Vision-language-action

Model-ready action trajectories for VLA systems

Structure VLA datasets as action-conditioned episodes connecting language goals, visual observations, grounded actions, state changes, and outcomes.

Synthetic schema sampleObservation → action → next state
01
00.000sObservergb.player_view
02
00.050sActmove.forward + camera.delta
03
00.100sTransitionpose.x +0.84 m
04
00.150sOutcomeprogress 0.31 → 0.36

The values are synthetic and illustrate the public specification. They are not measurements from a commercial dataset.

Direct answer

A VLA policy predicts what action to take from visual and language context. A world model predicts what follows from an action. A model-ready VLA trajectory preserves both sides: instruction and observation, action target, resulting state, and outcome.

A VLA chooses actions; a world model predicts consequences

Language expresses intent and visual observations describe the current scene. The action target still needs a declared interface: pose delta, joint target, end-effector command, gripper state, discrete skill, or another embodiment-specific control.

Recording the state and outcome after execution makes the same episode useful for policy learning, world-model training, rollout evaluation, and failure analysis. One caption over an entire session cannot provide that causal resolution.

  • Instruction, goal, and optional subgoal text
  • Camera and sensor observations with calibration
  • Action tokens plus physical units and control rate
  • Embodiment, task, success, failure, and intervention metadata

Action tokenization must remain reversible

A VLA may predict discretized tokens, continuous action chunks, or higher-level skills. The release should preserve the original command beside any quantized or normalized representation.

Token vocabulary, bin boundaries, chunk horizon, control frequency, coordinate frame, and clipping behavior belong in the schema. Otherwise identical token IDs can imply different motion across robots or datasets.

Cross-embodiment mixtures need explicit normalization

Large VLA programs often mix tasks and robot types. Positive transfer is possible, but morphology, sensors, action spaces, control rates, and success definitions do not become equivalent because they share one loader.

Keep source-dataset identity and embodiment metadata available during training and evaluation. Hold out robots, scenes, objects, and instruction families to measure generalization rather than memorization.

Questions this page should settle

What is a vision-language-action model?

A VLA model uses visual observations and language context to predict actions for an embodied agent, commonly a robot.

What fields belong in a VLA dataset?

Include language goals, time-aligned observations, original and model-ready actions, embodiment and calibration metadata, task state, interventions, and outcomes.

How do VLA models and world models relate?

A VLA model usually selects actions from visual and language context. A world model predicts possible consequences of actions. They can share the same trajectories when observations, actions, resulting state, and outcomes are aligned.

Bring us the training target.

We will map it to the observations, actions, state, outcomes, rights, and delivery evidence the dataset must contain.

Request access