Vision-Based Guidance for Tracking Dynamic Objects

TL;DR

This paper introduces a vision-based framework for tracking moving objects with unmanned aircraft, utilizing guidance laws and feature point estimators to handle occlusions, validated through simulations.

Contribution

It proposes a novel guidance law based on a rendezvous cone for monocular camera tracking of dynamic objects, including feature point estimators for occlusion management, and provides an open-source simulation environment.

Findings

Effective tracking of dynamic objects demonstrated in simulations

Guidance laws improve continuous tracking within sensor view

Feature point estimators handle occlusions successfully

Abstract

In this paper, we present a novel vision-based framework for tracking dynamic objects using guidance laws based on a rendezvous cone approach. These guidance laws enable an unmanned aircraft system equipped with a monocular camera to continuously follow a moving object within the sensor's field of view. We identify and classify feature point estimators for managing the occurrence of occlusions during the tracking process in an exclusive manner. Furthermore, we develop an open-source simulation environment and perform a series of simulations to show the efficacy of our methods.