# OpenDriveLab/DriveLM

[ECCV 2024 Oral] DriveLM: Driving with Graph Visual Question Answering

Repository: https://github.com/OpenDriveLab/DriveLM
Canonical: https://ross.abutalabs.com/products/drivelm
Homepage: https://opendrivelab.com/DriveLM/
Language: HTML
License: Apache-2.0
License Family: permissive
Topics: autonomous-driving, large-language-models, vision-language, chain-of-thought, graph-of-thoughts, llm, prompting, tree-of-thoughts, prompt-engineering
Last push: 2025-07-02T05:29:49+00:00

## Health v2 (maintenance only)
Score: 41/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 29, release rhythm 35, longevity 80
- inputs: {"age_days": 1121, "days_push": 427, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1338, forks 88 (observed 2026-08-28T04:04:25.818660+00:00)

## What it is
DriveLM is a research project and dataset for autonomous driving built around Graph Visual Question Answering (GVQA), where QA pairs about perception, prediction, and planning are connected in a graph structure. It includes DriveLM-Data (built on nuScenes and CARLA) and DriveLM-Agent, a VLM-based baseline for end-to-end driving, and served as the basis for the CVPR 2024 Driving-with-Language challenge.

## Use cases
- train vision-language models for autonomous driving reasoning
- build end-to-end driving agents with chain-of-thought planning
- evaluate VLMs on perception prediction and planning QA tasks
- research graph-of-thoughts prompting for driving scenarios
- participate in autonomous driving with language challenges
- generate what-if counterfactual reasoning about driving scenes

## When to choose
- you need annotated driving QA data with logical dependencies between questions
- you are researching VLM-based end-to-end autonomous driving
- you want a benchmark for structured visual reasoning in driving
- you are entering a driving-with-language challenge

## When to avoid
- you need a production-ready autonomous driving stack
- you need real-time vehicle control software
- your project has nothing to do with driving or embodied AI
- you lack GPU resources for training large vision-language models

## Facets
- artifact type: dataset
- maturity: active
- function: machine-learning, nlp, computer-vision, prompt-engineering, rag
- domain: autonomous-vehicles, large-language-models, artificial-intelligence, computer-vision
- platform: python
- tags: vision-language-models, graph-visual-question-answering, end-to-end-driving, chain-of-thought, nuscenes, carla, benchmark-dataset, eccv-2024, natural-language-processing, linux, gpu

## Member repositories
- OpenDriveLab/DriveLM (main) score 41

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:25.818660+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:43:58.097878+00:00, confidence not recorded.
  - readme: https://github.com/OpenDriveLab/DriveLM (fetched 2026-08-28T04:04:25.818660+00:00, sha 4237cf2813d4)
  - homepage: https://opendrivelab.com/DriveLM/ (fetched 2026-08-29T12:03:00.275904+00:00, sha a322fbd79e69)
- Data as of 2026-08-30T08:39:29.467469+00:00.