Models & learning
World-action model
A world-action model is an emerging, non-standardised name for a robot model that couples prediction of future world states with generation of executable actions, often by adapting a video or world-model backbone as a policy. It differs from an action-conditioned world model that only predicts consequences of supplied actions, and from Barrett’s WAM, the Whole Arm Manipulator.
Also known as: WAM, world action model
Updated
An emerging model category
World-action model is recent, non-standardised terminology rather than a settled architecture. The DreamZero paper uses it for a model built on a video diffusion backbone that jointly predicts future video and robot actions. NVIDIA’s technical account places such models between visuomotor policies and world models: they use a predictive visual or latent backbone while also emitting actions.
Other papers may use names such as video-action model for a similar joint objective. The label alone therefore does not specify the action representation, planning method or control frequency.
Not every action-conditioned model is a WAM
An action-conditioned world model receives a proposed action or action sequence and predicts its consequences. It can support planning without generating the action itself. In the narrower world-action usage, the model is also a policy: it proposes executable actions, often alongside the future states associated with them.
That distinction matters when assessing training data. A predictive model needs observations paired with the actions that caused subsequent states. A joint policy additionally needs action labels expressed in the target robot’s control space. Video-only data may help pretrain visual dynamics, but it does not directly provide humanoid joint or torque commands.
Not the Barrett WAM arm
WAM has a separate, older robotics meaning. Barrett’s WAM Arm is the Whole Arm Manipulator, a backdrivable physical robot arm offered in four- and seven-degree-of-freedom configurations. It is hardware, not a learned world-action model. Writing the full term on first use avoids confusion in papers, dataset records and equipment descriptions.
Sources
Related terms
Models & learning
World model
A world model is a learned predictive model of how an environment changes, often conditioned on a robot’s current state and candidate actions. It may forecast pixels, latent states, rewards or other task-relevant quantities. A policy or planner can use those predictions to compare possible futures before commanding the physical robot.
Models & learning
Policy
A policy is the decision rule that maps a robot’s current observations or estimated state, and sometimes a task instruction, to an action or probability distribution over actions. It can be hand-designed or learned from demonstrations, rewards or both. In humanoid robotics, its outputs may be joint targets, torques, end-effector changes or higher-level skills.
Models & learning
Vision-language-action model
A vision-language-action model is a robot policy that conditions on visual observations and natural-language instructions to predict executable actions. It adapts a pretrained vision-language model, or a similar multimodal backbone, using robot trajectories so that semantic knowledge is connected to an embodiment’s action space. The action representation may use discrete tokens or continuous values.
Data & collection
Trajectory
A trajectory is a time-ordered sequence of states or observations, actions and, where applicable, rewards generated as an agent or robot evolves. A complete episode or policy rollout often yields a trajectory, but the terms are not universally identical: trajectories may be partial, while episodes have dataset- or environment-defined boundaries.