UKF SLAM-Enhanced Perception and Navigation for Effective Area Coverage in Autonomous Driving

Reliability in decision-making, path planning, and motion control is significantly influenced by localization accuracy and integrity, which poses a fundamental challenge for au- tonomous ground vehicles. To address errors induced by external factors in GPS-based pose estimation, this study introduces a method that combines GNSS/IMU data with landmark-based environment mapping constructed from laser scan data. The integration of measurements from multiple sensors and the enhancement of accuracy and dependability in the estimation process were achieved using the Unscented Kalman Filter (UKF) for multi-sensor data fusion. Comprehensive experiments conducted across various datasets and under diverse conditions demonstrate that our proposed technique achieves high accuracy and robustness in state estimation. Its precision remains consistent with ground truth even in the presence of noise, underscoring its reliability. Moreover, the incorporation of a scan-matching approach significantly improves map estimation precision, contributing to enhanced overall pose estimation and localization accuracy. This capability underscores the effectiveness of our approach in navigating and mapping dynamic and noisy environments, making it a valuable tool for a variety of robotics and autonomous system applications.