This paper concerns the issue of enhancing the robustness in radio tomographic imaging (RTI) with sparse Bayesian learning (SBL), which aims at addressing the localization performance deficiency due to uninformative radio frequency (RF) data. Spatiotemporal RTI is developed to keep data informative and reliable for sparse signal recovery in localization issues. In addition, two robust sparse Bayesian learning algorithms are developed to handle with the low signal-to-noise-ratio (SNR) with heterogeneous noise. The localization results highlight advantages of applying proposed robust sparse Bayesian learning algorithms in addressing missing estimations and outlier errors, and finally improving indoor target DFL performance.