In　the　control　of　self-driving　vehicles,　PID　controllers　are　widely　used,　because　of　their　simple　structure　and　good　stability.　However,　in　a　complex　self-driving　environment,　different　scenarios　have　different　requirements　for　vehicle　control　accuracy.　At　this　time,　using　a　set　of　PID　parameters　may　not　meet　the　requirements.　In　order　to　solve　this　problem,　we　proposed　an　adaptive　fuzzy　PID　controller.　Specifically,　firstly,　we　added　a　rule　partition　table　on　the　basis　of　fuzzy　PID.　Secondly,　vehicle　sensors　recognize　the　scene.　Thirdly,　the　fuzzy　rules　are　determined　by　the　rule　partition　table,　and　the　PID　parameters　are　adjusted　in　different　scenarios　in　real　time.　Finally,　we　Use　this　method　to　test　in　Panosim　simulation　environment,　the　control　accuracy　is　better　than　PID　control　in　right-angle　corners,　overtaking　and　other　scenes.　©　2021　IEEE.