Scene-graph augmented data-driven risk assessment of autonomous vehicle decisions

This repository includes the code and dataset information required for reproducing the results in our paper.

In this repo, we have also integrated the source code of our baseline method, DeepTL-Lane-Change-Classification, into this repo. Infers the risk level of lane change video clips with deep learning. Utilizes deep transfer learning (TL) and spatiotemporal networks.

As for fabricating the lane-changing datasets, we use Carla CARLA 0.9.8 which is an open-source autonomous car driving simulator. Besides, we also utilized the scenario_runner which was designed for CARLA challenge event. For real-driving datasets, we used Honda-Driving Dataset (HDD) in our experiments. We published the converted scene-graph datasets used in our paper here. **brandon fix this.

The architecture of this repository is as below:

• sg-risk-assessment/: this folder collects all the core model/trainer/utilties used for our scene-graph based approach.
• baseline-risk-assessment/: this folder collects all the related source file that our baseline method requires
  • Mask_RCNN is the module that handle object detection and coloring on the image sequence.
• sg_risk_assessment.py: the script that triggers our scene-graph based approach.
• baseline_risk_assessment.py: the script that triggers our baseline algorithm.

To get started **brandon fix this.

We recommend our potential users to use Anaconda as the primary virtual environment.

$conda install python=3.6.8
$conda install -c anaconda cython=0.29.10
$conda install -c aaronzs tensorflow-gpu
$conda install git
$pip install git+https://github.com/philferriere/cocoapi.git#subdirectory=PythonAPI
$pip install -r requirements.txt

Since our primary working environment is Windows, we refer to this solution to have pycocotools to be installed.

Usages **brandon fix this.

To be filled.

• Information about pretrained coco model: Mask-RCNN pretrained weights can refer to this link and move it under [project folder]/pretrained_model (credit to [this repo](Mask R-CNN implementation by Matterport).

Citation

Please kindly consider citing our paper if you find our work useful for your research

@article{yu2020scene,
  title={Scene-graph augmented data-driven risk assessment of autonomous vehicle decisions},
  author={Yu, Shih-Yuan and Malawade, Arnav V and Muthirayan, Deepan and Khargonekar, Pramod P and Faruque, Mohammad A Al},
  journal={arXiv preprint arXiv:2009.06435},
  year={2020}
}