Physic Grounded Vision Foundation Models for Human Computer Interaction in Embodied Environments | Synapse