Robots in a multi-robot system cannot communicate with each other due to various reasons. The robots can only maximize their own individual performance without regard for the overall system performance. To guarantee high system revenue even when the robots are self-interested, an algorithm called the intermediary recruitment algorithm (IRA) is proposed. It imitates the operating mechanism of an intermediary recruitment market to allocate the tasks to the robots. The simulation results analyse the influence of the robots' social value orientation (SVO) on the system revenue,where SVO is used to measure the degree of self-interest, and the effectiveness of the IRA is also verified.