Relying on national key projects, the members of the research team combined the needs of the new generation of artificial intelligence technology and public service robots, and integrated and applied the research results obtained in each project to the experimental scene of the new generation of intelligent shopping guide unmanned convenience stores. The entire intelligent shopping guide robot system integrates advanced technologies such as human target re-identification based on human spatial structure information triple convolutional neural network, and interactive object behavior recognition and understanding for the fusion of vision and depth perception. At present, the intelligent shopping guide robot system can realize the function of five people shopping independently, and can automatically carry out commodity statistics, identification, billing and customer checkout.