Vision-based Omnidirection Detect and Avoid System for Multicopters

Abstract

This research aims to develop a Detect and Avoid (DAA) system that utilizes omnidirectional cameras as sensors. This study focuses on quantifying scenarios in which UAS must take evasive action based on images captured by omnidirectional cameras. Initially, other UAS are detected from images captured by the omnidirectional cameras. The detected UAS are then tracked through the omnidirectional cameras, and behavioral data of the UAS within the images are collected. Based on this data, we explore the possibility of determining the presence or absence of collision risks. The results show that it is possible to determine the presence or absence of collision risk from the data detected by the omnidirectional camera. However, it was necessary to set the detected distance and the speed of the UAS in order to indicate evasive maneuvers. In addition, the UAS and sensor would have to be quite close to detect prominent features.