DriveTeach-VLA: Teaching Vision-Language-Action Models What to See and Where to Look
jiqizhixin · x · 2026-08-18
Researchers from Beihang University, Tsinghua University, and DiDi present DriveTeach-VLA, a new training framework for autonomous driving models. Current AI drivers get overwhelmed by text-heavy reasoning data, losing focus on spatial clues needed for safe navigation.
DriveTeach-VLA changes this by teaching models a two-step vision: first, identifying what visual details matter for driving, then guiding attention based on real, feasible driving paths. This links seeing with acting. The result is SOTA performance on NAVSIM and nuScenes benchmarks, with code available on GitHub.
More from Embodied
- Former Google Research Scientist Alex Toshev joins Wayve to lead General Robotics Intelligence — alexgkendall · 2026-08-19
- CFP: Continual World Models Workshop @ NeurIPS 2026 — cindy_x_wu · 2026-08-19
- Robotics' biggest hurdle may be endurance, not intelligence — aftahi_ai · 2026-08-19
- Foxglove, the robotics data platform, goes agentic — Sethwinterroth · 2026-08-19
- DynaRobotics podcast: scaling robot pretraining toward 1 million hours of data — chris_j_paxton · 2026-08-19
- Entertainment Is the First Mass Market for Humanoid Robots — chris_j_paxton · 2026-08-19