Ant Group Proposes LingBot-VA 2.0 Video-Action Model
jiqizhixin · x · 2026-07-18
Ant Group introduced LingBot-VA 2.0, a video-action foundation model built from scratch specifically for real-world control.
Unlike standard video models designed for digital content, this model introduces four breakthroughs:
- Semantic-Action Tokenizer: Links the robot's visual input with the actions it should execute.
- Causal Training Method: Prevents the model from forgetting previously executed actions.
- Sparse MoE Backbone: Increases model capacity without compromising operational speed.
- Asynchronous Inference Scheme: Predicts future steps in real-time while executing the current action.
Results: Its few-shot generalization capabilities in complex manipulation tasks surpass previous video-action models.
More from Embodied
- ECCV26 Oral: Flow Matching Enables Single-Stage Multi-View Point Cloud Registration — ducha_aiki · 2026-09-11
- Polish developers build iPhone app that detects nearby Meta smart glasses — Low-Honeydew6483 · 2026-09-11
- Ant's Afu health AI hits 150M users, unveils AI+hardware health alliance at Bund Summit — APPSO · 2026-09-11
- Johns Hopkins Launches Full-Stack Hands-on Robot Learning Class with SO-101 Arm Kits — _krishna_murthy · 2026-09-11
- SyncWorld: In-Context Robot World Model Simulates Unseen Views and Embodiments Zero-Shot — ChongZzZhang · 2026-09-11
- A 3D Pose Dataset for Dogs Released — ducha_aiki · 2026-09-11