Low Latency Trajectory Predictions For Interaction Aware Highway Driving

by Cyrus Anderson at UM FCAV

Introduction

The paper proposes a probabilistic model to predict trajectories of vehicles in critical highway merging scenarios, where the number of observations may be limited. For more details, check out the published paper on IEEE Xplore or the arxiv version.

Dependencies

• NumPy
• SciPy
• PICOS
• CVXOPT
• pandas
• matplotlib

The proposed method's semidefinite program is formulated and solved with the PICOS interface for conic optimization. The solver used is CVXOPT. The other dependencies are used for loading data and plotting.

Here is an example of creating an environment named py37sdp with Anaconda having the dependencies:

conda create -n py37sdp python=3.7 numpy scipy pandas matplotlib pip 
pip install cvxopt
conda install -c picos picos

Datasets

The NGSIM dataset is used to evaluate the method, whose root folder should be set in utils.py. The default setup uses datasets as a symbolic link:

baselines/
datasets/
|__ngsim/
    |__i-80/
    |__us-101/

where the ngsim folder contains the NGSIM dataset (dataset portal and homepage).

(Note: There may be small errors when first loading the data due to small inconsistencies between the folder name formats/column names of the US-101 and I-80 datasets - manually changing them can solve this.)

Predicting

Running display_driver.py will evaluate each baseline on ramp merge scenarios from NGSIM.

Citation

If you find this paper helpful, please consider citing:

@ARTICLE{anderson2020lowlatency,
  author={C. {Anderson} and R. {Vasudevan} and M. {Johnson-Roberson}},
  journal={IEEE Robotics and Automation Letters}, 
  title={Low Latency Trajectory Predictions for Interaction Aware Highway Driving}, 
  year={2020},
  volume={5},
  number={4},
  pages={5456-5463},
}