Target tracking, as one of the core research fields of computer vision, has a huge potential of application value in the fields of UAV countermeasures and autonomous driving. Aiming at the lack of robustness of traditional single-modal trackers in extreme scenarios such as low illumination, target occlusion, and complex background interference, academics are increasingly focusing on the solution of visible and thermal infrared (RGBT) dual-modal fusion, which significantly improves the tracking performance in complex scenarios through the mechanism of cross-modal feature complementarity. In order to solve the problem that traditional target tracking algorithms cannot effectively enhance and extract the features of the two modalities and fuse them, we propose a multi-modal target tracking algorithm that employs interactive fusion with spatial a feature extraction. It is able to realize two modal features for bidirectional and multi-stage information fusion. Experiments of our method on Lasher and GTOT datasets achieve better performance and robustness.