DNN). These references include the traversal time and the quadrotor SE(3) traversal pose that encourage the robot to fly through the gate with maximum safety margins from various initial conditions. To cope with the difficulty of training in highly dynamic environments, we develop a reinforce-imitate learning framework to train the DNN efficiently that generalizes well to diverse settings. Furthermore, we propose a binary search algorithm that allows online adaption of the SE(3) references to dynamic gates in real-time. Finally, through extensive high-fidelity simulations, we show that our approach is robust to the gate's velocity uncertainties and adaptive to different gate trajectories and orientations.