Local Qwen models power a robot that tests 78 smartphones’ battery life
gappyvalley · reddit · 2026-07-27
A smartphone battery-testing video shows a battery-test robot using a vision-language agent to decide taps and navigation, with decisions made locally to avoid cloud latency and network variance.
Key details:
- The team built a custom inference server with an NVIDIA H20 and an RTX PRO 6000 Blackwell.
- It runs Qwen3.6-35B-A3B for fast, scroll-heavy tasks and Qwen3.6-27B for more precise actions.
- The two models are paired so the 27B model can step in when the faster model misreads the screen.
- The full image-to-path-planning loop reportedly finishes in under two seconds.
More from Embodied
- XPENG's IRON robot demos full-duplex speech with 9-mic array and lip reading — ChrisGPT · 2026-09-23
- Uber riders can now hail Waymo robotaxis on Austin freeways — ATTlKA · 2026-09-23
- M5 Ultra LLM test: 4x faster prompt processing, but double the power draw — DigitalguyCH · 2026-09-23
- Figure CEO: Humanoid Robotics Needs Four Stages and Eventually Hundreds of Billions — adcock_brett · 2026-09-23
- InstinctFlash: open-source runtime runs 8 robot model families in real time on one commercial GPU — chris_j_paxton · 2026-09-23
- M5 Ultra 96GB vs RTX 5090-6000 Pro: which rig for local video generation at $6.5K-$27K — MaxwellHusk · 2026-09-23