Smooth Trajectory Collision Avoidance through Deep Reinforcement Learning

TL;DR

This paper introduces a deep reinforcement learning approach for autonomous UAV navigation that emphasizes smooth trajectories and better generalization to unseen environments, using novel reward functions and state designs.

Contribution

It proposes new agent state and reward function designs that improve trajectory smoothness and model generalization in DRL-based UAV navigation.

Findings

Enhanced flight trajectory smoothness with minimized derivatives.

Improved collision avoidance in unseen environments.

Effective DRL-based navigation demonstrated through experiments.

Abstract

Collision avoidance is a crucial task in vision-guided autonomous navigation. Solutions based on deep reinforcement learning (DRL) has become increasingly popular. In this work, we proposed several novel agent state and reward function designs to tackle two critical issues in DRL-based navigation solutions: 1) smoothness of the trained flight trajectories; and 2) model generalization to handle unseen environments. Formulated under a DRL framework, our model relies on margin reward and smoothness constraints to ensure UAVs fly smoothly while greatly reducing the chance of collision. The proposed smoothness reward minimizes a combination of first-order and second-order derivatives of flight trajectories, which can also drive the points to be evenly distributed, leading to stable flight speed. To enhance the agent's capability of handling new unseen environments, two practical setups are proposed to improve the invariance of both the state and reward function when deploying in different scenes. Experiments demonstrate the effectiveness of our overall design and individual components.