The accurate information of road friction coefficient allows the control algorithm in vehicle activity system to adapt to external driving conditions. This paper developed a slip-slope friction coefficient estimation method based on Unscented Particle Filter. A 7-DOF non-linear vehicle dynamic model was established. The normal force of tire was approximately calculated from the vehicle dynamic model; the slip and longitudinal force of tire were estimated through a combination of tire mechanical model and UPF(Unscented Particle Filter) method. Finally slip-slopes of different adhesion level roads was obtained. Through virtual test environment in ADAMS/Car, the estimation method proposed was verified to be effective and reliable under various road condition. From the method the relationships between the slip-slope and road friction coefficient are achieved.