The traditional visual SLAM method has poor tracking performance in scenes with fewer textures and lacks robustness for situations such as severe shaking or fast turning. In response to the above issues, this article proposes the stereo vision SLAM algorithm based on adaptive fusion of point and line features. We improve the classic ORB-SLAM2 algorithm by fusing different types of line feature projection errors and designing flexible adaptive fusion factors. The weight of line features is dynamically adjusted based on the number of traceable feature points, achieving a more effective fusion of point and line features. This not only improves the tracking performance of the system but also improves the accuracy of camera trajectory positioning. The experimental results show that the algorithm proposed in this paper is effective in Integrating line features into the ORB-SLAM2 system reducing average positioning error by 16.9%.