Simultaneous learning of state-to-state minimum-time planning and control

TL;DR

This paper introduces a reinforcement learning framework that enables UAVs to learn minimum-time flight policies capable of navigating between any start and goal states, balancing agility and stability, with validation in simulation and real-world outdoor environments.

Contribution

It presents a novel RL-based approach that simultaneously learns planning and control for minimum-time UAV flights, generalizing beyond predefined tracks using curriculum learning and PMM trajectories.

Findings

The method outperforms NMPC in simulation.

Curriculum learning enhances training efficiency and generalization.

The policy is robust in outdoor real-world tests.

Abstract

This paper tackles the challenge of learning a generalizable minimum-time flight policy for UAVs, capable of navigating between arbitrary start and goal states while balancing agile flight and stable hovering. Traditional approaches, particularly in autonomous drone racing, achieve impressive speeds and agility but are constrained to predefined track layouts, limiting real-world applicability. To address this, we propose a reinforcement learning-based framework that simultaneously learns state-to-state minimum-time planning and control and generalizes to arbitrary state-to-state flights. Our approach leverages Point Mass Model (PMM) trajectories as proxy rewards to approximate the true optimal flight objective and employs curriculum learning to scale the training process efficiently and to achieve generalization. We validate our method through simulation experiments, comparing it against Nonlinear Model Predictive Control (NMPC) tracking PMM-generated trajectories and conducting ablation studies to assess the impact of curriculum learning. Finally, real-world experiments confirm the robustness of our learned policy in outdoor environments, demonstrating its ability to generalize and operate on a small ARM-based single-board computer.