C AR MA: Towards Personalized Automotive Tuning Tobias Flach 1, Nilesh Mishra 1, Luis Pedrosa 1, Christopher Riesz 2, Ramesh Govindan 1 1 University of Southern California 2 Rutgers University, Camden Campus Sensys 2011

2 Motivation Architecture Evaluation Conclusion Introduction Networked Sensing  Smart Grid  Smart Building Smartphones  Context awareness  Participatory sensing Automobiles  On-board sensing  Personalized tuning Introduction

3 Motivation Architecture Evaluation Conclusion Introduction Sensing and Control in an Automobile Introduction Automobiles are highly customizable but this capability is not exposed to the user

4 Motivation Architecture Evaluation Conclusion Introduction Current Practice: Fixed Control Parameters Driving behaviorTerrain TrafficWeather Introduction Factory defaults cannot cover all possible conditions

5 Motivation Architecture Evaluation Conclusion Introduction Our Goal: Flexible Personalized Tuning Bob Alice Introduction CarMA can optimize for individual trips

6 Motivation Architecture Evaluation Conclusion Introduction Can Personalized Tuning Be Effective? Impact of route characteristics? Impact of driving behavior? Interstate RouteHill RouteStreet route Motivation

7 Architecture Evaluation Conclusion Introduction Can Personalized Tuning Be Effective? Fuel economy for two cars on different routes (with different conditions) Intrinsic variability Extrinsic variability Exploitable by personalized tuning Motivation

8 Architecture Evaluation Conclusion Introduction Client device On-board diagnostics (OBD-II) Engine control unit (ECU) Background: Scanning and Tuning SensorValue Intake manifold absolute pressure32.0 kPa RPM1925 rev/min Speed16.0 km/h Ignition timing advance3.0 °(relative) Throttle position27.84 % Injector pulse width11.89 ms Coolant Temperature Transmission not engaged Transmission engaged -20°C1200 0°C °C °C Table: Desired Idle RPM Motivation Scanning…Tuning…

9 Motivation Architecture Evaluation Conclusion Introduction Contributions: CarMA Fosters ecosystem of car modifications Wireless connectivity Programmability Motivation C AR MA: A smartphone-based, make and model-independent platform providing programming abstractions for scanning and tuning On-Board Sensing

10 Motivation Architecture Evaluation Conclusion Introduction Contributions: Sample Apps Improve responsiveness for hilly route Maximize fuel efficiency Support user diversity Enforce safer driving behavior Motivation Implementation of sample apps that allow non-expert users to personalize and automatically optimize cars for different drivers and/or route conditions

11 Motivation Architecture Evaluation Conclusion Introduction Contributions: Benefits of Personalized Tuning A demonstration of personalized tuning that can exploit extrinsic variability to achieve significant fuel efficiency or responsiveness gains Motivation

12 Motivation Architecture Evaluation Conclusion Introduction C AR MA Architecture Architecture

13 Motivation Architecture Evaluation Conclusion Introduction Scanning Subsystem Architecture C AR MA Request: vehicle speed sensor Request: A F1 01 0D Response: B D 28 Response: Vehicle Speed Sensor = 40 km/h Vehicle Heterogeneity Protocol Optimizations Interface Heterogeneity

14 Motivation Architecture Evaluation Conclusion Introduction 2 2 Tuning Subsystem Architecture C AR MA ECU memory Download routine ECU firmware OBD-II interface prepare(); setRPMLimit(); commit(); setRPMLimit(); commit(); Reverse Engineering Level of Abstraction

15 Motivation Architecture Evaluation Conclusion Introduction ModResponseEconomySafety DisableEGR ✔✔ EnrichMixture ✔✔ IncreaseSparkTiming ✔✔ AdjustShiftDownPoints ✔✔✔ LimitRPM ✔✔✔ DecreaseShiftTime ✔ Tuning Subsystem: ModAPI Client codeC AR MA code tuner = ATuner.Stub.asInterface(service); tuner.loadCurrentSettings() tuner.applyMod(ModNames.LimitRPM, rpmVal) tuner.commitChanges() tuner = ATuner.Stub.asInterface(service); tuner.loadCurrentSettings() tuner.applyMod(ModNames.LimitRPM, rpmVal) tuner.commitChanges() new TableSetMod( ECUTableName.Disable_All_Injectors_High_MPH, 0, 0, 0, 2, 255), new TableSetMod( ECUTableName.Disable_All_Injectors_High_MPH, 1, 1, 0, 2, limit + 1), new TableSetMod( ECUTableName.Disable_1_Injector_High_MPH, limit), new TableSetMod( ECUTableName.Disable_2_Injectors_High_MPH, limit) new TableSetMod( ECUTableName.Disable_All_Injectors_High_MPH, 0, 0, 0, 2, 255), new TableSetMod( ECUTableName.Disable_All_Injectors_High_MPH, 1, 1, 0, 2, limit + 1), new TableSetMod( ECUTableName.Disable_1_Injector_High_MPH, limit), new TableSetMod( ECUTableName.Disable_2_Injectors_High_MPH, limit) Architecture tuner = ATuner.Stub.asInterface(service); tuner.loadCurrentSettings() tuner.applyMod(ModNames.LimitRPM, rpmVal) tuner.commitChanges() tuner = ATuner.Stub.asInterface(service); tuner.loadCurrentSettings() tuner.applyMod(ModNames.LimitRPM, rpmVal) tuner.commitChanges() new TableSetMod( ECUTableName.Disable_All_Injectors_High_MPH, 0, 0, 0, 2, 255), new TableSetMod( ECUTableName.Disable_All_Injectors_High_MPH, 1, 1, 0, 2, limit + 1), new TableSetMod( ECUTableName.Disable_1_Injector_High_MPH, limit), new TableSetMod( ECUTableName.Disable_2_Injectors_High_MPH, limit) new TableSetMod( ECUTableName.Disable_All_Injectors_High_MPH, 0, 0, 0, 2, 255), new TableSetMod( ECUTableName.Disable_All_Injectors_High_MPH, 1, 1, 0, 2, limit + 1), new TableSetMod( ECUTableName.Disable_1_Injector_High_MPH, limit), new TableSetMod( ECUTableName.Disable_2_Injectors_High_MPH, limit)

16 Motivation Architecture Evaluation Conclusion Introduction Evaluation Goals Does CarMA raise the level of abstraction?  Implementation of sample applications  Micro-benchmarks Does CarMA provide significant fuel efficiency gains?  Test runs using route-specific fuel efficiency optimizations on five different routes  Comparison of the recorded sensor data from baseline and modified runs Evaluation

17 Motivation Architecture Evaluation Conclusion Introduction Evaluation: Applications Scan App Tune WizardRoute Evaluator Valet ModeDriver Customizer 98 LOC 214 LOC 408 LOC 60 LOC 76 LOC Evaluation

18 Motivation Architecture Evaluation Conclusion Introduction Evaluation: Tune Wizard Evaluation Alice Enable EGR Increase torque timing Limit RPM to 5000 Enable EGR Increase torque timing Limit RPM to 5000 Limit RPM to 3000

19 Motivation Architecture Evaluation Conclusion Introduction Evaluation: Fuel Efficiency Gains Experimented with multiple drivers Recorded data for more than 1100 miles of driving Recorded data for baseline runs and runs using modifications Evaluation Speed limit Elevation gain Max. gradeCongestion Traffic lights

20 Motivation Architecture Evaluation Conclusion Introduction Experimental Results: Fuel Economy Fuel economy improved by 11% on average Fuel economy for one driver Evaluation Would save more than $18B in the US annually Trip

21 Motivation Architecture Evaluation Conclusion Introduction Experimental Results: Validating Applications RPM vs. throttle pos. for DriverCustomizer Aggressive driving behavior is prevented Aggressive driving behavior is prevented Evaluation

22 Motivation Architecture Evaluation Conclusion Introduction Related Work Evaluation GreenGPS Sensing ControlAnalysis Mechanics Firmware Upgrade Torque Powr Tuner eco:Drive C AR MA

23 Motivation Architecture Evaluation Conclusion Introduction Conclusions CarMA provides programming abstractions for scanning and tuning automobiles via mobile devices Usable by non-experts for personalized tuning CarMA apps can achieve significant fuel efficiency gains by inhibiting aggressive driving behavior and enforcing limits Conclusion

24 Motivation Architecture Evaluation Conclusion Introduction Future Work Extending CarMA for a broader range of makes and models Adding safety mechanisms to prevent accidental or malicious damage to vehicles using CarMA Conclusion

C AR MA: Towards Personalized Automotive Tuning Tobias Flach 1, Nilesh Mishra 1, Luis Pedrosa 1, Christopher Riesz 2, Ramesh Govindan 1 1 University of Southern California 2 Rutgers University, Camden Campus Sensys 2011

26 Motivation Architecture Evaluation Conclusion Introduction Improving the State of the Art Complexity Expressivity Fuel economy / Performance modes Specialized tuning software C AR MA

27 Motivation Architecture Evaluation Conclusion Introduction Implemented functionality  Checks for conflicting modifications  Definition of safe value ranges for parameters Additional ideas  Application certification  Mandatory "kill switch" Security Considerations Safe range Value range (Desired Idle RPM; single cell)

28 Motivation Architecture Evaluation Conclusion Introduction Experimental Results: Economy variability Impact of mods depends on “old” driving behavior Fuel economy for different drivers Evaluation

29 Motivation Architecture Evaluation Conclusion Introduction Experimental results: Speed and RPM limits Speed limit is enforced RPM long tail is cut Speed on trip 3RPM CDF for trip 4B

30 Motivation Architecture Evaluation Conclusion Introduction Background: Scanning & Tuning Motivation

31 Motivation Architecture Evaluation Conclusion Introduction OBD-II Port