NAVSIM baselines (TransFuser, LTF)
- AI model
- Cars
NAVSIM is a pseudo-simulation framework for evaluating autonomous driving planners, with baseline agents such as TransFuser and LTF. It targets fast, large-scale benchmarking of driving models.
Source on GitHub: github.com/autonomousvision/navsim (opens in a new tab). Facts read at commit 0a380a9 (opens in a new tab).
Licenses
- Code
- Apache-2.0 (opens in a new tab)
Read by Robotheca from LICENSE (opens in a new tab) (Apache-2.0) at commit 0a380a9 (opens in a new tab), on 16 Oct 2026.
- Weights
- Apache-2.0 (opens in a new tab)
Read on huggingface.co/autonomousvision/navsim_baselines (opens in a new tab).
- Commercial use
- Yes
What it does
- Pseudo-simulation augments real driving data with synthetic observations near the planned trajectory, combining open-loop efficiency with closed-loop robustness.
- Evaluates planners on all metrics open-loop, without sequential or interactive simulation, reportedly requiring about 6x less compute than closed-loop simulation.
- Includes baseline agents such as TransFuser and LTF, documented in the agents guide.
- Provides an Extended PDM Score metric and a two-stage pseudo closed-loop simulation with reactive traffic agents.
- Used as the primary evaluation framework for the AGC2025 NAVSIM end-to-end driving challenge, with public leaderboards.
Written by Robotheca from the project's README (opens in a new tab) at commit 0a380a9.
Getting started
- Difficulty
- Advanced
- How to start
- The project's README (opens in a new tab)
Maintenance
- Status
- Maintained: a commit in the last 12 months
- Newest commit
- 27 Oct 2025 on main
- Latest release
- NAVSIM v2.1.2 release (28 Apr 2025)
Checked on GitHub on 16 Oct 2026.
Links
Robots that can run it
No robot on Robotheca is linked to it yet. Links come from a robot's robot.yaml, from its dependency files read as text, or from Robotheca's editors.
Robotheca lists AI models as a reference: nothing is hosted or run here, and every fact links to where it was read.