We extend existing recurrent encoder-decoder models to be advantageously combined with anchor trajectories to predict vehicle behaviors on a roundabout. Drivers’ intentions are encoded by a set of maneuvers that correspond to semantic driving concepts. Accordingly, our model employs a set of maneuver-specific anchor trajectories that cover the space of possible outcomes at the roundabout. The proposed model can output a multi- modal distribution over the predicted future trajectories based on the maneuver-specific anchors. We evaluate our model using the public RounD dataset.
The RounD Dataset is a new dataset of naturalistic road user trajectories recorded at German roundabouts. Download the dataset to the 'rounD' directory, then run the following MATLAB script:
preprocess_rounD.m
This will do the required pre-processing, split the dataset into train, validation and test subsets, and save such subsets into the 'data' directory.
The default network arguments are in:
model_args.py
You can set the required experiment arguments in this script. For example:
-
args['ip_dim'] selects the input dimensionality (2D or 3D).
-
args['use_intention'] and args['use_anchors'] are Boolean variables that choose whether using intention prediction and anchor trajectories or not.
The model structure is coded in 'model.py'. After setting the required experiment arguments, create a 'trained_models' directory to save the trained models. You can start model training by running:
train.py
To test a trained model, first create an 'eval_res' directory, then run:
evaluate.py
which will load and test the trained model defined by the selected model arguments. The RMSE results will be saved as csv files to the 'eval_res' directory.
If you find this code useful for your research, please cite our work:
- Mohamed Hasan, Evangelos Paschalidis, Albert Solernou, He Wang, Gustav Markkula and Richard Romano, "Maneuver-based Anchor Trajectory Hypotheses at Roundabouts", preprint 2021.
This project is licensed under the MIT License - see the LICENSE.md file for details.