Robotics
VLA+
2026Experimental
Key
innovation
Extension of the VLA paradigm with tactile sensing and online learning from human corrective feedback, so the robot keeps adapting after deployment rather than only during training.
Category
Robotics
Abstraction level
Paradigm
Operation level
Robot controlSystemModel
Implementation
Implementation pitfalls
Brak web-skalowych korpusów dotykowychHigh
Dryf polityki przy online learningMedium
Hardware requirements
Primary
Sources
AI 101: VLA Models Explained: Architecture, Types & the Leap to VLA+
articleTuring Post
Gemini Robotics: Bringing AI into the Physical World
PaperarXiv
π0: A Vision-Language-Action Flow Model for General Robot Control
PaperarXiv
Helix: A Vision-Language-Action Model for Generalist Humanoid Control
BlogFigure