Enhancing Loop Closure Detection in Intelligent Transportation and Robotics through Semantic and IoT-Integrated Frameworks

Abstract

Accurate loop closure detection is a critical component of Simultaneous Localization and Mapping (SLAM) systems, directly impacting the reliability and consistency of autonomous navigation in intelligent transportation and robotic applications. This study proposes a semantic and IoT-integrated framework that enhances loop closure detection by combining deep feature representations with contextual scene understanding and multi-sensor data fusion. Semantic segmentation and object recognition modules provide high-level contextual information, enabling the system to distinguish between dynamic and static elements in the environment. Simultaneously, IoT-enabled sensors contribute additional spatial and environmental data, supporting robust feature matching and trajectory optimization. Experimental evaluations demonstrate that the proposed framework significantly improves loop closure precision and recall, reduces trajectory drift, and maintains global map consistency across diverse outdoor and urban scenarios. The results validate the potential of integrating semantic perception and IoT connectivity to advance autonomous navigation and intelligent transportation systems.