Researchers introduce T-Rex, a framework that unifies vision, language, and tactile sensing so robots can respond to physical contact in real time rather than relying on vision alone