Alibaba's Qwen-Drive 1.0 unifies perception, traffic Q&A and planning in one driving model
The Decoder · rss · 2026-09-07
Alibaba's research arm has released Qwen-Drive 1.0, a single AI model that handles environmental perception, traffic Q&A, and route planning together, with the goal of running both the cockpit and the driving system from one model. The researchers show that text-image models don't automatically understand 3D space—spatial awareness must be trained deliberately. The model can explain why it brakes, though its explanation may not always match the actual maneuver.
More from Embodied
- Cybercab vs Waymo spoof revives the classic 'I'm a Mac' ad format — bennash · 2026-09-07
- 9 steps to generate 3D-printable parts with llama.cpp, FreeCAD MCP and a local model — DevelopmentBorn3978 · 2026-09-07
- Huawei unveils Kirin 9050 Pro based on its Tau Scaling Law, claims 42% performance jump — pstAsiatech · 2026-09-07
- RTX 5090 owner seeks controllable segmented workflows for Wan2.2 local video generation — Legionhome81 · 2026-09-07
- Cyborg cockroaches with cameras and drug injectors built for disaster rescue — ChuckDBrooks · 2026-09-07
- Tesla Optimus 3 and Figure 4 humanoids to enter initial production by year-end, says analyst — davidpattersonx · 2026-09-07