The invention belongs to the technical fields of robot active visual perception, language interaction, radar obstacle avoidance and deep learning, and in particular relates to a robot human-computer interaction method. This method captures the RGB image and depth map of the environment, detects obstacle information to obtain a laser radar array, normalizes the acquired data, constructs a problem encoding network in human-computer interaction to encode the problem; constructs an image feature extraction network, Extract RGB image and depth image information into a feature matrix, splice lidar data, question code and feature matrix to get feature fusion matrix; use convolutional network to obtain data fusion matrix as the data fusion matrix of surrounding environment; train a cyclic neural network As a navigator, the network takes the data fusion matrix as input, outputs the navigation result, and controls the robot's movement direction. The method realizes the robot's self-navigation, self-exploration, human-computer interaction and other functions, and improves the intelligence of the robot.