Multi-Sensor Fusion on Automated Guided Vehicle

In the era of big data, Due to the insufficiencies of single-sensor systems in achieving reliable target recognition, semantic segmentation, and 3D point cloud reconstruction, multi-sensor fusion systems have risen as a significant trend in data acquisition. This paper proposes a multi-sensor fusion system mounted on an automated guided vehicle (AGV) platform, consisting of a multispectral camera, millimeter wave cascaded radar, lidar, and RGB camera. The system uses a linear motor stage to facilitate the formation of synthetic apertures for millimeter-wave cascaded radars. It uses a multispectral camera to obtain target spectral data, an RGB camera to obtain target optical images, and a lidar to obtain target spatial information. The system has low power consumption to collect various sensor data easily. Our platform has been utilized extensively for multi-sensor data acquisition, yielding high-resolution SAR images, high-quality optical and multispectral images, and detailed 3D point cloud data, which have enabled accurate identification of target materials, demonstrating the system's robust performance.