Adaptive pedestrian activity classification for indoor dead reckoning systems

摘要: A pedestrian activity classification (PAC) system classifies motion data into activities related to the usage of specific building facilities, such as going up on an escalator or descending a staircase. Recent studies confirm that use PAC significantly reduces indoor localization errors dead reckoning (PDR) exact facility locations in can be retrieved from floor map. However, complexity may become issue for resource constraint mobile devices. We propose novel that, instead using single complex classifier based large set features, employs multiple simple classifiers each trained classify only subset small number features. As moves around inside building, proposed adaptive-PAC dynamically switches right (simple) facilities exist within immediate proximity. By always classifier, has potential drastically reduce average PAC-aided PDR systems. Using experimental data, we quantify and compare performance against conventional PAC. find typical shopping centers, by 91–97% without any degradation accuracy rates.