Presentazione della ricerca "Advanced Driver Assistance Control
Transcript
Presentazione della ricerca "Advanced Driver Assistance Control
PROGRAMMA Ore 10.30 Saluti di apertura Giovanni Azzone, Rettore Politecnico di Milano Ore 10.45 Discussione della ricerca: “Sistemi predittivi di marcia” Modera: Federico Cheli, Dipartimento di Meccanica, Politecnico di Milano Intervengono: Maurizio Boiocchi, General Manager Technology Pirelli Tyre e Amministatore Delegato Pirelli Labs Elisabetta Leo, Dipartimento di Meccanica, Politecnico di Milano Ore 11.45 Tavola rotonda: “L’auto del futuro” Modera: Giampio Bracchi, Presidente Fondazione Politecnico di Milano Intervengono: Enrico Pisino, Head of Research & Innovation, Fiat Chrysler Automobiles Franco Cimatti, Responsabile Impostazione Prodotto Ferrari Guidalberto Guidi, Presidente Ducati Energia Umberto Bertelè, Dipartimento di Ingegneria Gestionale, Politecnico di Milano Marco Tronchetti Provera, Presidente e Amministratore Delegato Pirelli & C., Presidente Fondazione Silvio Tronchetti Provera Ore 13.00 Conclusioni Marco Tronchetti Provera, Presidente e Amministratore Delegato Pirelli & C., Presidente Fondazione Silvio Tronchetti Provera Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Federico Cheli Dipartimento di Meccanica, Politecnico di Milano PROJECT TARGET ADAS systems improvement via potential friction knowledge 2008…. … 2014 6 years sponsored research project Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 PROJECT TARGET: STUDIED ADAS SYSTEMS If I suppose to know the available potential friction which ADAS system can be improved and how? LONGITUDINAL LONGITUDINAL i.e. Autonomous Emergency Brake (AEB) could exploit friction knowledge: Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 LATERAL i.e. Curve Speed Warning (CSW) could improve their performances: By World Health Organization Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Implications Developing vehicles that: - optimize the traffic flow; - optimize fuel consumption; - provide safe and comfortable transportation and at the same time - have minimal impact on the environment. Satisfy these diverse and often conflicting requirements is a great challenge. Solution: electromechanical subsystems that employ sensors, actuators and feedback control. Technological advancement in these fields had an enabling role in promoting this trend. Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 The ADAS systems Many advanced driver assistance systems (ADAS) are being developed by automotive manufacturers to automate driving operations in order to reduce highway accidents. Examples include: • collision avoidance systems: which automatically detect slower moving preceding vehicles and provide warning and brake assist to the driver; • adaptive cruise control (ACC) systems: which are enhanced cruise control systems and enable preceding vehicles to be followed automatically at a safe distance; • lane departure warning systems; • lane keeping systems: which automate steering on straight roads; • vision enhancement/ night vision systems • driver condition monitoring systems: which detect and provide warning for driver drowsiness, as well as for obstacles and pedestrians Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 The ADAS systems Safety increasment Pollution reduction & energy saving ADAS Advanced Driver Assistance Systems Cheap sensing, actuating, computational capabilities Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Congestion reduction The ADAS systems From SBD - Automotive technology consultancy and research (www.sbd.co.uk/) Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 The ADAS systems By 2020 in Europe, over 70% of vehicle models sold will be fitted with at least one type of ADAS. EuroNCAP decided to include testing of ADAS as part of their five star rating. Updates in legislation in the USA from 2018 will result in a more aggressive growth in ADAS. ADAS is still in its infancy for China where currently the market is dominated with ADAS fitted on EU and USA imported vehicles: E.g Volvo From SBD - Automotive technology consultancy and research (www.sbd.co.uk/) Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Level 0 - no-automation: control of the vehicle completely entrusted to the driver at all times. Systems that provide safety warnings level 1 - function-specific automation: one or more specific control functions. The driver has overall control, but can choose to cede limited authority over a primary control level 2 - combined function automation: automation of at least two primary control functions designed to work together. The driver is still responsible for monitoring the roadway and safe operation and is expected to be available for control at all times and on short notice. level 3 - limited self-driving automation: the driver is enabled to cede full control under certain traffic or environmental conditions. The driver is expected to be available for occasional control, but with sufficiently comfortable transition time. level 4 - full self-driving automation: the vehicle is designed to perform all safety-critical driving functions and monitor roadway conditions for an entire trip. level 0 - no-automation: lane departure warning forward collision warning blind spot monitoring level 1 - function-specific automation: Adaptive Cruise Control Lane keeping Assist level 2 - combined function automation: Super Cruise Technology level 3 - limited self-driving automation: Platooning (SARTRE Project) level 4 - full self-driving automation: Google Car, University collaboration researches Active safety: Autonomous vehicles One Level MPC, with Risk Evaluation, Non Linear Vehicle Model and Safety/Comfortable Trajectory Planner Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Cooperation with Active safety: ADAS Camera, radar, navigation and sensors in general are stimulating a growing number of ADAS systems aimed to improve safety: Source: BMW, IPG Automotive GmbH Where I am, probable or selected route turns, slope, road signs, etc… Source: Porsche, IPG Automotive GmbH Traffic conditions, surrounding objects Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Wheels speed, steer angle,… Active safety: ADAS Are these information enough to implement a new ADAS or an autonomous vehicle? Or I need something else ? Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Active safety: ADAS I need something else !!! ONLY including potential friction estimation the new ADAS could provide previews of warning situations due to actual driving speed. Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Maurizio Boiocchi General Manager Technology Pirelli Tyre e Amministatore Delegato Pirelli Labs COMPANY OVERVIEW > One of the most innovative and profitable companies in the sector (13,4% tyre Ebit margin in 2013) > A strong Brand, enhanced by F1 involvement - official tyre supplier in ‘11-’16 > A unique positioning in the industry: • • • Growing focus on Premium segment (~ 54% of car sales in 2013) Strong exposure to the Replacement channel (75% of revenues), less exposed to Auto sector dynamics Balanced distribution of revenues, leading the development of Premium tyres in Emerging Market 6,15 €/bln revenues in 2013 SALES BY SEGMENT SALES BY REGION Car 63% Truck 20% Central And South America 36% Moto 6% Steelcord & others Other European Country 27% 11% 1 NAFTA 11% Asia/Pacific 8% MEAI 8% Italy Russia 4% 6% VEHICLE ELECTRONICS EVOLUTION: CAR MAKERS DEMAND > Investments cover all the car segments for new functionalities to maximize differenciation and fulfill customer’s demand in the • • • safety economy & infotainment areas through • • • • > ADAS systems Driver Information Fuel economy: Hybrid / electrical engine Connectivity and Infotainment Electronics value will increase, in Europe, from 20% of today to up 60% in 2018 of the total cost of a premium vehicle full opt (F&S source). 2 PIRELLI’S ANSWERS: CYBERTMTYRE NEW VEHICLE CONTROL LOGIC ENGINEERING FEATURE EXTRACTION SENSOR 3 TYRE EVOLUTION: ELECTRONICS KICKS IN > Pirelli New instrumented Tyres now have a DIRECT roles in Vehicle • • PREVENTIVE Safety PERFORMANCE Driving specifically mapped in • • • • > ADAS systems Fuel economy: Hybrid / electrical engine Connectivity and Infotainment Driver Information Cyber Tyre dynamically provides to the Vehicle unique informations of grip margin and actual forces exchanged to the ground to exploit fully preventive safety paradigms and performance driving 4 CYBER TYRE IN A NUTSHELL TYRE DYNAMICS • Vertical Load • Longitudinal & Lateral Forces • Speed margin before hydroplaning conditions • Grip & Grip margin before sliding • Patch Area • Slip Ratio • Slip Angle • Camber • Tyre Aligning Moment ROAD CONDITION • Road surface texture TYRE CONDITION • Tyre Temperature & Pressure • Tyre wearing • Number of tyre revolutions • Tyre ID 5 NEXT GENERATION APPLICATIONS ENABLED BY CYBER TYRE Patch Are Control for Maximasing Cornering Performance Aero-Loads control Warning Information (Tyre Wearing, Aquaplanning, Grip margin) Active Suspension Control Performance Enhancement New Applications Electronic Stability Program Enhanced ADAS (Adaptive Speed Limit) Advanced Vehicle State Estimator (combined Active Controls) Traction Control, E-Differential 6 THE ADAS MISSING LINK: CYBERTMTYRE ‘MAXIUM AVAILABLE GRIP’ CLARKE, D. ET AL. FATAL VEHICLE-OCCUPANT COLLISIONS: AN IN-DEPTH STUDY. 2007. ROAD SAFETY RESEARCH REPORT NO. 75. DEPARTMENT FOR TRANSPORT, LONDON. • 1185 fatal vehicle occupant from 1994 to 2005. • Over 65% of the accidents examined involved driving at excessive speed Speeding accidents were examined, and divided into separate types, based predominantly on the levels of risktaking by drivers. Type 3 – ignorance of speed, for example failing to realize that wet road conditions increase the likelihood of skidding 2006 EUROPEAN PROJECT. FINAL REPORT “…Along with driver behavior and alertness, friction is one of the remaining key unknowns in the algorithms of future ADAS (Advanced Driver Assistance Systems) that calculate the risk of collision, or safe speed. For example, if a Collision Mitigation System assumes high friction, it will have poor performance on snow, since it will brake too late..” TITLE: Sensor Data Fusion Based Estimation of Tyre-Road Friction to Enhance Collision Avoidance DATE: 2010 7 Elisabetta Leo Dipartimento di Meccanica, Politecnico di Milano PROJECT TARGET ADAS systems improvement via potential friction knowledge 2008…. … 2014 … future development 4 partecipations to international conferences 4 master thesis Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 6 years sponsored research project PROJECT TARGET: STUDIED ADAS SYSTEMS If I suppose to know the available potential friction which ADAS system can be improved and how? LONGITUDINAL LONGITUDINAL i.e. Collision Avoidance Warning(CAW) could exploit friction knowledge: Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 LATERAL i.e. Curve Speed Warning (CSW) could improve their performances: PROJECT TARGET ONLY including potential friction estimation the new ADAS could provide previews of warning situations due to actual driving speed. The CT Friction IDentification algorithm has been used to verify the possibility to increase the performance via potential friction knowledge Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 PROJECT MAIN RESULTS Three steps has been followed to reach the target for both CSW e CAW • Logics improvement via CT potential friction knowledge • Numerical verification • On-board tests. LONGITUDINAL Collision Avoidance Warning (CAW) Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 CAW: how it works For instance CAW systems work as follows: - Phase 1: warning is provided to the driver; V - Phase 2: the obstacle is closer and braking system is pre-charged to assure fast reaction; - Phase 3: braking pressure is applied to avoid accident. Radar Visibility Range V CAW Phase 1 (warning) CAW Phase 2 (pre-charge) CAW Phase 3 (brake) time Obstacle distance d(t) Pre-assigned REFERENCE potential friction μREF Brake pre-charge Vehicle deceleration Relative speed v(t) Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Driver warning a(t) CAW: how it works Simulations of CAW system with pre assigned value of μREF = 0.7: Legend: μREF = 0.7 1 = avoided accident ice 0 = accident occurs Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 wet stones concr. asphalt track CAW: how it works Simulations of CAW system with pre assigned value of μREF = 0.7: Legend: μREF = 0.7 1 = avoided accident check residual distance “d” 0 = accident occurs check impact speed “v” Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 CAW: how it works μREF = 0.2 μREF = 0.4 μREF = 0.7 Residual distance [m] Simulations about how an CAW system will work with different values of μREF: μREF - More cases in which accident occurs. Higher impact speeds. BAD EFFECTIVENESS Compromise is needed… BAD EFFICIENCY - Higher residual distances. More cases in which accident is avoided. μREF μREF Real road μP Impact speed [km/h] μREF Accident avoided index Real road μP μREF Legend: μREF = 0.7 μREF = 0.4 μREF = 0.2 CAW: benefits including CT-F.ID algorithm Exploiting the CT-F.ID algorithm the ECU is able to refresh the value of μREF The CT Potential Friction Identification algorithm (CT-F.ID) provides road friction μ within 5 ranges; Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Are such conditions suitable to assure ADAS improvements? CAW: benefits including CT-F.ID algorithm Comparison between CAW system with fixed value of μREF = 0.7 and CAW exploiting friction knowledge: residual distance Identified - μCT-F.ID More cases in which accident is avoided. Lower impact speeds (both maximum and average). Still short residual distances when accident is avoided. BETTER EFFECTIVENESS & GOOD EFFICIENCY Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Legend: Avoided accidents Low impact speed μREF = 0.7 μREF = μCT CAW: on-board implementation 4 instrumented tyres RADAR INTERNAL CAMERA Real time PC. CAW logic has been implemented and integrated with CT data Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 CAW: on-board implementation. Example on urban concrete surface. NOTE: expected potential friction on dry concrete surface = 0.6 ÷ 0.9 Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 CAW: on-board implementation. Example on urban concrete surface. My vehicle speed[km/h] Relative speed [km/h] Safe Warning Danger Relative distance [m] Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Out of range Identified potential friction [-]. CAW: on-board implementation. Example on urban concrete surface. Time: 4 seconds My vehicle speed: 42km/h = 11,6m/s Relative speed : 23km/h = 6,4m/s Relative distance : 20m Potential friction: 0.7 Maximum deceleration: 7m/s2 Vmy = Vmy vehicle Vtarget = V in front vehicle Vmy > Vtarget CAW ALGORITHM: - Vmy < Vtarget - No object in the radar range DANGER condition Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 SAFE condition One target is to reduce to zero the relative speed between the two vehicles. It considers an usual human time delay before braking (i.e. 1s) @42km/h = 11.6m It considers to avoid to fully brake at the beginning, the initial deceleration imposed is lower than the max. CAW: on-board implementation. Example on urban concrete surface. Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 LATERAL Curve Speed Warning (CSW) Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 CSW system: how it works Operation scheme: 1. The driver sets the path 2. The CSW logic evaluates the safety speed on the path 3. If the vehicle speed overcomes the safety speed a signal warns the driver Current system’s weakness Potential friction unknown 1st possible approach: to impose usual potential friction conditions (i.e. µ=0.7-0.8) 2nd possible approach: to impose high Safety margin to impose a low potential friction Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 μREF μREF Risk: wrong safety speed prediction on low-µ surface. BAD effectiveness Risk: warning always ON. BAD efficiency CSW : benefits including CT-F.ID algorithm Computed safety speed as a function of trajectory radius and potential friction coefficient Constant radius curve (120m) Safety speed changing the potential friction. Potential friction Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Trajectory radius [m] CSW : benefits including CT-F.ID algorithm Simulations about how a CSW system will work with different values of μREF: μREF If critical road conditions occur (i.e. ice, snow) high overestimated saefty speed μREF = 0.7 Computed Safety speed Real potential friction If μ0 =0.2 the safety speed is 43.6% LOWER If μ0 =1 the safety speed is 21.8% HIGHER Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 BAD EFFECTIVENESS CSW : benefits including CT-F.ID algorithm Simulations about how a CSW system will work with different values of μREF: μREF If critical road conditions occur (i.e. ice, snow) high overestimated saefty speed BAD EFFECTIVENESS Compromise is needed… Computed Safety speed BAD EFFICIENCY μREF = 0.4 μREF Real potential friction If μ0 =0.2 the safety speed is 19.8% LOWER Conservative approach BUT if usual medium-high road conditions occur (i.e. dry asphalt) high underestimated safety speed If μ0 =1 the safety speed is 78.1% HIGHER Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 CSW : benefits including CT-F.ID algorithm Simulations about how a CSW system will work with different values of μREF via CT. μREF = CT-F.ID Max UNDERESTIMATED speed: 13.2% Max OVERESTIMATED speed: 12.1% Real potential friction Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Lower difference between real safety speed and the safety speed computed by the algorithm Both in safety and unsafety regions BETTER EFFECTIVENESS & GOOD EFFICIENCY CSW Implementation: on-board implementation Potential friction µ0 Vertical load: Fz [N] Vehicle position Trajectory CSW Algorithms SAFETY STATUS Safety speed E-Horizon Carthography Vehicle speed [km/h] Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Safe Warning Danger CSW Implementation: on-board implementation Potential friction µ0 Vertical load: Fz [N] Vehicle position Trajectory CSW Algorithms SAFETY STATUS Safety speed E-Horizon Carthography Vehicle speed [km/h] How is it possible to verify the on-board procedure introducing digitalized maps? Safe Warning Danger CSW Implementation: on-board implementation Co-simulation CSW Algorithms SAFETY STATUS Safety speed Potential friction µ0 Safe Warning Danger Vehicle speed [km/h] E-Horizon My Vehicle position Potential friction µ0 Vehicle speed [km/h] Vehicle position Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 VEHICLE MODEL 2° turn 1° turn Potential friction [-] CSW Implementation. Example on dry asphalt 2° turn 1 0.5 0 950 1000 1050 s [m] 1100 1150 1000 1050 s [m] 1100 1150 1050 s [m] 1100 Hypothesis: potential friction constant between my position and the next curve. Curve Radius [m] 0 -20 -40 -60 -80 -100 950 100 Safety speed My vehicle speed V [km/h] 80 60 40 20 0 950 Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 1000 1150 CSW Implementation: on-board implementation Safe Warning Danger Simulated world – vehicle model Real world – camera Blue line: maximum safety speed Red dots. Vehicle speed Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 CONCLUSIONS - The knowledge of road friction coefficient plays a key role in ADAS systems improvements, both in terms of effectiveness and efficiency. - Two ADAS systems have been considered to verify their improvement introducing potential friction knowledge: Collision Avoidance and Curve Speed Warning - The friction knowledge has been supposed to be identified via CT-F.ID with 5 ranges. It has been verified to be adequate for ADAS systems performances improvements: 1) Numerical simulations of both CAW and CSW systems behavior proved how 5 ranges of μ can improve both system’s effectiveness and efficiency. 2) The algorithms developed togheter with CT potential friction identification have been implemented on-board Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 FUTURE DEVELOPMENTS. Adaptive Speed Limiter (ASL): IDEA TODAY Speed Limiter click click Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 As a function of road speed limit the driver set the maximum speed of the vehicle with its speed limiter. Ok, I fix 60km/h FUTURE DEVELOPMENTS. Adaptive Speed Limiter (ASL): IDEA TOMORROW Adaptive Speed Limiter Via road speed limit + road µ condition infos, the system fix my max speed Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014 FUTURE DEVELOPMENTS. TARGETS TARGETS: 1. To develop a simulator exploiting numerical vehicle dynamic code. WHY? • • • To impose offline traffic conditions. To test the HumanMachineInterface. To verify different drivers behaviour with same test scenario. • To impose a fixed virtual path exploiting google maps or other navigation system. TEST CONDITION REPETEABILITY. TARGET 1 2. To equip the PIRELLI vehicle with ADAPTIVE SPEED LIMITER logic TARGET 2 Advanced Driver Assistance Control | Politecnico di Milano | 15 ottobre 2014