Service-oriented robot for realizing hierarchical semantic map by convolutional neural network

The invention relates to a service robot for realizing a hierarchical semantic map by using a convolutional neural network, which comprises a voice unit for inputting a voice instruction; the visual unit is used for recording environment images; the graph database comprises a grid map for recording path nodes, a topological map or a mixed map combining the grid map and the topological map; the semantic database comprises a semantic map generated according to the graphic database; the computing unit is connected with the voice unit, the visual unit, the graphic database and the semantic database, discriminates the semantics of the voice instruction through a convolutional neural network, compares the current environment image with the characteristics of the path node, and guides the robot to move to a destination; and the control unit is connected with the calculation unit and moves the robot according to the calculation result. 本发明是一种以卷积式神经网络实现阶层式语意地图的服务型机器人，该机器人包括有：语音单元，用来输入语音指令；视觉单元，用来记录环境图像；图形数据库，包括有记录路径节点