Journal of Intelligent and Robotic Systems

, Volume 55, Issue 4, pp 267–297

Architectures for Cooperative Airborne Simultaneous Localisation and Mapping

Authors

    • ARC Centre of Excellence in Autonomous Systems, Australian Centre for Field RoboticsUniversity of Sydney
  • Salah Sukkarieh
    • ARC Centre of Excellence in Autonomous Systems, Australian Centre for Field RoboticsUniversity of Sydney
Article

DOI: 10.1007/s10846-008-9303-9

Cite this article as:
Bryson, M. & Sukkarieh, S. J Intell Robot Syst (2009) 55: 267. doi:10.1007/s10846-008-9303-9

Abstract

This paper develops active Simultaneous Localisation And Mapping (SLAM) trajectory control strategies for multiple cooperating Unmanned Aerial Vehicles (UAVs) for tasks such as surveillance and picture compilation in Global Positioning System (GPS)-denied environments. Each UAV in the team uses inertial sensor and terrain sensor information to simultaneously localise the UAV while building a point feature map of the surrounding terrain, where map information is shared between vehicles over a data fusion network. Multi-vehicle active SLAM control architectures are proposed that actively plan the trajectories and motions of each of the vehicles in the team based on maximising information in the localisation and mapping estimates. We demonstrate and compare an ideal, centralised architecture, where a central planning node chooses optimal actions for each UAV, and a coordinated, decentralised architecture, where UAVs make their own control decisions based on common shared map information. The different architectures involve varying degrees of complexity and optimality through differing communications and computational requirements. Results are presented using a three-UAV team in a six-degree of freedom multi-UAV simulator.

Keywords

SLAMActive SLAMCooperative controlUAVsInformation
Download to read the full article text

Copyright information

© Springer Science+Business Media B.V. 2009