Tsinghua AIR and UC Berkeley BAIR propose ODEWorld, a continuous-time embodied world model
AIthat learns the velocity field of a compact dynamic latent space rather than predicting the next frame. The model integrates an ODE solver to generate future states at arbitrary times, which the authors say enables arbitrary frame-rate generation, filling in frames missing from training video, and reverse rollouts.
AIWhy it matters.
The paper replaces frame-by-frame prediction with a learned velocity field, a different approach to robot world models that readers building embodied systems can compare with video-prediction methods.
Source details
Tsinghua AIR · air.tsinghua.edu.cnPublished · added here
Original title: AIR科研｜连续时间具身世界模型：高效生成、任意补帧，还能“倒着推演”