Draw Forces, Get Plans: Goal Force for Physics‑Savvy Video Models
What if you could guide a robot by drawing arrows that show pushes and pulls? Goal Force is a new way to steer video world models with physics, not vague text prompts.
Instead of describing goals in words or target images, users sketch force vectors and intermediate dynamics. The model is trained on simple “causal primitives” (like elastic collisions and falling dominos) so it learns to propagate forces through time and space.
Despite training only on synthetic physics, Goal Force generalizes zero-shot to complex, real-world scenes—tool use, multi-object cause-and-effect—acting as an implicit neural physics simulator. The result: precise, physics-aware planning without external engines.
Paper and demos: https://arxiv.org/abs/2601.05848v1
Paper: https://arxiv.org/abs/2601.05848v1
Register: https://www.AiFeta.com
AI Robotics VideoGeneration WorldModels Physics Simulation Planning ComputerVision OpenSource
