Xiaomi Unveils Robotics World Model with Factory Training Demo
CyberRobooo · x · 2026-07-14
Xiaomi has reportedly released an autoregressive world model, Xiaomi-Robotics-U0. A core capability is Real Robot Improvement, which uses style-transfer demonstrations generated by U0 alongside real-world data for policy post-training. This significantly enhances the robot's generalization across unfamiliar backgrounds, lighting changes, and visual distractions.
The same update also noted:
- Since Q1 this year, Xiaomi's humanoid robots have been training at dual-sided workstations in an auto factory.
- They have achieved a 98% success rate at self-tapping screw stations.
- CyberOne Gen2 can autonomously handle repetitive tasks in the general assembly logistics area, such as part sorting and material box folding.
More from Embodied
- DONIC hits 86.79% on bimanual dexterous tracking, crushing prior SOTA's 9.43% — liuziwei7 · 2026-10-27
- Bedrock Robotics raises $600M as its fully autonomous excavators do paid work on live job sites — Sethwinterroth · 2026-10-27
- AW-0 hits 80.8% on RoboTwin tasks, beating strongest baseline by 16 points — trevordarrell · 2026-10-27
- AW-0's code stream fixes world model prediction: picks mango where baseline grabs apple — trevordarrell · 2026-10-27
- AW-0 runs at 230ms on a single RTX 5090, 2.4x faster than autoregressive code planning — trevordarrell · 2026-10-27
- AW-0 debuts: one generative world action model plans in code, predicts, and controls robots — trevordarrell · 2026-10-27