Multi-sensor Data Fusion Target Tracking Algorithm

Data fusion technology was a new information processing subject rising in the last decade. It is widely used in military and civil field such as target tracking and identifying, medical diagnosis, traffic control and industrial robot. Yet, target tracking and identifying are the most important task that we use this technique to deal with.This thesis concerns the problem of target tracking by data fusion technology. Firstly, we state the basic definition, application and art method of data fusion which is a burgeoning field of information processing. At the same time the application in target tracking of multi-sensor data fusion is summarized. Secondly, the essentials of target tracking, such as the choosing of reference frame, target maneuvering model setup, tracking gate information and Kalman filter are discussed in detail. Then two algorithms are studied in detail.The algorithm based on then weighted mean method is straightforward and often used. From the Lagrange Multiplier Method the optimal weight distribution principle is derived in this paper. In consideration of the asynchronous problem of data from two sensors, a time aligning between the asynchronous data based on least-square technique is presented at the same time. The simulation results show that the tracking performance of multi-sensor fusion is much better.The Interacting Multiple Model (IMM) approach is used in tracking high maneuvering target and the Probabilistic Data Association (PDA) filter is adapted for clutter environments. Bar-Shalom present an algorithm composed of IMM and PDA that can track high maneuvering target in clutter. In this paper, it is extended to multi-sensor application. The results of Monte Carlo simulation show the performance of this method. Lastly, the MS-IMMPDA algorithm is applied in the state's fixed delay smoothing.For each algorithm, the validity is found and the merits and defects are pointed out from Monte Carlo simulation. How to verifying the algorithms through real hardware system design is the emphasis of next stage works.