A Humanlike Lane Change Decision Strategy for Mixed Traffics with Multi-objects

In mixed traffics, humans are easily confused by the non-humanlike lane change of autonomous vehicles. This may even bring traffic accidents. According to the naturalistic driving analysis results that there exist significant influences of multi-obstacles on lane change behavior, a humanlike lane change decision strategy for scenarios with multi-obstacles is presented. The driver tolerance force is proposed to establish the relationship between the longitudinal social force and lateral lane change behavior by using the driver's speed requirement. Then the visual attenuation coefficient is designed to reflect the influence of social forces generated by different traffic participants. Compared with other methods, the statistical results based on the naturalistic driving data indicate that the proposed strategy has a higher decision accuracy of target lane and start point.