Physically Grounded Vision-Language Models for Robotic Manipulation | IEEE Conference Publication | IEEE Xplore