Optimized Sensor Fusion for Local Positioning System of Autonomous Vehicles
Cao Tuan Kiet, Nguyen The Bang, Le Huu Phuoc, Phan Xuan Anh Tu, Nguyễn Quang Huy, Nguyen Dang Nhat Minh, Long Dang Tran, Quang T.D. Pham · 2024
The relentless advancement in autonomous vehicle technology necessitates sophisticated local positioning systems (LPS) to ensure precise navigation and safety without GPS. This paper introduces an optimized sensor fusion model designed to enhance the accuracy and reliability of the LPS in autonomous vehicles. By integrating data from a variety of sensors, including IMU, Encoder and Camera. Our proposed technique employs advanced algorithms to mitigate the inherent limitations of each sensor type, such as signal degradation in adverse weather conditions and LPS multipath errors in urban environments. We introduce a novel optimization framework that leverages techniques including data fusion, Kalman filter and zero velocity updater to dynamically weight sensor inputs based on their current reliability and accuracy. This approach significantly reduces the system's susceptibility to individual sensor failures or inaccuracies, thereby improving overall positioning precision. Extensive simulations and real-world testing scenarios demonstrate that our optimized sensor fusion model achieves superior performance in terms of accuracy, robustness, and computational efficiency compared to traditional sensor fusion techniques. Our findings suggest that this optimized sensor fusion approach holds significant promise for enhancing the safety and efficiency of autonomous vehicles by providing a more reliable LPS under a wide range of operating conditions.