On importance sampling in sequential Bayesian tracking of elderly
Shahram Payandeh · 2016
Caring for elderly is a task which is facing various communities. Enabling seniors to live with dignity and security is one of the main goals in providing the care they deserve. Living in independent dwellings or in care giving facilities with minimum supervision and intervention can enable our elderly to maintain such dignity. Being able to monitor movements and activities of the elderly through various available sensing modalities are the key requirements for promoting such sense of independence. However, due to various limitations of the current sensing technology (i.e. either due to the lack of privacy, distributed and coarseness of the sensed information), the tracking information are subject to occlusions or occasional black-outs. It has been shown that the Sequential Bayesian approach can offer a suitable framework for tracking targets with the expected state-space definitions where their trajectories can follow a non-Gaussian distribution. However, the general approach requires to distribute various sample estimates of the motion in order to capture the prior distribution of the expected trajectories. This paper presents an approach which can be used as a part of such a priori distributions of sample trajectories in order to capture the predicted movements of the elderly in sequential framework. As such, it is possible to reduce the number of samples and offer a more computationally efficient approach.