<p indent=0mm>Because of the direct contact between the tire and pavement, the skid resistance of asphalt pavement is a basic factor affecting the braking safety of autonomous vehicles. The road skid resistance mainly depends on the texture characteristics of the road surface. Considering the pavement friction characteristics, building a braking model of autonomous vehicles is an effective method to improve the braking safety of autonomous vehicles. In order to develop the safety braking method for autonomous vehicles based on the road surface friction characteristic, an automatic close-range photogrammetry system (ACRP system) was proposed and built based on the circle arranged three cameras close-range photogrammetry (CRP) technology to obtain the asphalt pavement surface texture accurately in real time. Automatic image acquisition and 3D reconstruction were achieved by the ACPR system. Sand patch testing method and laser scanning method (ZGScan) were applied to collect the on-site asphalt pavement texture comparing with the results of ACRP system. It shows the texture data obtained by ACPR system have relatively high accuracy and efficiency with recognition accuracy close to 0.02 mm. Then, the peak adhesion coefficients under different road conditions were calculated by considering the skid resistance contribution of the road surface texture. Based on CarSim/Simulink joint simulation, a braking model for autonomous vehicles was established. The characteristics of the autonomous vehicles during straight normal braking, emergency braking and driving on curved sections were analyzed. Then, the safety braking distance under different conditions was put forward. The research shows that the variation trends of friction coefficient curves under different pavement conditions (dry and wet pavement) are basically the same, which both decrease significantly with the increase of relative slipping ratio. When the speed exceeds <sc>40 km/h,</sc> the curve tends to be gentle, indicating that the actual contact area of the tire-road comes to a stable status when the speed is relatively high. Based on the tire hydroplaning model built by ABAQUS, the peak adhesion coefficient curves for different pavement conditions were obtained. It can be seen that the peak adhesion coefficient of the road surface is convex parabolic distribution at different driving speeds. Moreover, the peak adhesion coefficient gradually decreases as the speed increases. During normal braking, the vehicle should keep relatively low braking deceleration considering passenger comfort. As for emergency braking, autonomous vehicle should be equipped with professional short-range radar, long-range radar and high-definition camera to detect the environment surrounding the vehicle. On rainy days, the braking distance is averagely increased by about 45% compared to sunny days. When the braking distance between vehicles is large enough, it is recommended to brake with a braking force of <sc>4−6 MPa.</sc> Autonomous vehicles should keep a certain safe distance, which requires 1.1−1.2 times of the simulated average value of safety braking distance during the driving process on dry or wet pavement. As the radius of curved sections increases, the acceleration interference values under different road conditions basically reduce. When the radius is greater than or equal to<sc>100 m,</sc> the decrease rate of the acceleration interference value is more significant. In order to improve the braking comfort of the vehicles, the radius of the curved road should be controlled above <sc>100 m.</sc> In this paper, the safety strategies for autonomous vehicles considering pavement surface texture characteristics can provide references for the theoretical design of braking system and braking safety evaluation for autonomous vehicles.