This paper proposes a human hand detection method for gesture recognition used for communication of a robot with a human. A human hand motion can be recognized as a different meaning according to a situation in the communication between the human and robot. First, we propose a steady-state genetic algorithm for extracting a time series of human hand position, and next, a gestures recognition method of a robot composed of a spiking neural network and self-organizing map. Finally, we discuss the effectiveness of the proposed method through several experimental results.