Multisensor fusion-based localization system and experimental design

With the rapid development of intelligent robotics and autonomous driving technologies, accurate and robust vehicle localization has become one of the fundamental prerequisites for environmental perception, motion planning, and autonomous decision-making. However, in practical applications, localization systems operating in complex environments are frequently affected by abnormal measurements, multipath interferences, and sensor uncertainties, which considerably degrade localization accuracy and system reliability. In particular, Global Navigation Satellite System (GNSS)–based positioning methods are highly susceptible to environmental disturbances in urban canyons, indoor–outdoor transition zones, and occluded scenarios, resulting in severe localization drift or temporary signal loss. Therefore, multisensor fusion has become an important research direction for improving the localization accuracy, robustness, and environmental adaptability in intelligent unmanned systems. To enhance students’ understanding of vehicle localization technologies and the theoretical foundations of fusion-based estimation, a multisensor-based localization experimental platform was designed as part of robotics-related courses, focusing on a collaborative localization system integrating multiple heterogeneous sensors.

This experiment introduces a fusion localization framework for an intelligent unmanned vehicle that integrates ultra-wideband (UWB), GNSS, and inertial measurement unit (IMU) sensors. This experiment aims to bridge the gap between theoretical learning and engineering implementation by enabling students to understand the practical workflow of sensor fusion localization, including data acquisition, state estimation, information fusion, and robustness analysis under disturbed environments. Specifically, GNSS measurements provide global positioning information, UWB measurements compensate for localization degradation under signal occlusion and interference, and IMU measurements enable continuous motion-state propagation and short-term state estimation. By exploiting the complementary characteristics of these heterogeneous sensor measurements, the proposed framework effectively integrates multidimensional information to improve localization accuracy, reliability, and environmental adaptability. Through the collaborative utilization of multisource sensor data, the system leverages the complementary properties among different measurements, thereby improving localization accuracy, estimation consistency, and robustness against abnormal observations and environmental disturbances. The experiment also enables students to gain a deeper understanding of the practical implementation of fusion filtering algorithms, including state prediction, measurement update, and uncertainty propagation in unmanned vehicle localization tasks. In addition, students can intuitively analyze the influence of sensor noise, measurement uncertainties, and environmental interference on localization performance through experimental observations and comparative analyses.

Experimental results demonstrate that the proposed multisensor fusion localization scheme can effectively suppress the influence of abnormal measurements and compensate for localization degradation caused by environmental interference. Compared with single-sensor localization methods, the proposed fusion framework exhibits superior localization accuracy, stronger robustness, and improved stability under complex environments. The designed experimental platform not only provides an effective educational tool for robotics and intelligent vehicle courses but also offers practical guidance for understanding the engineering applications of multisensor fusion localization in autonomous systems.