vision-cair / unlikelihoodmotionforecasting

Motion Forecasting with Unlikelihood Training inContinuous Space

This repository contains the code for Motion Forecasting with Unlikelihood Training in Continuous Space. Here we applied the unlikelihood loss with Trajectron++ and test on the nuScenes dataset. Code is based on its official implementation.

Installation

Environment Setup

conda create --name unlike python=3.6
conda activate unlike
pip install -r requirements.txt

Dataset Setup

A preprocessed nuScenes Dataset is provided here In case you'd like to preprocess the dataset by youself, first download the nuScenes dataset (this requires signing up on their website. We use v1.0 following Trajectron++).
Then, download the map expansion pack (v1.1) and copy the contents of the extracted maps folder into the maps folder of the dataset. Finally, process them into a data format that our model can work with. We use CuPy, a cuda version of numpy, to speed up the data preprocessing. You can install it following here

cd experiments/nuScenes
python process_data.py --data=./v1.0 --version="v1.0" --output_path=/path/to/save/

Model Training

To train a model on the nuScenes dataset, you can execute one of the following commands from within the trajectron/ directory, depending on the model version you desire.

Model	Command
Traj.++ with L_unlike	python train.py --conf ../experiments/nuScenes/models/int_ee_me/config.json --train_data_dict /path/to/nuScenes_train_full.pkl --eval_data_dict /path/to/nuScenes_val_full.pkl --offline_scene_graph yes --preprocess_workers 12 --log_dir /path/to/log_root --train_epochs 35 --node_freq_mult_train --log_tag unlike --map_encoding --augment
Traj.++	python train.py --conf ../experiments/nuScenes/models/int_ee_me/config.json --train_data_dict /path/to/nuScenes_train_full.pkl --eval_data_dict /path/to/nuScenes_val_full.pkl --offline_scene_graph yes --preprocess_workers 12 --log_dir /path/to/log_root --train_epochs 35 --node_freq_mult_train --log_tag traj++ --map_encoding --augment --unlike_type no

Model Evaluation

To evaluate a trained model's performance on forecasting vehicles, you can execute a one of the following commands from within the experiments/nuScenes directory. We provide three pretrained models in experiments/nuScenes/models/. int_ee_me\ contains the original pretrained Trajectron++ same as the one in their repository. trajectron_pp\ contains the pretrained Trajectron++ with our hyperparameters that lead to a better performance. unlikelihood\ contains the pretrained model trained with the unlikelihood loss.

python evaluate.py --model /path/to/model/ --checkpoint the_one_you_like --data /path/to/nuScenes_test_full.pkl --prediction_horizon 8