One-shot imitation performance remains flat despite model scale
animesh_garg · x · 2026-08-30
Animesh Garg compares 2020 and 2026 robotics techniques: while modern generalist models (like GEN-1.5 and Skild) achieve "emergent" physical in-context learning directly from video prompts, the performance rates for one-shot imitation remain roughly the same as those achieved in 2020 using learned Neural Task Graphs (NTG). He argues that task specification is effectively a code generation task and predicts high-performance systems will rely on code generation (e.g., ProgPrompt) using atomic skills.
Related event: Robot One-Shot Imitation: Bigger Models, Little Real Progress(2 posts)→
More from Embodied
- SHOES: 10 minutes of real-robot RL lets quadrupeds walk in shoes, CoRL 2026 paper — kscottz · 2026-10-16
- Tesla Cybercab robotaxis logged zero crashes in first two weeks of public rides, NHTSA data shows — elonmusk · 2026-10-16
- Robotics LLMs tested on 16 tasks: GPT-6 Astra hits 17.2% success vs Fable's 9.1% — pabbeel · 2026-10-16
- WEDGE paper generates training environments conditioned on navigation agent failures — _akhaliq · 2026-10-16
- Viral meme nails AI's irony: it can write symphonies but can't do the dishes — briannekimmel · 2026-10-16
- Knödel Watch: animated vocabulary cards on ESP32-S3, open-sourced 4 days after unboxing — alexcovo_eth · 2026-10-16