Enhancing Automated Vehicle Safety Through Testing with Realistic Driver Models [supporting dataset]
-
2023-12-07
-
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
DOI:
-
Resource Type:
-
Geographical Coverage:
-
Corporate Publisher:
-
Abstract:Driver process models play a central role in the testing, verification, and development of automated and autonomous vehicle technologies. Prior models developed from control theory and physics-based rules are limited in automated vehicle applications due to their restricted behavioral repertoire. Data-driven machine learning models are more capable than rule-based models but are limited by the need for large training datasets and their lack of interpretability. In this project we developed a novel car following modeling approach using active inference, which has comparable behavioral flexibility to data-driven models while maintaining interpretability. We assessed the proposed model, the Active Inference Driving Agent (AIDA), through a benchmark analysis against several benchmarks. The models were trained and tested on a real-world driving dataset using a consistent process. The testing results showed that the AIDA predicted driving controls significantly better than the rule-based Intelligent Driver Model and had similar accuracy to the data driven neural network models in three out of four evaluations. Subsequent interpretability analyses illustrated that the AIDA's learned distributions were consistent with driver behavior theory and that visualizations of the distributions could be used to directly comprehend the model's decision-making process and correct model errors attributable to limited training data.
The total size of the zip file is 75.3 MB. The .csv, Comma Separated Value, file is a simple format that is designed for a database table and supported by many applications. The .csv file is often used for moving tabular data between two different computer programs, due to its open format. The most common software used to open .csv files are Microsoft Excel and RecordEditor, (for more information on .csv files and software, please visit https://www.file-extensions.org/csv-file-extension). The .txt file type is a common text file, which can be opened with a basic text editor. The most common software used to open .txt files are Microsoft Windows Notepad, Sublime Text, Atom, and TextEdit (for more information on .txt files and software, please visit https://www.file-extensions.org/txt-file-extension). The Portable Document Format (PDF) file format was developed by Adobe Systems and represents two-dimensional documents in a device-independent and resolution-independent format. There are PDF readers available on many platforms, such as Xpdf, Foxit, and Adobe's own Adobe Acrobat Reader. PDF readers/viewers or online services for basic functions are generally free (for more information on .pdf files and software, please visit https://www.file-extensions.org/pdf-file-extension).
-
Content Notes:Data Scope:
We use the publicly available INTERACTION dataset. The INTERACTION dataset is collected using drones on fixed road segments in the USA, Germany, and China. The dataset provides a lanelet2 format map and a set of time-indexed trajectories of the positions, velocities, and headings of each vehicle in the scene in the map's coordinate system at a sampling frequency of 10 Hz, and the vehicle's length and width for each road segment.
Data Specification:
Complete description of the INTERACTION dataset can be found on the official website: https://interaction-dataset.com/. This is the location of the original, public dataset. For this project, the team used this dataset, pre-processed the data, and created their model. This pre-processed data by the team is the data on the Dataverse page.
Model: To view and download the simulation model, visit the project's GitHub: https://github.com/ran-weii/interactive_inference
National Transportation Library (NTL) Curation Note: As this dataset is preserved in a repository outside U.S. DOT control, as allowed by the U.S. DOT’s Public Access Plan (https://doi.org/10.21949/1503647) Section 7.4.2 Data, the NTL staff has performed NO additional curation actions on this dataset. This dataset has been curated to CoreTrustSeal's curation level "C. Initial Curation." To find out more information on CoreTrustSeal's curation levels, please consult their "Curation & Preservation Levels" CoreTrustSeal Discussion Paper" (https://doi.org/10.5281/zenodo.8083359). NTL staff last accessed this dataset at its repository URL on 2024-05-06. If, in the future, you have trouble accessing this dataset at the host repository, please email NTLDataCurator@dot.gov describing your problem. NTL staff will do its best to assist you at that time.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type: