Itisdifficulttoobtaintheroadsurfaceadhesioncoefficientinrealtimeandaccu-rately during the running of the vehicle.Aiming at this problem， a four-hub drive co-simula-tion electric vehicle platform combined with a three-degree-of-freedom dynamic model and aDug off modified tire force model was built， and a double cuba ture Kalman observer withstrong timeliness and robustness for road adhesion coefficient was designed.The observeralgorithm uses the singular value decomposition to solve the error covariance matrix， whichcan update the vehicle speed information in realtime， and correlate the vehicle driving stateobserver information with the adhesion coefficient observer cycle to form a closed-loop feed-back correction to realize real-time estimation of the road surface adhesion coefficient.Thelow-attachment road surface and bi sectional road surface condition was setted in the four-wheel hub drive co-simulation electric vehicle platform to verify double cuba ture Kalman fil-ter algorithm and compared with the cuba ture Kalman filter observer data.The results showthat the algorithm has faster response speed， higher accuracy and better real-time perform-ance in estimating road adhesion coefficient.