Robust PCA is as the name implies a more robust version of PCA. Extremely useful for large but sparse datasets (many biological areas e.g. ecology, metabolomics, transcriptomics; and computer science e.g. visual recognition).
Provides a PCA which is less susceptible to outliers.