Jiaolong / trajectory-prediction

Trajectory Prediction with Noisy Topometric Map for Autonomous Navigation

Trajectory Prediction for Autonomous Driving with Topometric Map

Repository for the paper "Trajectory Prediction for Autonomous Driving with Topometric Map".

@inproceedings{traj-pred:2022,
  title={Trajectory Prediction for Autonomous Driving with Topometric Map},
  author={J. Xu, L. Xiao, D. Zhao etal},
  booktitle={ICRA},
  year={2022}
}

Requirements

• python 3.6+

• pytorch 1.4+

Install build requirements

pip install -v -e .  # or "python setup.py develop"

Pretrained models

Pre-trained weights can be downloaded here(code: uf9g)

Dataset

├── datasets
│   └── KITTI_RAW
        └── trajectory_prediction
            ├── 07
            └── 10

Testing dataset kitti-10 can be downloaded here(code: kbuf)

Train & Test

Train

• Train with multiple GPUs:

sh tools/scripts/dist_train.sh ${NUM_GPUS} -c ${CONFIG_FILE}

• Train with a single GPU:

python tools/train.py --cfg config/trajectory_prediction/transformer.yaml

Test

• Test with a pretrained model:

python tools/test.py --cfg config/trajectory_prediction/transformer.yaml --ckpt cache/transformer_epoch_120.pth