A scale adaptive analysis using drift of mean position as the metric and universal thresholding as the statistical tool to separate motor-driven active motion and passive Brownian diffusion in intracellular active transport is developed. It is validated by comparison with three existing methods, as well as by manual inspection. The method makes no assumption about the nature of the transport and can automatically adapt to heterogeneities among trajectories. The fully automated method allowed the analysis of a large intracellular active transport dataset with varying cell lines, cargo types, microtubule and cytoskeleton treatments to investigate the regulatory mechanisms of active transport.