Openmind Unveils Embodied AI Base and HumanGPT
机器之心 · wechat · 2026-07-20
At WAIC, Openmind (spun off from Efort) made its debut. Instead of showcasing acrobatic stunts with physical robots, it released a universal technology base for intelligent robots and its self-developed operational world model, HumanGPT, directly addressing the high thresholds, high costs, and low efficiency of embodied AI development.
Acting like an "Android" for the robotics industry, the base comprises three layers: OpenmindOS handles hardware adaptation downwards and provides a unified runtime environment upwards; the HALO skill suite and the Dayan data platform manage the collection and governance of human operation data; and the Modu IDE offers low-code development. The HumanGPT model extracts skill representations and causal relationships from real human operation data, endowing robots with a "mental sandbox" for deduction, thereby achieving large-and-small brain collaboration for long-horizon complex tasks.
Currently, the base has been validated in industrial and commercial scenarios like smart spraying and medical triage, and has completed ecosystem integrations with domestic computing power from Huawei and multiple robot bodies from Siasun and Leju.
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