An Indoor Dead-Reckoning Algorithm with Map Matching

Step counting based dead-reckoning has been widely accepted as a cheap and effective solution for indoor pedestrian tracking using hand held device equipped with motion sensors. To calibrate the accumulating error in a dead-reckoning tracking system, extra techniques are always fused to form a hybrid system. Currently various particle filter based map filtering algorithms have been proposed in the literature. However, more improvement can be achieved by using map information. In this paper, a map matching involved step counting algorithm is proposed. It is shown that by applying map matching in an indoor environment with corridors, not only can the estimated location be calibrated, but also the estimated sensor's orientation and the walking direction, which reduces the error in the ensuing location update process. The computational complexity is also less than the particle filter. Experimental results show that map matching algorithm returns more accurate results than the particle filter given the same map information. Further experiments indicate that when incomplete map information is given, the particle filter is much easier in returning a drifted path compared to map matching. This indicates that less map information is required by map matching to achieve a robust performance. This reduced reliance on accurate map information implies less labour cost and error in map maintenance.