Department of Civil and Environmental Engineering, Carnegie Mellon ...
Fuel Economy Testing of Autonomous Vehicles Avi Mersky ([email protected]), Constantine Samaras Department of Civil and Environmental Engineering, Carnegie Mellon University Automated Vehicles Symposium 2015, MI, Ann Arbor, July 20-23, 2015 Summary
Autonomous and Connected Driving Rulesets Simulated
• A method was derived for testing the fuel economy of Autonomous and Connected Vehicles (AVs & CVs), using the existing EPA test procedure • This simulations showed fuel economy differences for the AV/CV tests from 3% to 10%, compared to the current EPA testing procedure
•
Fully Autonomous •
HeadwayACC
• Fuel economy testing is required for CAFE compliance, emissions standards and fuel economy labeling and advertising • The same set of tests is used for all three, with different weights, and is standard for all vehicles and features • Partial AV/CV systems are already commercially available • When these features are used, they change the fuel economy characteristics of the vehicle in a way that can not be advertised, utilized or incentivized
•
Velocity Schedule of the EPA HWFET (Freeway) Drive Cycle Velocity (km/hr)
Velocity (km/h)
120 100 80 60 40 20 0 0
500
1000
1500
0
100
Time (s)
200
300
𝑏𝑜𝑢𝑛𝑑 𝑥, 𝑦, 𝑧
•
•
400
500
600
700
800
Time (s)
Simulated AV/CV Testing Addition
(1+𝑠𝑡𝑒𝑝2∗ℎ𝑒𝑎𝑑𝑤𝑎𝑦𝑡𝑎𝑟𝑔𝑒𝑡) 𝑣𝑙𝑒𝑎𝑑−𝑣𝑠𝑒𝑙𝑓 𝑠𝑝𝑎𝑐𝑒 𝑏𝑢𝑓𝑓𝑒𝑟
Normal Maximum Acceleration Deceleration Bounds (m/s2) for Safety (m/s2)
HeadwayACC 1 +/- 2 HeadwayACC 2 +1 / -1.5 𝑎𝑡𝑎𝑟𝑔𝑒𝑡 = 0.35 ∗= max min 𝑥, 𝑦 , 𝑧 HeadwayACC 3 +/- 1 𝑖𝑓 𝑎𝑠𝑎𝑓𝑒 ≤ 𝑎𝑡𝑎𝑟𝑔𝑒𝑡 𝑎𝑛𝑑 𝑎𝑠𝑎𝑓𝑒 ≤ 0: • 𝑡ℎ𝑒𝑛 𝑎𝑛𝑒𝑤 = 𝑏𝑜𝑢𝑛𝑑(𝑎𝑠𝑎𝑓𝑒,+𝑎𝑏𝑜𝑢𝑛𝑑,−𝑎𝑏𝑜𝑢𝑛𝑑) VelocityACC 1 +/- 2 • 𝑒𝑙𝑠𝑒: 𝑎𝑛𝑒𝑤 = 𝑏𝑜𝑢𝑛𝑑(𝑎𝑡𝑎𝑟𝑔𝑒𝑡,+𝑎𝑏𝑜𝑢𝑛𝑑,−𝑎𝑏𝑜𝑢𝑛𝑑) VelocityACC 2 +/- 1.5 VelocityACC 3 +/- 1 Speed Control PlannedACC 1 +/- 2 • 𝑣𝑒 = 𝑣𝑠𝑒𝑙𝑓 − 𝑣𝑑 PlannedACC 2 +/- 2 • 𝑎 = 𝑎𝑠𝑐 = 𝑏𝑜𝑢𝑛𝑑 −0.4 ∗ 𝑣𝑒, 𝑎𝑀𝑎𝑥, 𝑎𝑀𝑖𝑛 𝑠𝑡𝑒𝑝2
+
𝑠𝑡𝑒𝑝
−
-2 -2 -2 -2 -2 -2 N/A N/A
𝑠𝑡𝑒𝑝2
Plan Planning Ahead Interval Time (s) (s)
Target Headway (s)
Minimum Headway (s)
Minimum Safe Distance (m)
N/A N/A N/A N/A N/A N/A 3 5
3 3 3 3 3 3 N/A N/A
N/A N/A N/A N/A N/A N/A 1 1
1 1 1 1 1 1 5 5
N/A N/A N/A N/A N/A N/A 2 3
Gap Control • • •
𝑠𝑑 = 𝑇𝑑 ∗ 𝑣𝑠𝑒𝑙𝑓 𝑠𝑒 = 𝑠 − 𝑠𝑑 𝑎 = 𝑏𝑜𝑢𝑛𝑑 𝑠 + 0.25 ∗ 𝑠𝑒, 𝑎𝑠𝑐, 𝑎𝑀𝑖𝑛
Connected Autonomous Proxy •
PlannedACC • •
• 5 separate pre-set drive cycles, with specified velocity schedules • Tested on a dynamometer (vehicular treadmill) with fuel consumption measured from the tailpipe (except for EVs where SOC is used) • These are then weighted for regulation and rating purposes
100 90 80 70 60 50 40 30 20 10 0
•
VelocityACC •
Current Fuel Economy Testing Method
Velocity Schedule of the EPA FTP (Urban) Drive Cycle
𝑎𝑠𝑎𝑓𝑒 ≥
• •
Background
𝑠𝑝𝑎𝑐𝑒+(𝑣𝑙𝑒𝑎𝑑−𝑣𝑠𝑒𝑙𝑓)∗𝑠𝑡𝑒𝑝 ∗ℎ𝑒𝑎𝑑𝑤𝑎𝑦𝑡𝑎𝑟𝑔𝑒𝑡−𝑣𝑠𝑒𝑙𝑓
•
Rule Set Parameters Tested
Given perfect information about the lead vehicle’s location for the next X seconds find the maximum acceleration that does not take the following vehicle past 1 second of headway at any time over the next X seconds Repeat every Y seconds • Y<X-1
Fuel Economy Estimation •
Fuel consumption was estimated using the Virginia Tech Comprehensive Fuel Consumption Model for several different vehicles 2010 Honda Accord: Fuel Economy Estimation Results Sample FTP Urban Cycle (22)* HWFET Freeway Cycle (31)* HeadwayACC City 1 HeadwayACC Freeway 1 VelocityACC City 2 VelocityACC Freeway 2 PlannedACC City 2 PlannedACC Freeway 2
Simulated Fuel Economy (MPG) 25.1 43.3 25.6 43.1 25.3 43 26.6 43.9
% Change from EPA N/A N/A 2% 0% 1% -1% 6% 1%
*Rated fuel economies are notably lower than simulated, due to usage of the extra 3 cycles for the rated fuel economies
First the ruleset the AV will follow will be abstracted to function in a simulation, with only information pertaining to its own performance and visible information about the car in front of it. Lateral control will be ignored. The road will be assumed to be straight, single lane, and level, with only two vehicles and no traffic control systems. The simulated vehicle will start 5 meters behind another “lead vehicle”. At time 0 the lead vehicle will start to obey the EPA drive cycle for either FTP or HWFET conditions. The simulated AV will then make decisions about how to best follow the lead vehicle until the end of the test. The test will end at the completion of the EPA cycle, when the lead vehicle has stopped, regardless of whether or not the AV has stopped. The velocity profiles for both the Urban and Freeway simulations will then be recorded. The simulation results can be audited by a physical experiment, with the lead vehicle following the EPA drive cycles. These “derived drive cycles” could then be used in standard dynamometer testing as a supplemental tests, to be weighed in the official fuel economy ratings
Conclusion
Future Work
• Autonomous and connected driving features can have a significant effect on fuel economy • Current testing methodologies cannot captures this effect • As some of these features already exist on mass market vehicles evaluating them for policy decisions is important • A standardized method for deriving drive cycles for dynamotor testing from the current drive cycles can be used to capture some of the effects of autonomous and connected features
• More advanced rulesets, with less simplifications • Testing protocols for situation specific features (such as cooperative adaptive cruise control platoons) • More vehicles and drivetrain variety • Simulation engines that are able to capture the fuel economies of non-conventional drive train types (electric & advanced stop-star) • Aggregate effects
Acknowledgments This work was supported in part by the Dwight David Eisenhower Transportation Fellowship Program, and by the Center for Climate and Energy Decision Making (SES-0949710) through a cooperative agreement between the NSF and Carnegie Mellon University. We would also like to thank Professor Hesham Rakha, his students, and the team that created the Virginia Tech Comprehensive Fuel Consumption Model, for allowing its use and providing helpful comments.
Autonomous and Connected Driving Rulesets Simulated
• A method was derived for testing the fuel economy of Autonomous and Connected Vehicles (AVs & CVs), using the existing EPA test procedure • This simulations showed fuel economy differences for the AV/CV tests from 3% to 10%, compared to the current EPA testing procedure
•
Fully Autonomous •
HeadwayACC
• Fuel economy testing is required for CAFE compliance, emissions standards and fuel economy labeling and advertising • The same set of tests is used for all three, with different weights, and is standard for all vehicles and features • Partial AV/CV systems are already commercially available • When these features are used, they change the fuel economy characteristics of the vehicle in a way that can not be advertised, utilized or incentivized
•
Velocity Schedule of the EPA HWFET (Freeway) Drive Cycle Velocity (km/hr)
Velocity (km/h)
120 100 80 60 40 20 0 0
500
1000
1500
0
100
Time (s)
200
300
𝑏𝑜𝑢𝑛𝑑 𝑥, 𝑦, 𝑧
•
•
400
500
600
700
800
Time (s)
Simulated AV/CV Testing Addition
(1+𝑠𝑡𝑒𝑝2∗ℎ𝑒𝑎𝑑𝑤𝑎𝑦𝑡𝑎𝑟𝑔𝑒𝑡) 𝑣𝑙𝑒𝑎𝑑−𝑣𝑠𝑒𝑙𝑓 𝑠𝑝𝑎𝑐𝑒 𝑏𝑢𝑓𝑓𝑒𝑟
Normal Maximum Acceleration Deceleration Bounds (m/s2) for Safety (m/s2)
HeadwayACC 1 +/- 2 HeadwayACC 2 +1 / -1.5 𝑎𝑡𝑎𝑟𝑔𝑒𝑡 = 0.35 ∗= max min 𝑥, 𝑦 , 𝑧 HeadwayACC 3 +/- 1 𝑖𝑓 𝑎𝑠𝑎𝑓𝑒 ≤ 𝑎𝑡𝑎𝑟𝑔𝑒𝑡 𝑎𝑛𝑑 𝑎𝑠𝑎𝑓𝑒 ≤ 0: • 𝑡ℎ𝑒𝑛 𝑎𝑛𝑒𝑤 = 𝑏𝑜𝑢𝑛𝑑(𝑎𝑠𝑎𝑓𝑒,+𝑎𝑏𝑜𝑢𝑛𝑑,−𝑎𝑏𝑜𝑢𝑛𝑑) VelocityACC 1 +/- 2 • 𝑒𝑙𝑠𝑒: 𝑎𝑛𝑒𝑤 = 𝑏𝑜𝑢𝑛𝑑(𝑎𝑡𝑎𝑟𝑔𝑒𝑡,+𝑎𝑏𝑜𝑢𝑛𝑑,−𝑎𝑏𝑜𝑢𝑛𝑑) VelocityACC 2 +/- 1.5 VelocityACC 3 +/- 1 Speed Control PlannedACC 1 +/- 2 • 𝑣𝑒 = 𝑣𝑠𝑒𝑙𝑓 − 𝑣𝑑 PlannedACC 2 +/- 2 • 𝑎 = 𝑎𝑠𝑐 = 𝑏𝑜𝑢𝑛𝑑 −0.4 ∗ 𝑣𝑒, 𝑎𝑀𝑎𝑥, 𝑎𝑀𝑖𝑛 𝑠𝑡𝑒𝑝2
+
𝑠𝑡𝑒𝑝
−
-2 -2 -2 -2 -2 -2 N/A N/A
𝑠𝑡𝑒𝑝2
Plan Planning Ahead Interval Time (s) (s)
Target Headway (s)
Minimum Headway (s)
Minimum Safe Distance (m)
N/A N/A N/A N/A N/A N/A 3 5
3 3 3 3 3 3 N/A N/A
N/A N/A N/A N/A N/A N/A 1 1
1 1 1 1 1 1 5 5
N/A N/A N/A N/A N/A N/A 2 3
Gap Control • • •
𝑠𝑑 = 𝑇𝑑 ∗ 𝑣𝑠𝑒𝑙𝑓 𝑠𝑒 = 𝑠 − 𝑠𝑑 𝑎 = 𝑏𝑜𝑢𝑛𝑑 𝑠 + 0.25 ∗ 𝑠𝑒, 𝑎𝑠𝑐, 𝑎𝑀𝑖𝑛
Connected Autonomous Proxy •
PlannedACC • •
• 5 separate pre-set drive cycles, with specified velocity schedules • Tested on a dynamometer (vehicular treadmill) with fuel consumption measured from the tailpipe (except for EVs where SOC is used) • These are then weighted for regulation and rating purposes
100 90 80 70 60 50 40 30 20 10 0
•
VelocityACC •
Current Fuel Economy Testing Method
Velocity Schedule of the EPA FTP (Urban) Drive Cycle
𝑎𝑠𝑎𝑓𝑒 ≥
• •
Background
𝑠𝑝𝑎𝑐𝑒+(𝑣𝑙𝑒𝑎𝑑−𝑣𝑠𝑒𝑙𝑓)∗𝑠𝑡𝑒𝑝 ∗ℎ𝑒𝑎𝑑𝑤𝑎𝑦𝑡𝑎𝑟𝑔𝑒𝑡−𝑣𝑠𝑒𝑙𝑓
•
Rule Set Parameters Tested
Given perfect information about the lead vehicle’s location for the next X seconds find the maximum acceleration that does not take the following vehicle past 1 second of headway at any time over the next X seconds Repeat every Y seconds • Y<X-1
Fuel Economy Estimation •
Fuel consumption was estimated using the Virginia Tech Comprehensive Fuel Consumption Model for several different vehicles 2010 Honda Accord: Fuel Economy Estimation Results Sample FTP Urban Cycle (22)* HWFET Freeway Cycle (31)* HeadwayACC City 1 HeadwayACC Freeway 1 VelocityACC City 2 VelocityACC Freeway 2 PlannedACC City 2 PlannedACC Freeway 2
Simulated Fuel Economy (MPG) 25.1 43.3 25.6 43.1 25.3 43 26.6 43.9
% Change from EPA N/A N/A 2% 0% 1% -1% 6% 1%
*Rated fuel economies are notably lower than simulated, due to usage of the extra 3 cycles for the rated fuel economies
First the ruleset the AV will follow will be abstracted to function in a simulation, with only information pertaining to its own performance and visible information about the car in front of it. Lateral control will be ignored. The road will be assumed to be straight, single lane, and level, with only two vehicles and no traffic control systems. The simulated vehicle will start 5 meters behind another “lead vehicle”. At time 0 the lead vehicle will start to obey the EPA drive cycle for either FTP or HWFET conditions. The simulated AV will then make decisions about how to best follow the lead vehicle until the end of the test. The test will end at the completion of the EPA cycle, when the lead vehicle has stopped, regardless of whether or not the AV has stopped. The velocity profiles for both the Urban and Freeway simulations will then be recorded. The simulation results can be audited by a physical experiment, with the lead vehicle following the EPA drive cycles. These “derived drive cycles” could then be used in standard dynamometer testing as a supplemental tests, to be weighed in the official fuel economy ratings
Conclusion
Future Work
• Autonomous and connected driving features can have a significant effect on fuel economy • Current testing methodologies cannot captures this effect • As some of these features already exist on mass market vehicles evaluating them for policy decisions is important • A standardized method for deriving drive cycles for dynamotor testing from the current drive cycles can be used to capture some of the effects of autonomous and connected features
• More advanced rulesets, with less simplifications • Testing protocols for situation specific features (such as cooperative adaptive cruise control platoons) • More vehicles and drivetrain variety • Simulation engines that are able to capture the fuel economies of non-conventional drive train types (electric & advanced stop-star) • Aggregate effects
Acknowledgments This work was supported in part by the Dwight David Eisenhower Transportation Fellowship Program, and by the Center for Climate and Energy Decision Making (SES-0949710) through a cooperative agreement between the NSF and Carnegie Mellon University. We would also like to thank Professor Hesham Rakha, his students, and the team that created the Virginia Tech Comprehensive Fuel Consumption Model, for allowing its use and providing helpful comments.
