Robotics researcher: post-training and recursive improvement are the next bottleneck
animesh_garg · x · 2026-09-08
Robotics researcher Animesh Garg on the MTS podcast: he's excited about the return of RL in new packaging, and argues that once the data problem is (somewhat) solved, post-training / recursive improvement becomes the real bottleneck again.
- Cheap, standardized hardware + modern RL methods could accelerate progress faster than people expect
- Sub-$500 toys like the microduck let high schoolers play with cutting-edge robotics RL
- "The future of robotics is going to be very bright."
The retweeter adds that recursive improvement is the key lever once data is no longer limiting.
More from Embodied
- Robotics researcher: harness + tool calls beat raw VLA on control benchmarks, at 2.3x lower cost — rbhar90 · 2026-09-08
- Grok bot plays a Roland piano, composing romantic-style music from a text prompt — yunta_tsai · 2026-09-08
- 360CityArena at ECCV 2026: best AI agent scores 17.1% vs humans' 77.3% on city navigation — CSProfKGD · 2026-09-08
- Open-source C-based LLM engine runs Gemma voice conversations on Jetson Orin, faster than llama.cpp — cortexist · 2026-09-08
- Robotics veteran Jitendra Malik challenges LLM robotics hype with a dexterity benchmark — JitendraMalikCV · 2026-09-08
- Self-driving demo runs on nothing but logic gates, moves like an insect — burny_tech · 2026-09-08