Using an A-priori learnt motion model with particle filters for tracking a moving person by a linear infrared array network

An infrared sensor has been primarily used as a proximity sensor, its use being mostly limited because of imprecise measurements attributing to the non-linearity of the device as well as its dependence on the reflectivity of the surrounding objects. However, one cannot overlook the fact that these sensors are quite low-cost, can be easily mounted on small robotic systems and are computationally very efficient. In this paper, we try to use an infrared sensor array network to detect a person in its environment and also track the person. A traditional particle filter algorithm using a given motion model poses challenges for tracking a person using infrared sensors, primarily because the motion model might fail to keep up with complex dynamic changes in motion directions coupled with the fact that in the presence of noisy readings or missed detections from the infrared sensor data, small errors in position estimation could add up over time making the particle filter completely lose track of the person. In this paper, instead of using a fixed motion model, we propose to learn a motion model statistically from the initial target motion data and subsequently use this model with the particle filtering approach in order to track the person. In addition, the learnt motion model is regularly updated so as to support the particle filtering approach in establishing a more accurate track of the person.