There are three kinds of conflicts in a traffic flow at the traffic circle because of track changing. We establish the driver model by regarding each vehicle as an intelligent agent which constitutes the multi-agent system by integrating the cellular automation model at the traffic circle. Besides, we put forward the problem how to eliminate the conflicts between the vehicle agents by priority strategy and dynamic negotiation mechanism. Take the traffic flow at the classic traffic circle of the suburban area of a small-medium city as the sample data to simulate and verify the accuracy of the model, and study the influences of some factors on the flow of motor vehicle, such as the proportion of different driver type and track changing probability, etc. The experimental result shows that the model constructed is consistent with the actual situation. Meanwhile, it is found that the driver characteristics have obvious influences on the flow, and the track changing probability follows, which does not have obvious influences on the fluctuation in the flow.