Report an accessibility problem

Engineering | Undergraduate Research

Automated Vehicle Driving Score for Selected Scenarios

, , , , ,
Automated Vehicle Driving Score for Selected Scenarios

Automated driving system-equipped vehicles (AVs) can improve roadway safety through better execution of driving plans, reducing risks intrinsic to human drivers such as impulsive speeding and distracted or intoxicated driving. Another area with potential for promoting safer and more efficient integration of AVs into the road transportation system is to help design safe and effective driving policies so that the driving plans are themselves safe.

The goal of this project is to create a driving score for vehicles in scenarios that are relatively common in urban driving. The output of this project will be a metric or driving score for the specific scenario that can be used to rank drivers from the most aggressive to the most conservative. For example, in a car-following scenario, one might use the time headway measurement to differentiate between drivers who are aggressive (short time headways) and those who are more conservative (long time headways). Further, one might conduct an analysis of a naturalistic data set to determine the distribution of drivers, as described by their time headway, to determine what time headway an autonomous vehicle might maintain.

In this project, the student will work with Dr. Jeffrey Wishart, the sponsoring professor as well as industry partner Motional to license and make available datasets from LevelX to conduct the aforementioned analysis. LevelX uses drones to observe interesting urban driving from above. Each dataset focuses on one or a few scenarios (e.g., unsignalized intersections) and annotates road actors and their positions in machine-readable form. Each dataset is hours long and consists mostly of continuous observation.

The student will analyze one of the datasets and use one or more metrics to sort the interactions from most aggressive to least, and use that information to set a driving baseline for AVs. The choice of dataset as well as the metric(s) to be employed will be made through consultation between the sponsoring professor, Motional, and the student, and will draw on the work of the Institute of Automated Mobility (IAM), SAE On-Road Automated Driving (ORAD) Committee’s V&V Task Force (the sponsoring professor is lead on the IAM Metrics Project and chair of the V&V Task Force), and other sources such as the Automated Vehicle Safety Consortium (AVSC).

Spring 2022, FURI/MORE eligible, 10 hours/week

Contact Jeffrey Wishart (




Posted on