Decentralised data fusion applied to a network of unmanned aerial vehicles

Abstract

Recent progress made in the application of decentralised data fusion (DDF) algorithms to a network of heterogeneous sensors, mounted aboard a group of unmanned aerial vehicles (UAVs) is reported. The paper refers to work conducted as part of the Autonomous Navigation and Sensing Experimental Research programme; a collaborative demonstration programme undertaken by BAE Systems and the Australian Centre for Field Robotics at the University of Sydney. The UAVs are equipped with GPS and inertial payload sensors, in addition to multiple terrain sensors chosen from passive vision systems and active mm-wave radar. The sensor nodes are constructed in a fully modular fashion and the decentralised paradigm is applied to both the intra and inter-UAV fusion algorithms. In the first instance, the complete test facility comprising flight platforms, ground station, test-site and simulation environment is used to demonstrate the operational benefits of DDF in a range of multiple-vehicle ground-target tracking scenarios. The advantages of decentralising the fusion algorithms around a sensor network, in terms of system flexibility, scalability and integrity are discussed. Issues of interfacing fundamental data fusion technology with operational and system requirements are also addressed.