Research map / Vision-language and multimodal / Robot action and navigation
DrivingGPT: Unifying Driving World Modeling and Planning with Multi-modal Autoregressive Transformers
Yuntao Chen, Yuqi Wang, Zhaoxiang Zhang
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 |
|---|---|---|---|
| replace diffusion SVD fine-tuning with autoregressive DrivingGPT for video generation | video world model · navtest video generation | FVD: 227.54 -> 142.61 | Improved |
| replace diffusion SVD fine-tuning with autoregressive DrivingGPT for video generation | video world model · navtest video generation | FID: 24.03 -> 12.78 | Improved |
| use DrivingGPT instead of SVD for 64-frame long video generation | video world model · navtest long video generation (64 frames) | FVD: 1079.28 -> 506.95 | Improved |
| fine-tune SVD on navtrain instead of using it zero-shot | SVD · navtest video generation | FVD: 483.76 -> 227.54 | Improved |
| keep predicted longitudinal x action instead of copying last history action (ablation) | DrivingGPT planner · NAVSIM navmini planning | PDMS: 53.5 -> 82.4 | Improved |
| keep predicted lateral y action instead of copying last history action (ablation) | DrivingGPT planner · NAVSIM navmini planning | PDMS: 79.4 -> 82.4 | Improved |
| keep predicted yaw action instead of copying last history action (ablation) | DrivingGPT planner · NAVSIM navmini planning | PDMS: 73.1 -> 82.4 | Improved |
| train on curated NAVSIM data (104k sequences) rather than 651k nuPlan uniform sequences (ablation) | DrivingGPT planner · NAVSIM navmini planning | PDMS: 74.6 -> 82.4 | Improved |
| add 1D rotary position embedding to action tokens as well as image tokens (ablation) | DrivingGPT planner · NAVSIM navmini planning | PDMS: 65.3 -> 82.4 | Improved |
| use Llama-Gen tokenizer instead of OpenMAGVIT2 (ablation) | DrivingGPT visual tokenizer · navtest tokenizer reconstruction | rFID: 6.70 -> 5.67 | Improved |
| use Llama-Gen tokenizer instead of Cosmos Tokenizer (ablation) | DrivingGPT visual tokenizer · navtest tokenizer reconstruction | PSNR: 25.05 -> 23.09 | Mixed |
| DrivingGPT without auxiliary supervision versus LAW | DrivingGPT · NAVSIM navmini planning | PDMS: 82.7 -> 82.4 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports DrivingGPT as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- DriveWorld-VLA: Unified Latent-Space World Modeling with Vision–Language–Action for Autonomous Driving (ICML 2026 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Driving into the Future: Multiview Visual Forecasting and Planning with World Model for Autonomous Driving (CVPR 2024 · similarity 0.90)
- DriveGPT: Scaling Autoregressive Behavior Models for Driving (ICML 2025 · similarity 0.90)
- From Forecasting to Planning: Policy World Model for Collaborative State-Action Prediction (NeurIPS 2025 · similarity 0.89)
- Copilot4D: Learning Unsupervised World Models for Autonomous Driving via Discrete Diffusion (ICLR 2024 · similarity 0.89)
- Generating Multimodal Driving Scenes via Next-Scene Prediction (CVPR 2025 · similarity 0.89)
- ChainFlow-VLA: Causal Flow Planning with Vision-Language Models (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.
