Autonomous Vision-based UAV Landing with Collision Avoidance using Deep Learning

TL;DR

This paper presents a vision-based autonomous UAV landing system that employs deep learning to ensure collision avoidance during simultaneous landings without communication.

Contribution

It introduces a novel deep learning approach for collision avoidance in autonomous UAV landings using vision-based sensing.

Findings

Successful demonstration of collision-free autonomous landing

Effective deep learning model for real-time collision avoidance

Improved safety in multi-UAV landing scenarios

Abstract

There is a risk of collision when multiple UAVs land simultaneously without communication on the same platform. This work accomplishes vision-based autonomous landing and uses a deep-learning-based method to realize collision avoidance during the landing process.