T2-z: Test-Time Optimization of Environment Latents for Robotic World Models
Presents novel architecture for robotic perception and adaptation to unseen environments through latent space optimization.
AI Summary
Researchers introduce T2-z, a method for test-time optimization of environment latents in robotic world models that improves prediction accuracy and action success rates across simulations and real-world tasks.
Excerpt
Predicting how the world responds to an action requires more than what can be seen: robots must account for environmental factors that are neither directly observable nor explicitly specified, from physical properties such as friction and mass to human-defined environmental rules. We introduce T2-Z for test-time optimization of environment latents, which augments a robotic world model with a learnable environment latent to condition world predictions. This compact latent captures underlying envi