To solve the problem of boundary oscillation and difficulty in realizing the fission of the classic algorithm for multi-agent system motion control based on the principle of Separation, Alignment and Cohesion (SAC), a motion control algorithm based on attention-following mechanism and Focus-Improved Artificial Physics (F-IAP) is proposed.The alignment area and the aggregation area of the classical SAC algorithm are combined in this algorithm, and the gravitational coefficient is adjusted according to the distance and the speed, so as to change the effects of the mimic physical force.The problem of boundary oscillation of the SAC algorithm is solved, and the cohesion, alignment and obstacle avoidance of the multi-agent system are realized.According to the characteristics of biological attention in neuroscience, the agent that directly perceives environmental stimulation is selected as the follow-up object.By using the IAP algorithm, the fission of the multi-agent system is realized in multi-objective environment.The feasibility of the algorithm and the superiority of its performance are verified in simulation.