This paper deals with an emotion-based approach to the task allocation of multi-robots, in which one robot requests the assistance of another robot in order to complete a task. A stochastic model of emotion is adapted to achieve a method for well organization of team work. With the emotional capability, each robot can distinguish the changed environment and can react with it with adaptability and also can understand colleague robot's state. The proposed computational architecture to model emotion is based on Markov modeling theory. In our approach, emotional state plays a role in the task allocation among the robots of a team in a productive and efficient manner.