In the field of channel modeling, a cluster is usually used to describe the spatial characteristics of the dense components of the channel, and the time-varying trajectory is used to describe the evolution and birth and death of the multipaths. However, these two parts are usually studied separately. In recent years, researchers have tried to consider both temporal domain and spatial domain characteristics, but the main methods are still limited to two-stage frameworks. In this paper, a one-stage spatio-temporal cluster tracking method is proposed to achieve the generalized clustering in the space-time domain. Out proposal takes into account the temporal and space tolerances as well as adaptive selection of multipath components distance thresholds. The proposed method is verified by means of an innovative spatio-temporal cluster evaluation index applied to measured data, exhibiting better performance than currently available approaches. The proposed method poses the basis for the unified development of spatially-consistent channel models for future integrated communication and sensing standards.