GlanceWAM: Sparse Test-Time Imagination for World-Action Models
GlanceWAM: Sparse Test-Time Imagination for World-Action Models
GlanceWAM is a world-action model that enables real-time inference and high task success rates by asynchronously generating visual imagination in the background while decoding actions in latent space. It outperforms synchronous baselines in both simulation and real-world robotic tasks, achieving 72.2% on RoboCasa and 99.0% on LIBERO with significantly reduced control latency.
This paper introduces GlanceWAM, a world-action model that decouples visual imagination from control to achieve real-time inference and improved task success rates. The model uses asynchronous generation and latent-space decoding to reduce latency while maintaining performance, outperforming synchronous baselines in both simulation and real-world robotic tasks.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
control latency
24
View original evidence
latent space without blocking. Enabled by a non-interfering attention mask that isolates video representations and staleness-robust horizon training that accommodates asynchronous lookahead aging, GlanceWAM breaks the speed-success dilemma. Trained purely on demonstrations, it attains 72.2% on the 24-task RoboCasa kitchen benchmark (vs. 67.1% for synchronous Cosmos Policy) and 99.0% on LIBERO while cutting per-chunk
Open source S4
- code: Available Artifact link Open source S5
- code: Available Artifact link Open source S5
- code: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
latent space without blocking. Enabled by a non-interfering attention mask that isolates video representations and staleness-robust horizon training that accommodates asynchronous lookahead aging, GlanceWAM breaks the speed-success dilemma. Trained purely on demonstrations, it attains 72.2% on the 24-task RoboCasa kitchen benchmark (vs. 67.1% for synchronous Cosmos Policy) and 99.0% on LIBERO while cutting per-chunk
Open source S4
control latency 24x relative to synchronous world-action models (48 ms on one A100). In single-arm and bimanual real-robot manipulation, it achieves higher average success than $\pi_{0.5}$ without any robot-data pretraining. Code is available at https://github.com/linhanwang/GlanceWAM and real-robot videos at https://linhanwang.github.io/glancewam/.
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org