Post on 11-Jul-2020
AVL List GmbH (Headquarters)
Public
Energy-Efficient Cooperative
Adaptive Cruise ControlDr. Stephen J. Jones, AVL List GmbHstephen.jones@avl.com +43 664 850 9172
N. Wikström, A. Ferreira Parrilla, R. Patil, E. Kural, A. Massoner, AVL List GmbH
A. Grauers, Chalmers University of Technology
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 2Public
Energy-Efficient CACCContents
1. Introduction to Predictive Energy Management
2. Traffic Light Assistant
3. Energy-Efficient Cooperative Adaptive Cruise Control
a) Problem Overview
b) Model Predictive Control
c) Results
4. Summary & Conclusion
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 3Public
Introduction4 Pillars of ADAS/AD Engineering Services
Trusted Engineering Service Provider & Development Partner at ADAS & Autonomous Driving with long term references at several OEMs
System Designsystem engineering,
use & test cases, architecture, component & function
specification
Tailored Control & SW Development
concept & series development customer features, modification/
adaptation
Advanced Predictive Functions improving vehicle attributes e.g. energy or fuel efficiency
Calibration, Testing & Validation
derivative integration, optimization & assessment, testing from lab, XiL to road
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 4Public
IntroductionMarket Drivers / Customer Requirements
▪ Accident free driving active safety functions e.g. emergency braking, lane keeping assistant
▪ Driver relief & comfort functions e.g. parking assistant, adaptive cruise control
▪ Connectivitye.g. smart phone interaction, real time traffic information, V2X, cloud computing
▪ Fuel/energy efficiency e.g. EV driving range, Fuel saving by predictive functions & platooning
▪ Operating cost: Driver substitution as TCO argument at mainly transport & shared mobility business
Key importance
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 6Public
IntroductionPredictive Energy Management Leveraging ADAS Data
PREDICTIVE CHARGINGTRAFFIC LIGHT
ASSISTANT
PREDICTIVE THERMALMANAGEMENT
PREDICTIVE ADAPTIVE CRUISE CONTROL
ECOROUTING
COASTING ASSISTANT
Sources: www, AVL
PREDICTIVE GEARSHIFT
ADAS/ADHMI
CONNECTED
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 7Public
Energy-Efficient CACCContents
1. Introduction to Predictive Energy Management
2. Traffic Light Assistant
3. Energy-Efficient Cooperative Adaptive Cruise Control
a) Problem Overview
b) Model Predictive Control
c) Results
4. Summary & Conclusion
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 8Public
Market Example: TLA Audi
• Audi announces first V2I service in 10 US cities (as of 19/11/18) with Traffic Light information system. System available in 2017 on Q7, A4 & A4 Allroad.
• Current system: Real-time communication with city’s traffic management system, displays traffic light status (red/orange/green) & “time-to-green”.
• Future dvpt: integration of system with start/stop function, Green Light Optimized Speed Advisory (GLOSA), optimized navigation routing & other predictive services.
https://media.audiusa.com/en-us/releases/241#gallery
https://media.audiusa.com/en-us/releases/92https://www.audi-mediacenter.com/en/audimediatv/video/audi-connect-online-traffic-light-information-2543
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 11Public
Market Example: TLAJaguar Land Rover
• GLOSA (Green Light Optimal Speed Advisory) system: V2X communication with traffic lights to advise driver of optimal approach speed to avoid having to stop at traffic light junctions.
• “Prevents drivers from racing to beat lights & improves air quality by reducing harsh acceleration or braking near lights.”
• Trials on Jaguar F-Pace.
• Works in conjunction with other ADAS: Adaptive Cruise Control, Intersection Collision Warning.
https://media.jaguarlandrover.com/news/2018/11/jaguar-land-rover-gets-green-light-solve-150-year-old-problem?q=&start=0&brand=corporate
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 13Public
▪ Uses V2I (Vehicle-to-Infrastructure) data to intelligently approach multiple Traffic Light (TL):• Goal to find most energy efficient velocity.
▪ Model Predictive Control (MPC) formulation: • Receding horizon approach.• Real-time optimization by cost function minimization & constraints.
Traffic Light Assistant (TLA)AVL’s TLA Visualized
C2I
Communication
Digital Map
GPS Navigation
System
Electronic
Horizon
Recommended cruising speed
Vo=
40km/h
Optimal speed profile
System
activation
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 14Public
EnergySavings*
Time Savings*
17% 3.8%
Battery SoC as metric of EV energy savings
Traffic Light AssistantResults of AVL’s 1st Gen. TLA for 4WD EV in 2013
* Depends on test case, baseline, etc.
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 15Public
Energy-Efficient CACCContents
1. Introduction to Predictive Energy Management
2. Traffic Light Assistant
3. Energy-Efficient Cooperative Adaptive Cruise Control
a) Problem Overview
b) Model Predictive Control
c) Results
4. Summary & Conclusion
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 16Public
Energy-Efficient CACCContents
1. Introduction to Predictive Energy Management
2. Traffic Light Assistant
3. Energy-Efficient Cooperative Adaptive Cruise Control
a) Problem Overview
b) Model Predictive Control
c) Results
4. Summary & Conclusion
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 17Public
Energy-Efficient CACC – Problem OverviewWhat is Cooperative Adaptive Cruise Control?
Cruise Control (CC): Longitudinal speed control with set speed defined by human driver.
Adaptive Cruise Control (ACC): Adapts speed based on distance to & speed of preceding vehicle, e.g.
measured using on-board sensors such as RADAR or Camera.
Cooperative Adaptive Cruise Control (CACC): ACC extension supported by communication with
surrounding traffic & infrastructure, possibly also other off-board data sources e.g. cyclists, pedestrians.
Adaptive Cruise Control(ACC)
Ego vehicle Preceding
vehicle(s)
Cruise Control(CC)
Ego vehicle
Cooperative Adaptive Cruise Control (CACC)
Ego vehicle Preceding
vehicle(s)Infrastructure
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 18Public
Energy-Efficient CACC – Problem OverviewInputs & Constraints
Ego vehicle Preceding vehicles
… …
Following vehicle
Infrastructure-to-vehicle
(I2V)
Radar
Traffic light
1
Traffic light
2
Vehicle-to-Vehicle
(V2V)
V2V / I2V
Vehicle following known path
Route contains altitude, curvature, traffic lights, ...
Includes time-dependent constraints such as traffic
light signal phases
Holistic approach needed!
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 19Public
Energy-Efficient CACC – Problem OverviewApproach
▪ Holistic & full range predictive speed control strategy (CACC) including ego-vehicle & its static
& dynamic powertrain characteristics, uses V2X derived RT traffic, infrastructure & route data.
▪ Optimizes in real-time trade-off between energy efficiency, driver comfort & safety.
Velocity 𝒗
Acceleration 𝒂
Powertrain Model
Gear 𝑮
Road Inclination 𝜽
Ego Vehicle Preceding Vehicle
V2X (Vehicle to Anything)
On-Board Sensors
(e.g. radar)
Traffic Light
Vehicle-to-Vehicle (V2V)
Energy savings up to 30%*
Energy Consumption Map computed Online in real-time Optimization
Vehicle-to-Infrastructure (V2I)
…
Inertia, Drag, Rolling Res., Gravity
* Depends on test case, baseline, etc.
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 20Public
Energy-Efficient CACCContents
1. Introduction to Predictive Energy Management
2. Traffic Light Assistant
3. Energy-Efficient Cooperative Adaptive Cruise Control
a) Problem Overview
b) Model Predictive Control
c) Results
4. Summary & Conclusion
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 21Public
Energy-Efficient CACC – MPCIntroduction to Model Predictive Control (MPC)
𝑢
𝒖𝒐𝒑𝒕 = 𝑢 0 , 𝑢 1 ,… , 𝑢 𝐻𝑇𝑇
Optimal sequence of control inputs over prediction horizon 𝐻𝑇
* i.e. vehicle & driving environment
*
** vehicle states, traffic light information, etc.
**
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 22Public
Energy-Efficient CACC – MPCHybrid Model Predictive Control
▪ Hybrid* Model Predictive Control (MPC) dynamically incorporates descriptions of upcoming traffic & road conditions as constraints in receding horizon.
▪ Non-linear constraints like energy consumption, gear shifts, full load, & road attributes (e.g. gradient, curvature, speed limits) modelled.
▪ eHorizon & V2X used for better predictions of preceding traffic & infrastructure, including traffic lights, variable speed limits, delivery & bus stops.
Energy consumption mapincluding gear shifting
Acceleration limitsincluding road gradient
Road segmentation for topology, speed limits, etc.
*Note Hybrid here refers to modelling technique, not the powertrain type
1st2nd
3rd4th5th
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 29Public
Energy-Efficient CACCContents
1. Introduction to Predictive Energy Management
2. Traffic Light Assistant
3. Energy-Efficient Cooperative Adaptive Cruise Control
a) Problem Overview
b) Model Predictive Control
c) Results
4. Summary & Conclusion
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 31Public
Energy-Efficient CACC – Simulation ResultsGraz Route Simulation without Traffic
Energy savings: 25% without traffic,without an increase in travel time
Adjustable travel time & driver comfortability
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 32Public
Energy-Efficient CACC – Simulation ResultsGraz Route Simulation with Traffic
Energy savings: 16% with traffic,without an increase in travel time
Adjustable travel time & driver comfortability
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 33Public
Energy-Efficient CACC – Testbed ResultsAustrian FFG TASTE Project
“Traffic Assistant Simulation & Testing Environment“.10.2015 – 06.2017
▪ Virtual test environment for ADAS, including real communication units.▪ RT interaction / communication of traffic control infrastructure & cars.▪ Specific testbed setting for specialized application.▪ Testbed & Road testing with real vehicle & V2X units.
EECACC controlled test case achieves lower fuel consumption by end of the maneuver (up to 25% diesel consumption savings measured).
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 34Public
Energy-Efficient CACCContents
1. Introduction to Predictive Energy Management
2. Traffic Light Assistant
3. Energy-Efficient Cooperative Adaptive Cruise Control
a) Problem Overview
b) Model Predictive Control
c) Simulation Results
d) Testbed Results
4. Summary & Conclusion
Stephen Jones, N. Wikström, A. Ferreira Parrilla | DS | 28 February 2019 | 35Public
Energy-Efficient CACCSummary
▪ Increasing interest in V2X communications to intelligently connect conventional & electrified vehicles.
▪ V2X supported ADAS such as simple Traffic Light Assistants, now starting to be introduced in market.
▪ Efficiency, safety & convenience all benefit from optimized vehicle speed profiles.
▪ AVL’s Energy-Efficient Cooperative Adaptive Cruise Control (EECACC) reduces energy consumption by
up to 30%* in simulated city scenario, 25% on testbed.
▪ EECACC considers static layout, sizing & efficiency of powertrain, as well as dynamic state (e.g. SoC,
temperature) of powertrain, road topology, traffic, & traffic light signal phasing & timing (SPAT) data.
▪ Benefits of EECACC extended to other functions e.g. predictive hybrid powertrain mode selection.
* Like most predictive functions, the benefits depend on specific use case, baseline taken, etc.
www.avl.com
Thank You