趋势解读：World Action Models give robots the ability to，解读最新研究结论

A recent review paper introduces a systematic framework for "World Action Models" (WAMs), a model class for robotics that enables AI systems to be trained using unlabeled everyday videos. Unlike conventional approaches, WAMs don't just learn which action should follow a given camera image. They also simulate how the environment will change as a result of that action, effectively building an internal model of the physical world. The roughly one hundred papers analyzed in the review fall into two main architectural categories. One line of work first generates a predicted future video and then derives control commands from it, while the other processes visual input and actions simultaneously in parallel. Today's robotics AI has a basic weakness: models learn to map camera images directly to movements. But they don't understand how the world actually changes as a result of their actions. A new survey paper from Fudan University, the Shanghai Innovation Institute, and the National University of Singapore is the first to systematically catalog a class of models designed to close that gap: World Action Models. Existing vision-language-action models mostly learn direct mappings from observations to matching actions. World Action Models go further. They also model how the environment will likely change, then couple that prediction to action generation. Ad The payoff is practical, the authors say. A model that simulates the consequences of a movement before executing it generalizes better to unfamiliar objects and settings. More importantly, it can learn from video footage where no robot actions are labeled at all—everyday first-person videos, for example. That kind of data was nearly useless for traditional robotics AI. Ad DEC_D_Incontent-1 Pure video generators can produce plausible future frames, but they aren't tied to control signals. A research team at Peking University recently drew exactly that distinction in its unified definition of world models . World Action Models meet both conditions at once. The researchers sort about a hundred papers into two architectural lines. The first, Cascaded WAMs, works in two steps. A world model first generates an image or video of what the scene should look like next. Then a second module pulls the right control commands from that output. Early work like UniPi generates complete videos and derives motion through a learned inverse model. Ad Other approaches like AVDC or 3DFlowAction use motion fields from which the robot's trajectory can be computed geometrically. Still others - VPP or LAPA, for instance - skip visible images entirely and predict the future in compressed, abstract representations. That saves the compute otherwise needed to render every single pixel. The second line, Joint WAMs, combines both tasks in a single model. Work like GR-1, GR-2, or WorldVLA treats images and actions as a unified token sequence. Diffusion-based variants such as PAD, UWM, or DreamZero generate the future frame and the movement in parallel. Nvidia's Cosmos Policy can use the same architecture as a controller, a simulator, or an evaluation model. Ad DEC_D_Incontent-2 Nvidia pursues a similar dual role with DreamDojo , a world model that takes control commands and generates a simulated visual future from them. The survey also discusses π0.7 , which uses the world model not as a replacement but as a supplier. It feeds imagined future frames into the context of a pretrained robotics AI, which then generates the movement. Ad