Capture, classification and path-planning of safe routes using an aerial camera platform

Authors Organisations
Type

Student thesis: Doctoral ThesisDoctor of Philosophy

Original languageEnglish
Awarding Institution
Supervisors/Advisors
Award date2018
Show download statistics
View graph of relations

Abstract

In real world situations it is sometimes impractical for a human being to enter or traverse an area due to safety or feasibility constraints. In these areas, robotic vehicles are commonly used to navigate and achieve tasks in place of a human subject. Unfortunately ground and waterborne vehicles suffer from a limited world view which may limit their ability to plan safe or optimal routes. By working in tandem with an aerial vehicle, these ground based robots can
increase their world view with a resultant improvement in safety, speed or optimality.

We have developed a closed loop system for the navigation of unknown environments using a two vehicle team. A Helikite aerial vehicle captures and performs texture-based conversion of the aerial images as well as localizing the ground vehicle. The ground robot receives the texture and location information and uses it to plan and execute routes in a closed loop manner using texture difference as a measure of safety.

Results of the tests on the individual components as well as the combined system are presented in terms of speed, optimality and general performance showing the successes and limitations of this approach. A discussion of the results is then presented noting the constraints of the proposed system with suggestions for future work.