RoboTTT Extends Robot Memory to 5 Minutes
seanmcdonaldxyz · x · 2026-07-16
This reply introduces RoboTTT: a robotic policy that embeds test-time training (TTT) directly inside the model.
Key points include:
- Robotic models typically remember only extremely short time slices (under 0.1 seconds), forgetting immediate events almost instantly.
- RoboTTT expands the context to 8,000 timesteps, equivalent to roughly 5 minutes of "muscle memory."
- The author states this is achieved at a constant inference cost, outperforming current SOTA capabilities by 3 orders of magnitude.
- The core concept involves having a tiny model "live" inside a larger model, continuously updating as it processes each new input.
Related event: RoboTTT brings test-time training to robot policies(6 posts)→
More from Embodied
- Amazon and Google sold 600M+ smart speakers, so why no AGI-era successor? — julianlehr · 2026-09-11
- ECCV26 Oral: Flow Matching Enables Single-Stage Multi-View Point Cloud Registration — ducha_aiki · 2026-09-11
- Polish developers build iPhone app that detects nearby Meta smart glasses — Low-Honeydew6483 · 2026-09-11
- Ant's Afu health AI hits 150M users, unveils AI+hardware health alliance at Bund Summit — APPSO · 2026-09-11
- Johns Hopkins Launches Full-Stack Hands-on Robot Learning Class with SO-101 Arm Kits — _krishna_murthy · 2026-09-11
- SyncWorld: In-Context Robot World Model Simulates Unseen Views and Embodiments Zero-Shot — ChongZzZhang · 2026-09-11