United States Department of Transportation
i
Risk Analysis of Autonomous Vehicles in Mixed Traffic Streams
-
2017-05-01
[PDF-3.32 MB]
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Report, 6/1/15 - 5/31/17
-
Corporate Publisher:
-
Abstract:The objective of this study was to identify the risks associated with the failure of autonomous vehicles in mixed traffic streams and develop strategies to minimize these risks. Three distinct and interconnected phases were used to conduct the risk analysis; i) risk identification, ii) risk estimation and iii) evaluation. To identify the risks, the autonomous vehicle system was first disintegrated into vehicular components (i.e., sensors, actuators and communication platforms). Because an autonomous vehicle will share the roadways with human drivers for many years after their deployment, transportation infrastructure components play an important role in the final risk analysis. A fault tree model was developed for each vehicular component failure and each transportation infrastructure component failure. The failure probabilities of each component were estimated by reviewing relevant literature and publicly available data. The fault tree analysis revealed the autonomous vehicle failure probability to be about 14% resulting from a sequential failure of vehicular components (i.e., particularly those responsible for automation) in the vehicle’s lifetime. Subsequently, the failure probability due to autonomous vehicle components and due to transportation infrastructure components were combined. An overall failure probability of 158 incidents per 1 million miles of travel was determined possible as a result of malfunctions or disruptions in vehicular or infrastructure components, respectively. To validate the results, real-world data from the California Department of Motor Vehicles autonomous vehicle testing records were utilized in this study. The most critical combinations of events that could lead to failure of autonomous vehicles, known as minimal cut-sets, were also identified and ranked based on their corresponding failure probabilities. Based on the fault tree analysis, 22 strategies were identified that would minimize the failure probability of autonomous vehicles. Finally, these identified strategies were evaluated using benefit-cost analysis.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type:
