OpenBMB Open-Sources MiniCPM-Robot: Embodied AI That Keeps Tracking Offline
iamfakhrealam · x · 2026-07-24
OpenBMB has open-sourced its first embodied AI model series, MiniCPM-Robot, at WAIC. The release features a wild live demo where unplugging a robot dog's network cable doesn't interrupt its tracking at all.
The series includes three main components:
- MiniCPM-RobotTrack: A 0.9B parameter model built on MiniCPM4-0.5B. It runs entirely on the robot's onboard hardware (e.g., Unitree Go2 Edu) with no internet required, achieving 180ms end-to-end latency at 5+ Hz.
- MiniCPM-RobotManip: A 1.5B general-purpose VLA model. Using efficient visual token compression, it matches the performance of Qwen-VLA and π0.5 while using roughly half the visual-processing compute.
- Inference Optimization: With PhyAI’s CUDA Graph optimization and custom Triton fused kernels, inference throughput jumps from 10 Hz to 33 Hz, and up to 36 Hz on NVIDIA H20.
Related event: OpenBMB Open-Sources MiniCPM-Robot Embodied AI Series(8 posts)→
More from Embodied
- Podcast preview: HUG explores human universal grasping for robots — chris_j_paxton · 2026-07-25
- Waymo-like cars for personal ownership should be a bigger push, author says — _xjdr · 2026-07-24
- EngineAI’s T800 robot draws T-1000 and Torment Nexus jokes — Darpinian · 2026-07-24
- Robot builder says Bob died after a battery mod fried its Raspberry Pi brain — chrismatthieu · 2026-07-24
- Autonomous Labs Employee Teases Unreleased AI Hardware Prototype — dee_hw · 2026-07-24
- Brazilian surgeon operates on a patient 12,034 km away with 199 ms latency — aakashgupta · 2026-07-24