Efficient Industrial AGV Pose Estimation System in Both Indoor and Outdoor Scenario

With recent advancements in mobile robotics and autonomous driving, multi-sensor fusion technologies have produced significant improvements for precise pose estimation system of mobile devices. However, due to the high computational requirements of multi-sensor fusion algorithms, there are challenges in integrating multiple sensors into automated guided vehicles (AGVs) because computing resources are limited these platforms. Balancing high localization accuracy with low computational load has become a major issue in AGV development. To address this problem, this paper introduces a novel multi-sensor fusion framework employing the error state Kalman filter. The framework combines the inertial measurement unit, Light Detection and Ranging, and ultra-wideband sensor data. Special sensor placement and algorithm optimization in the system are used to achieve reliable 6-DOF pose estimation. The experimental results show that the proposed pose estimation system is more efficient and accurate compared to existing multi-sensor fusion methods.