Plan-R1: Safe and Feasible Trajectory Planning as Language Modeling

---

Table of Contents

• Results
• Setup
• Datasets
• Training
• Validation
• Simulation
• Acknowledgements
• Citation

---

Results

Comparison with SOTAs on nuPlan benchmark.
NR/R: non-reactive/reactive mode. *: with rule-based post-processing.

Type	Planner	Val14 NR	Val14 R	Test14-hard NR	Test14-hard R	Test14-random NR	Test14-random R
Expert	Log-Replay	93.53	80.32	85.96	68.80	94.03	75.86
Rule-based & Hybrid	IDM	75.60	77.33	56.15	62.26	70.39	72.42
	PDM-Closed*	92.84	92.12	65.08	75.19	90.05	91.64
	PDM-Hybrid*	92.77	92.11	65.99	76.07	90.10	91.28
	Gameformer*	79.94	79.78	68.70	67.05	83.88	82.05
	PLUTO*	92.88	89.84	80.08	76.88	92.23	90.29
	PlanAgent*	93.26	92.75	72.51	76.82	-	-
	Diffusion Planner*	94.26	92.90	78.87	82.00	94.80	91.75
	Carplanner*	-	-	-	-	94.07	91.10
	Plan-R1* (Ours)	94.72	93.54	78.46	81.70	94.64	93.71
Learning-based	UrbanDriver	68.57	64.11	50.40	49.95	51.83	67.15
	PDM-Open	53.53	54.24	33.51	35.83	52.81	57.23
	PlanTF	84.27	76.95	69.70	61.61	85.62	79.58
	PLUTO	88.89	78.11	70.03	59.74	89.90	78.62
	Diffusion Planner	89.87	82.80	75.99	69.22	89.19	82.93
	Plan-R1 (Ours)	88.98	87.69	77.45	77.20	91.23	90.04

---

Setup

1. Create Environment

conda create -n planr1 python=3.9
conda activate planr1
conda install pytorch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 pytorch-cuda=12.1 -c pytorch -c nvidia
conda install pytorch-lightning==2.0.3
pip install torch-scatter==2.1.2 -f https://data.pyg.org/whl/torch-2.1.0+cu121.html
pip install torch_geometric==2.3.1

2. Download the pre-trained model weights and organize the directory as:

Plan-R1
├── ckpts
│   ├── pre-training.ckpt
│   └── fine-tuning.ckpt
├── ...

---

Datasets

1. Download nuPlan Dataset and organize the directory as:

~
├── Plan-R1
└── nuplan
    └── dataset
        ├── maps
        │   ├── nuplan-maps-v1.0.json
        │   ├── sg-one-north
        │   │   └── 9.17.1964
        │   │       └── map.gpkg
        │   ├── ...
        └── nuplan-v1.1
            └── splits
                ├── train
                ├── val
                └── test
                    ├── 2021.05.25.12.30.39_veh-25_00005_00215.db
                    ├── ...

2. Install nuplan-devkit

cd ~/nuplan
git clone https://github.com/motional/nuplan-devkit.git && cd nuplan-devkit
pip install -e .
pip install pip==24.0
pip install -r requirements.txt
pip install tensorboard
pip install numpy==1.24.4

---

Training

1. Preprocess Dataset

Preprocessing may take a long time (~30 hours).

python preprocess_dataset.py

2. Pre-training

python train.py --config config/train/pred.yaml

3. Fine-tuning

python train.py --config config/train/plan.yaml

---

Validation

To visualize results during validation, set val_visualization = True in config/val/pred.yaml or config/val/plan.yaml.

# For Pre-training
python val.py --config config/val/pred.yaml
# For Fine-tuning
python val.py --config config/val/plan.yaml

---

Simulation

bash simulation/run_simulation.sh <sim_type> <planner> <split> <ckpt_path>

Config

Argument	Description / Options
<sim_type>	closed_loop_nonreactive_agents, closed_loop_reactive_agents, open_loop_boxes
<planner>	planr1_planner, planr1_planner_with_refinement
<split>	val14, test14-random, test14-hard

Example

bash simulation/run_simulation.sh closed_loop_nonreactive_agents planr1_planner test14-random ckpts/fine-tuning.ckpt

Visualization with NuBoard

python run_nuboard.py

---

Acknowledgements

We thank the following works for their contributions and inspiration to this project:

• nuPlan
• SMART
• PDM
• PLUTO
• Diffusion Planner
• STR
• HPNet

---

Citation

If Plan-R1 has been helpful in your research, please consider citing our work:

@article{tang2025plan,
  title={Plan-R1: Safe and Feasible Trajectory Planning as Language Modeling},
  author={Tang, Xiaolong and Kan, Meina and Shan, Shiguang and Chen, Xilin},
  journal={arXiv preprint arXiv:2505.17659},
  year={2025}
}