A mobile sensor configuration and multi-object tracking algorithm

This paper proposes a mobile sensor configuration algorithm for multi-object tracking issues. The configuration process involves the selection of sensor and sensor controlling of direction. We will show that the configuration process will affect the precision of observation and further the estimated performance. Previous studies rarely consider the sensor configuration and multi-object tracking jointly under cluttered environment. We first built the dynamical and controlling models for the mobile sensors, where the observational noises are depended on the relative state changes, or specifically, the relative difference of distance and velocity between sensors and objects in x and y axles respectively. Based on our previously metric of Power and Cost (PaC), we further proposed a new criterion named probability of missing detection, power and cost (PMD-PaC) and describe the configuration as the linear program issue and derive the trajectories for the mobile sensors. The association and object estimation are respectively given by the proposed sequence multiple hypothesis tractor (S-MHT) and Kalman filter. The final experiment is given to verify the proposed algorithm.