Research map / Vision-language and multimodal / Robot action and navigation
Bird's-eye-view Informed Reasoning Driver
Yinuo Wang, Mining Tan, Yuanxin Zhong, Wang zhitao, Siyuan Cheng
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Spatial Localization dataset (ablation) | BIRDriver VLM (Qwen2.5VL-3B) · Test14-hard clips | x-error: 4.17m -> 3.76m (x), 1.28m -> 1.08m (y), 4.20° -> 3.80° (phi) | Improved |
| add Driving Scene Stepwise dataset (ablation) | BIRDriver VLM (Qwen2.5VL-3B) · Test14-hard clips | x-error: 4.27m -> 4.17m (x), 1.35m -> 1.28m (y), 4.23° -> 4.20° (phi) | Improved |
| use full composite dataset vs KeyPoint only (ablation) | BIRDriver VLM (Qwen2.5VL-3B) · Test14-hard clips | key point error: reduced by 11.9%, 20.0%, and 10.2% | Improved |
| add weighted SFT loss (ablation) | BIRDriver VLM (Qwen2.5VL-3B) · Test14-hard clips | key point error: 4.22m -> 3.76m (x), 1.19m -> 1.08m (y), 4.13° -> 3.80° (phi) | Improved |
| use RDP key points instead of final trajectory point only (ablation) | BIRDriver (InternVL2.5-4B) · InterPlan | CLS-R: 34.72 -> 53.81 | Improved |
| scale VLM from InternVL2.5-2B to 4B (ablation) | BIRDriver VLM · Key point error | x-error: 6.59m -> 3.71m (x), 2.79m -> 1.44m (y), 11.20° -> 4.89° (phi) | Improved |
| scale VLM from Qwen2.5VL-3B to 7B (ablation) | BIRDriver VLM · Key point error | x-error: 3.76m -> 3.60m (x), 1.08m -> 1.08m (y), 3.80° -> 3.83° (phi) | No clear effect |
| add VLM key point guidance to PLUTO | PLUTO · Test14-hard | CLS-NR / CLS-R: 80.03 -> 80.56 (CLS-NR), 76.92 -> 80.33 (CLS-R) | Improved |
| add VLM key point guidance to PLUTO | PLUTO · InterPlan | CLS-R: 48.92 -> 55.29 | Improved |
| add VLM key point guidance to PLUTO | PLUTO · Test14-random | CLS-NR: 91.87 -> 91.46 | Hurt |
| add VLM key point guidance to PLUTO | PLUTO · Test14-random | CLS-R: 90.03 -> 91.26 | Improved |
Closest papers by meaning
- Drive-R1: Bridging Reasoning and Planning in VLMs for Autonomous Driving with Reinforcement Learning (AAAI 2026 · similarity 0.89)
- SGDrive: Scene-to-Goal Hierarchical World Cognition for Autonomous Driving (CVPR 2026 · similarity 0.89)
- SpaceDrive: Infusing Spatial Awareness into VLM-based Autonomous Driving (CVPR 2026 · similarity 0.88)
- Intend, Reflect, Refine: An Adaptive Multimodal Reflection Framework for Autonomous Driving (NeurIPS 2026 · similarity 0.88)
- Explicit Geometric Chain-of-Thought for Vision-Language-Action in Autonomous Driving (NeurIPS 2026 · similarity 0.88)
- OneVL: One-Step Latent Reasoning and Planning with Vision-Language Explanation (NeurIPS 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
