Action with Visual Primitives

Weilong Guo, Yuchen Wang, Renping Zhou +5

Vision-Language-Action (VLA) models have emerged as a promising paradigm for generalist robotic manipulation. A common design in current architectures maps language instructions an…