Improved wireless tracking using radio frequency and video sensors

The use of wireless localisation and tracking is widespread in outdoor environments and provides improved functionality and benefits to many applications. Extending the same to indoor environments is a critical prerequisite for many applications that require location awareness. Wireless systems that use radio frequency (RF) time-of-arrival (TOA) technology for tracking is recognised as the best technology for indoors, however, multipath interference significantly impacts the performance of TOA systems. In this paper, we consider data fusion of an RF tracking system with video tracking for improved indoor positioning for applications that allow the use of the two systems simultaneously. An important distinction of the two systems is that the RF system does not exhibit measurement origin uncertainty, whereas the video camera does. Further, the RF system generally provides biased positions while indoors. The algorithm that we propose effectively solves the measurements origin uncertainty in the video camera and accounts for the bias in the RF measurements. In a challenging simulation scenario it achieves a 0.2 m positional accuracy and 85% data association accuracy. Preliminary field trial results also show the effectiveness of the proposed algorithm.