A Risk Analysis Uses the Probability of Failure of AVs in Mixed Traffic to Study Potential Component Failure Risks Such as Hardware Breakdown or Weather.
Nationwide, United States
Risk Analysis of Autonomous Vehicles in Mixed Traffic Streams
Summary Information
This study estimated risks associated with the failure of autonomous vehicles in mixed traffic streams to develop mitigation strategies. The three components of the analysis included risk identification, risk estimation and evaluation. Since autonomous vehicles will share the roadways with human drivers for a certain number of years after their deployment, transportation infrastructure is an important component of the final risk analysis. A graphical tool in the form of a fault tree model was developed to explore the causes of system level failures for each vehicular component failure (i.e., sensors, actuators and communication platforms) and each transportation infrastructure component failure. The failure probabilities were estimated by a literature review, publicly available information, and survey results from subject matter experts. Based on the cut sets (hierarchical sequence of events) from the fault tree analysis, 22 strategies that would minimize the failure probability of autonomous vehicles were identified and evaluated using benefit-cost (B/C) analysis.
Methodology
This study considered three performance measures from the 22 risk mitigation strategies: 1) number of traffic deaths, 2) level of congestion, and 3) environmental emissions. The current costs of the back-up sensors and DSRC devices used for the benefit-cost analysis were collected from research, analysis and expert interviews conducted by Boston Consulting Group (see report).. Net present values (NPV) of costs and benefits with a discount rate of 6.5% was used for the benefit-cost analysis. All traditional vehicles were assumed to be replaced by autonomous vehicles by the year of 2050 and a regression model based on the crash data of the past 50 years was developed to estimate future crash rates.
Four major exhaust pollutants emitted from tailpipe were considered in this analysis—carbon dioxide (CO2), nitrogen oxides (NOX), volatile organic compounds (VOC), and particulate matter (PM10).
The analysis focused on two measures to minimize failures: 1) installation of additional sensors (e.g., Lidar) as back-up, and 2) a regulation requiring Dedicated Short Range Communications (DSRC) device installation. Table 1 shows a benefit-cost analysis for the year 2030 with a market penetration of 10 Percent Autonomous Vehicles.
Table 1: Benefit-Cost Analysis for 2030 with a Market Penetration of 10 Percent Autonomous Vehicles
|
Sensors/ Devices |
Lives Saved |
Values of Lives Saved (in million $) |
Value of Travel Time Saved (in million $) |
Value of Emission Reduction (in million $) |
Total Benefits (in million $) |
Net Present Value of Benefits (in million $) |
Net Present Value of Total Costs (in million $) |
B/C Ratio |
|
Lidar |
284.87 |
1766.18 |
356.13 |
144.15 |
2266.46 |
20429.47 |
247520.00 |
0.08 |
|
Radar |
165.26 |
1024.64 |
356.13 |
144.15 |
1521.91 |
13745.24 |
4643.66 |
2.96 |
|
Video Camera |
208.90 |
1295.19 |
356.13 |
144.15 |
1795.47 |
16184.0 |
6188.00 |
2.62 |
|
GPS Device |
149.30 |
925.63 |
356.13 |
144.15 |
1425.90 |
12852.88 |
185640.00 |
0.07 |
|
Wheel Encoder |
194.98 |
1208.85 |
356.13 |
144.15 |
1709.12 |
15405.74 |
5346.43 |
2.88 |
|
DSRC |
108.29 |
671.43 |
356.13 |
144.15 |
1171.70 |
10561.53 |
10829.00 |
0.97 |
Table 2 shows a benefit-cost analysis for the year 2050 with a market penetration of 100 percent Autonomous Vehicles.
Table 2: Benefit-Cost Analysis for 2050 with a Market Penetration of 100 Percent Autonomous Vehicles
|
Sensors/ Devices |
Lives Saved |
Values of Lives Saved (in million $) |
Value of Travel Time Saved (in million $) |
Value of Emission Reduction (in million $) |
Total Benefits (in million $) |
Net Present Value of Benefits (in million $) |
Net Present Value of Total Costs (in million $) |
B/C Ratio |
|
Lidar |
1603 |
9,936.68 |
10,683.75 |
432.45 |
21,052.88 |
285,826.66 |
2,891,200 |
0.09 |
|
Radar |
930 |
5,764.71 |
10683.75 |
432.45 |
16,880.91 |
229,185.49 |
54,180.97 |
4.23 |
|
Video Camera |
1175 |
7,286.85 |
10683.75 |
432.45 |
18,403.05 |
249,851.03 |
72,280.00 |
3.46 |
|
GPS Device |
840 |
5,207.67 |
10683.75 |
432.45 |
16,323.87 |
221,622.78 |
2,168,400.00 |
0.10 |
|
Wheel Encoder |
1097 |
6,801.06 |
10683.75 |
432.45 |
17,917.26 |
243,255.64 |
43,368.00 |
5.61 |
|
DSRC |
60 |
3,777.51 |
10683.75 |
432.45 |
14,893.70 |
202,205.97 |
126,490.00 |
1.60 |
Key Takeaway
The results of the benefits-costs analysis indicate that installing radar, video cameras, or a wheel encoder as back-up sensors could be cost effective (B/C ratio > 1) even with only 10 percent market penetration rate for autonomous vehicles. Similarly, installing DSRC devices could also be beneficial if the market penetration rate increases to 100 percent in 2050.
