What is Vision-language-action model (VLA)?

Vision-language-action model (VLA)

Vision-language-action model (VLA): A model that maps camera input and an instruction directly to robot actions, the current approach to general-purpose manipulation.

Related terms

Vision-language model (VLM)
Large language model (LLM)
Program-aided language model (PAL)
Small language model (SLM)
Autoregressive model
Cascade (model cascade)
Computer vision
Consistency model

Browse the full glossary →