Autonomous Delivery Vehicle as a Disruptive Technology: How to Shape the Future with a Focus on Safety? (05-087)

Project Description: The goal of this project was to perform safety implication analysis and safety impact analysis of ADVs. This project used five different datasets to perform the analysis. ADV related operation design domain (ODD) scenarios were determined to examine the real-world collision data. This study generates a total of 80 association rules with high likelihood measures for these datasets. The rules can be used as the prospective benchmark rules to examine how these rule-based risk patterns can be replaced by ADVs by eliminating human-driven grocery related trips. The research team used five separate databases to perform the analysis: Fatality Analysis Reporting System (FARS) [2016-2020] Crash Report Sampling System (CRSS) [2016-2020] California AV and ADV collision Data [2014-2020] Waymo Open Data Third party ADV Trajectory Data (sample data) Data Scope: FARS ADV ODD Scenario dataset (FARS_ADV_ODD_Scenario_VehicleLevel_2016_2020.csv) was used to determine the benchmark rules from real-world FARS data. California AV collision associated with delivery inclusive AV companies (California_Collisions_Delivery_Inclusive_AV_Companies_2014_2021.csv) was used to determine the benchmark rules from real-world collision data. Third-party ADV trajectory sample data (ThirdParty_ADV_Trajectory_2021_Jan_Jun.xlsx) with top 99 trips are listed with event information statistics (mean and rank based on mean values) for hard brake, over speeding, and excessive over speeding. Data Specification: See Table 1-3 for description of each variable included in the datasets.