Embodied AI shifts from leaderboards to deployment: Annu Intelligence cuts robot rollout costs by 80%
智东西 · wechat · 2026-09-07
Zhidongxi profiles Annu Intelligence's full-stack embodied AI system as the industry pivots from benchmark chasing to real-world delivery. Key releases:
- ActiWorld: an action-anchored world model (inverse dynamics head, bidirectional prediction, motion gating) that filters out "visually plausible but action-insensitive" hallucinations. On an AgiBotG2 real-robot sorting task it lifted the frozen ACoT policy's success rate from 56% to 62% at zero inference overhead.
- AnuVerse-0.5: a distilled MoT-architecture vertical model running in 100ms on a single consumer GPU, already deployed for industrial tote-handling tasks.
- Nexus: a deployment system unifying model outputs into 24D absolute actions, decoupling VLA planning from 100Hz servo control, with an independent FK safety shield; cuts deployment time 30–50% across robot bodies.
Supported by ANUSimLab soft-body simulation, Helios long-video generation for synthetic data, and RoHIL/EvoHIL real-robot RL. At Fulin Precision's plant, first workstation took 4 months to deploy, now down to 3 days, MTBF over 300 hours, and industrial deployment costs cut 80%.
Related event: Embodied AI Shifts to Deployment as Annu Ships Six Models in a Week(2 posts)→
More from Embodied
- FLASH: GPU-native deformable sim learns cloth folding in minutes on one RTX 5090 — ChongZzZhang · 2026-10-26
- Luke Wroblewski spots it: iPhone Duo's corner radius mirrors the original iPhone — LukeW · 2026-10-26
- Humanoids shattered forecasts: half-marathon record in 2026, 100m faster than Bolt — ben_j_todd · 2026-10-26
- Humanoid walking is continuous correction — watch recovery, not locomotion, in demos — Olivier__OG · 2026-10-26
- CircuitHub unveils The Grid, a factory-scale robot making PCBs in the US twice as fast as Shenzhen — garrytan · 2026-10-26
- Shanghai AI Lab-led team releases 100-hour egocentric wearable dataset for humanoid training — ChuckDBrooks · 2026-10-26