Goal recognition for autonomous vehicles.
#Setup Make sure you are using Python 3.6 or later.
Install Lanelet2 following the instructions here.
Clone this repository:
git clone https://github.com/cbrewitt/av-goal-recognition.git
Install with pip:
cd av-goal-recognition
pip install -e .
Extract the datasets into the same directory which av-goal-recognition is placed.
Alternatively, edit the config files in av-goal-recognition/scenario_config to specify different directories.
Preprocess the data and Extract features:
cd core
python data_processing.py
Train the decision trees:
cd ../decisiontree
python train_decision_tree.py
Calculate evaluation metrics on the test set:
cd ../evaluation/
python evaluate_models_from_features.py
Show animation of the dataset along with inferred goal probabilities:
python run_track_visualization.py --scenario heckstrasse --goal_recogniser trained_trees