Promoting autonomous vehicles (AVs) requires ensuring that vehicles can safely and smoothly pass through various ice-covered complex road conditions. Theoretical modeling and simulation tests on the curve braking control of AVs based on the vehicle-ice-road coupling mechanism are conducted. This study presents an ice-snow two-degree-of-freedom model (I-TDOFM) and a braking control algorithm designed to address the low adhesion coefficient of ice-snow roads. Three examples of ice and snow turning road conditions are simulated on the CarSim-Simulink joint simulation platform, and the research results reveal the variation patterns of the yaw rate (YR) and sideslip angle (SA) during the turning process of AV. The simulation shows that under various steering conditions, speed significantly impacts the AV’s driving performance on icy and snowy roads; AVs require a larger SA to complete steering maneuvers at low speeds, and the YR increases with increasing driving speed. The proposed I-TDOFM and braking control algorithm can constrain AV to travel along the target path and maintain good steering stability on ice-snow roads. This study provides theoretical and simulation experimental references for the optimization control of AVs on ice/snow roads in the future.