<p>Driving safety field (DSF) model is a virtual force method, which has great potential in autonomous driving risk assessment and decision making. However, the large number of parameters of the DSF model brings great difficulties to the parameter calibration, thereby constraining further application of the DSF model. In this paper, a novel parameter calibration method of the DSF model based on the traffic evolution scenarios is proposed and applied on path planning. Firstly, based on the distribution law of traffic behavior, the longitudinal and horizontal initial scenarios are established, then Monte Carlo method is used to simulate and generate 1000 traffic evolution scenarios reflecting the uncertain behaviors of other traffic participants. Secondly, the risk value is calculated based on the severity and possibility of collision in these traffic evolution scenarios, and all the parameters of the DSF model are calibrated based on the calculated risk value. Finally, a path planning algorithm based on the calibrated DSF model is established and compared with the distance-based path planning algorithm. The simulation and hardware-in-loop experimental results show that the path planning algorithm based on the calibrated DSF model is superior to the distance-based path planning algorithm on planning speed, path safety and comfort. The parameter calibration method of the DSF model based on the traffic evolution scenarios simplifies the calibration process under the condition of ensuring the calibration accuracy and verifies the accuracy and feasibility of the practical application of the DSF model.</p>