This paper discusses a communication robot system based on image processing and voice recognition. We have developed the communication robot system Hakuen which consists of a multimedia robot with stereo cameras, a wheeled mobile robot and a PC with a microphone. What makes our robot unique is that the robot interacts with people in the same way the human beings do. The robot, for example, approaches and holds its hand out to someone based on the defined voice commands. The robot detects a person’s face based on the pixel values of the flesh tint in the color image. Since the system must calculate the distance between the robot and the person rapidly, we use this disparity. Experimental results clarified the effectiveness of our system.