Aiming at the problem of deterioration of vehicle longitudinal dynamic control performance caused by variable road surfaces in off-road environments, an adaptive control strategy integrating Extended Kalman Filter(EKF) and Random Forest (RF) is proposed. This strategy adopts a multimodal perception framework. Firstly, it uses the RF algorithm to identify the road surface type in real time, and then takes the result as the prior knowledge of the EKF estimator to achieve precise estimation of the peak adhesion coefficient of the road surface. Verification shows that this strategy can effectively adapt to different road surfaces. It can significantly suppress excessive wheel slippage on both high and low adhesion roads, stabilize the slip ratio within the optimal range, which is conducive to improving the vehicle's acceleration and braking performance, shortening the braking distance, and enhancing the safety of off-road driving.