A human-computer interaction system comprises the following components: a projection device which displays a virtual desktop through projection according to the projection information; an infrared emitter which emits the infrared light that is parallel with the virtual desktop and covers the virtual desktop; an image acquisition device which receives the infrared light that is reflected from an interaction gesture; a robot which responses the interaction gesture in the virtual desktop; and a processor which is respectively connected with the projection device, the infrared emitter, the image acquisition device and the robot electrically for transmitting the projection information and generating response instruction according to the infrared light from the interaction gesture. The virtual desktop is established in the human-computer interaction system, and the interaction between the user and the robot is realized on the virtual desktop. The required operation can be finished through contacting with the virtual desktop for the robot. Therefore, the structure complexity of the robot structure can be effectively reduced. The intelligence, speed and efficiency of the interaction process are effectively improved.