Scheduled Auxiliary Control (SAC-X) is a learning paradigm for AI-powered robots at DeepMind. This type of paradigm rewards AI robots for completion of any given command.[1]
