Determining the track of a moving object by Kalman and bootstrap method with multisensor data

Because data from multiple sensory sources always includes system errors and random errors, the estimation precision of the track is affected. A method which can reduce those two kinds of errors is presented. The method considers the data from two measuring sources. Kalman filtering theory is used to estimate the system errors of the two instruments. System errors are then compensated. As a result, a group of multiple tracks including only random errors is obtained. Bootstrap methods are used to estimate the actual track of the moving object. The method adopts a linear model and avoids the nonlinear problem in the moving equation. Experiments indicate that the estimation precision is satisfactory