Research on Efficient Localization Methods for Unmanned Vehicles in Complex Environments

Accurate localization of unmanned vehicle (AGV) in complex construction environments presents multiple challenges. This paper proposes a method that combines multi-sensor preliminary positioning with fixed-camera visual precision positioning to enhance localization accuracy and system robustness. Initially, GNSS and IMU are utilized for preliminary positioning of the vehicle, determining its approximate location. Subsequently, fixed cameras are employed to achieve high-precision visual positioning, acquiring accurate positional information. Experimental results demonstrate that fixed-camera-based visual positioning significantly outperforms traditional methods in complex environments, with the visual trajectory closely aligning with the actual trajectory. This validates the method's practicality and reliability in meeting high-precision localization requirements.