Moving Target Localization/Tracking Algorithms based on Measurement Fusion

Abstract

Positioning is one of key technology which is widely used in many types of equipment such as smartphones and autonomous vehicles. Therefore, high-level accuracy and reliability in positioning are required in various indoor and outdoor environments. In addition, it is emerging as a very important technology to track the position of a moving target in real time rather than simply estimating the position of a stopped target. It is the core technology of the automobile industry and the future service industry that manages unmanned moving targets such as autonomous vehicles and drones by using stable and always available positioning technology regardless of environmental constraints. There are a lot of positioning techniques which are used now. The Global Navigation Satellite System (GNSS) and Differential GNSS (DGNSS), which are widely used for location estimation in outdoor environments, are vulnerable to other radio interference and are difficult to use in environments where the LOS (Line of Sight) is not guaranteed. WiFi-based technology can be used indoor environment only because of the large difference in positioning accuracy depending on the number of nearby APs and installation interval. Inertial Measurement Unit (IMU) -based positioning system is independent of radio reception conditions and provides a precise relative position in short travel distance. However, due to the limitation of accumulation-based location estimation method, as the travel distance increases, a cumulative error occurs. Radar and Lidar can measure relative distance, angle by using the time it takes for the transmitted radio waves to reflect back to the object. However, as the distance from the object increases, the measurement reliability decreases. A vision-based navigation system can detect objects from continuous images obtained through cameras, calculate navigation information, and recognize traffic lights, curbs, pedestrians, and so on. However, there is a problem that the detection performance is deteriorated in a situation where the weather is blurry or the object is covered by another vehicle. UWB (Ultra Wide Band) signal provides more precise distance information with improved resolution compared with a general radio wave, but it is difficult to measure the distance of the object at a relatively long distance because the measurable distance is short. Therefore, single measurement based positioning has limitation.

Current industry trends are changing rapidly due to integration and mutual fusion of each technology, and change of trend requires a deep understanding of specific fields and mutual complement/convergence of technologies. Therefore, it is required to develop a sensor fusion technique that obtains high accuracy and reliability by effectively using various sensor information, rather than simply using one technique in position estimation.

In this thesis, we have investigated the measurement fusion based localization/tracking algorithms for a moving target in wireless networks. First, the study about the problem of the 3D target tracking with Bayesian filtering method when multiple measurements are used was carried out. The particle filter-based 3D target tracking method with measurement fusion of time difference of arrival (TDoA), frequency difference of arrival (FDoA), and angle of arrival (AoA) was proposed. And the performance of the proposed method was analyzed by the MATLAB simulations. From the simulation results, it was confirmed that the performance enhancement in the localization accuracy compared to the single measurement based tracking and extended Kalman filter. Second, the study of the sensor-fusion based pedestrian tracking with Bayesian filtering method was carried out. Pedestrian dead reckoning (PDR), and UWB was fusioned by the Kalman filter-based tracking method. The performance of the proposed method was analyzed by experiments on the first floor of the Advanced Materials & Chemical Engineering Building, Hanyang University. The detailed explanations are in the thesis.