PokeBot CTO predicts zero-shot generalization on 80% of household tasks within a year
suchenzang · x · 2026-09-12
In a Zhihu Frontier podcast, Tsinghua IIIS professor Huazhe Xu and PokeBot CTO Zhengrong Xue argue 80% of common household tasks could see zero-shot generalization within a year. Key points: their 3-month mapo-tofu end-to-end demo (long-horizon, contact-rich tasks are hardest); failed trajectories are undervalued data — quality beats quantity; sparse 0/1 rewards work best in homes; action chunking was a watershed; touch sensing may be overrated; they await a GPT-3-style turning point that makes some problems moot.
More from Embodied
- Adam Dorr asks if US should ban humanoid robot exports to all foreign countries — adam_dorr · 2026-09-12
- GPT-6 Astra shows 'step change' in spatial reasoning, solving 7/100 robot tasks vs zero for rival — The Decoder · 2026-09-12
- China's snake-like grid robot draws power from live wires, runs 24/7 inspections — prasanna_says · 2026-09-12
- Headsets hit 100-200g yet the criticism never stops: a dev's take on the VR doom loop — Scobleizer · 2026-09-12
- Houston Methodist tests Apple Vision Pro to give surgeons better views in real procedures — Scobleizer · 2026-09-12
- Weekend robot fights at autonomous labs: train your bot by chatting with an AI coach — dee_hw · 2026-09-12