information society technologies (ist) programme...lcm light control module recu retarder electronic...

48
INFORMATION SOCIETY TECHNOLOGIES (IST) PROGRAMME FP6 - IST - 2004 - 4 - 027006 D9: Description of commercial vehicle demonstration system Dissemination PU = Public Work package WP6 – System Integration Authors Stefan Nord, VTEC, VTEC_SN Due date 01.10.2008 Delivery date 03.11.2008 Status Final This document FRICTION_Deliverable_D9.doc Short description This document contains a description of the commercial vehicle demonstration system.

Upload: others

Post on 11-Feb-2021

1 views

Category:

Documents


0 download

TRANSCRIPT

  • INFORMATION SOCIETY TECHNOLOGIES (IST)PROGRAMME

    FP6 - IST - 2004 - 4 - 027006

    D9: Description of commercial vehicle demonstration system

    Dissemination PU = PublicWork package WP6 – System Integration

    Authors Stefan Nord, VTEC, VTEC_SN

    Due date 01.10.2008

    Delivery date 03.11.2008

    Status Final

    This document FRICTION_Deliverable_D9.doc

    Short description This document contains a description of the commercial vehicledemonstration system.

  • Deliverable D9 Dissemination PU 2 (48)

    PART 0 - Preliminaries

    Authors

    EditorMr. Stefan Nord VTEC VTEC_SN

    ContributorsMr. Stefan Nord VTEC VTEC_SNMr. Johan Casselgren VTEC VTEC_JCMs. Anna Heyden VTEC VTEC_AHDr. Matti Kutila VTT VTT_MKMr. Michael Koehler IBEO IBEO_MIKMr. Ari Tuononen Helsinki University of Technology TKK_AT

    Consortium participants:Technical Research Centre of Finland, VTT (FIN)Centro Ricerche FIAT S.C.p.A, CRF (ITA)IBEO Automobile Sensor GmbH, IBEO (D)Rheinisch-Westfaelische Technische Hochschule Aachen, ika (D)Magneti Marelli Sistemi Elettronici S.p.A, MM (ITA)Nokian tyres plc, NR (FIN)Pirelli Tyre S.p.A, PI (ITA)Siemens AG, SI (D)Helsinki University of Technology, TKK (FIN)Volvo Technology Corporation, VTEC (SWE)

  • Deliverable D9 Dissemination PU 3 (48)

    Revision history

    Version Date Description Author0.0 2008-07-24 Added headings and some initial

    informationVTEC_SN

    0.1 2008-07-30 Added general information about theselected commercial vehicle.Added information about the HMI.

    VTEC_SN

    0.2 2008-08-14 Added information and pictures aboutsensors

    VTEC_SN

    0.3 2008-08-21 Added details about COMPOSE CMbB VTEC_SN0.4 2008-09-08 Added details about Road eye and the

    encapsulation tubesVTEC_JC

    0.5 2008-09-16 Added more info to chapter 2 and addedheaders for VTT Cameras and IBEO LaserScanner.

    VTEC_SN

    0.6 2008-09-17 Some minor changes in chapter 6 and 7. VTEC_SN0.7 2008-09-22 Added info about IMU VTEC_SN0.8 2008-09-25 The IcOR system description added VTT_MK0.9 2008-09-25 Corrected some figures and figure texts. VTEC_SN0.10 2008-09-26 The IcOR system description updated VTT_MK0.11 2008-09-29 Compression of figures and correction of

    Roadeye to Road eyeVTEC_SN

    0.12 2008-10-24 Added description of the IBEO laserscanner(taken from D8)

    VTEC_SN

    0.13 2008-10-29 Spell check of document VTEC_SN0.14 2008-10-30 Minor corrections VTEC_SN0.15 2008-10-31 Completed Laserscanner desription IBEO_MIK0.16 2008-11-03 Updated tyre sensor description with input

    from TKK_ATVTEC_SN

    1.0 2008-11-03 Version finalized VTEC_SN

  • Deliverable D9 Dissemination PU 4 (48)

    Document structureThe document is divided as follows:

    PART 0 – PreliminariesContains meta information about the document and its contents.

    PART 1 – General Commercial Vehicle DescriptionContains a description of the Volvo FH12 460 Tractor with an overview of the truck, itsperformance and physical properties.

    PART 2 – Demonstration System DescriptionDescribes the commercial vehicle demonstration system.

    PART 3 – Safety Application DescriptionDescribes the selected safety application from the PReVENT project COMPOSE.

    PART 4 – AppendicesContains references.

  • Deliverable D9 Dissemination PU 5 (48)

    Abbreviations and acronyms

    ABS Anti-lock Braking SystemACC Adaptive Cruise ControlADAS Advanced Driver Assistance SystemAEB Automatic Emergency BrakingALA Alasca laserscannerASIC Application Specific Integrated CircuitBAS Brake AssistCAN Controller Area NetworkCAS Collision Avoidance SystemCMbB Collision Mitigation by BrakingCOR CorrevitCW Collision WarningEFF Environmental Feature FusionEMC Electromagnetic CompatibilityESP Electronic Stability ProgramFOV Field Of ViewHMI Human Machine InterfaceHW HardwareIMU Inertial Measurement UnitIP Integrated ProjectsIVSS Intelligent Vehicle Safety SystemsLSB Least Significant BitMSB Most Significant BitN/A Not ApplicableOEM Original Equipment ManufacturerPSD Position Sensitive DiodeREY Road eyeRFE Road Friction EstimationRPU Rapid Prototyping UnitRWIS Road Weather Information SystemSAE Society of Automotive EngineersSRIS Slippery Road Information SystemSTREP Specific Targeted Research ProjectSUV Sport Utility VehicleSW SoftwareTCS Traction Control SystemTFF Tyre Feature FusionTMC Traffic Message ChannelTYR Tyre Sensor secondary RPUVSC Vehicle Stability ControlWP Work PackageWT Work Task

  • Deliverable D9 Dissemination PU 6 (48)

    Table of contents

    PART 0 - PRELIMINARIES.............................................................................................................................2

    AUTHORS............................................................................................................................................................2REVISION HISTORY.............................................................................................................................................3DOCUMENT STRUCTURE ....................................................................................................................................4ABBREVIATIONS AND ACRONYMS .....................................................................................................................5TABLE OF CONTENTS .........................................................................................................................................6

    PART 1 – GENERAL COMMERCIAL VEHICLE DESCRIPTION ........................................................7

    1 COMMERCIAL VEHICLE DESCRIPTION ....................................................................................................71.1 Overview........................................................................................................................................71.2 TEA2 – Truck Electronic System Architecture............................................................................81.3 Vehicle Based Sensors ..................................................................................................................9

    PART 2 – DEMONSTRATION VEHICLE SYSTEM DESCRIPTION ..................................................11

    2 DEMONSTRATOR VEHICLE NETWORK ARCHITECTURE .........................................................................113 ADDITIONAL SENSORS ...........................................................................................................................13

    3.1 Environmental Sensor – Road eye .............................................................................................133.2 Environmental Sensor - IcOR.....................................................................................................173.3 Environmental Sensor – IBEO Laser Scanner..........................................................................193.4 IMU ..............................................................................................................................................223.5 Tyre Sensor..................................................................................................................................24

    4 HMI HARDWARE ...................................................................................................................................375 FRICTION ESTIMATION SYSTEM..........................................................................................................38

    PART 2 – SAFETY APPLICATION DESCRIPTION................................................................................39

    6 PREVENT APPLICATIONS WITHIN THE FH12 TRUCK..........................................................................397 TRUCK DEMONSTRATOR SAFETY APPLICATION ..................................................................................398 HMI APPLICATION.................................................................................................................................42

    8.1 Previous studies...........................................................................................................................428.2 Problems concerning friction value and HMI presentation.....................................................448.3 HMI simulation............................................................................................................................468.4 Other areas of use .......................................................................................................................47

    PART 4 – APPENDICES..................................................................................................................................48

    REFERENCES.....................................................................................................................................................48

  • Deliverable D9 Dissemination PU 7 (48)

    PART 1 – General Commercial Vehicle Description

    1 Commercial Vehicle Description

    This chapter describes the chosen commercial vehicle within the FRICTION project.

    1.1 OverviewA Volvo FH12 460 4x2 High Tractor was chosen as commercial vehicle demonstrator forthe FRICTION project. It is equipped with a 12 litre 460 hp diesel engine. It is 2600 mmin width and the wheel base is 3600 mm. The truck can be equipped with a ballast inorder to achieve a more distributed load profile between the front and rear axle (seeFigure 1 below). Without the ballast the load is approximately 5.5 tons on the front axleand 2 tons on the rear axle respectively. With ballast the load is approximately 6.3 tonson the front and the rear axle respectively.

    Figure 1 The Commercial Vehicle Demonstrator equipped with ballast

    Furthermore, within the project the truck has been equipped with new 315/70R22.5 tyressupplied by Pirelli.

    This specific truck has also been used for the integrated subprojects within PReVENTlike e.g. SASPENCE, APALACI and COMPOSE. It has also been used as demonstratorfor the FP6 project AIDE. This means that systems developed within those projects areavailable to be used in conjunction with FRICTION applications if needed.

  • Deliverable D9 Dissemination PU 8 (48)

    1.2 TEA2 – Truck Electronic System Architecture

    The communication in the truck is run on two different networks that follow two differentstandards. The control network follows the SAE J1939 standard and the differentcommands to other units are sent on this network. This network is segmented in totally 5different CAN busses. Diagnostics and information is sent on the other networkaccording to the SAE J1708 standard. This link also gives a system redundancy andworks as a back up in case of failure on the J1939. All vital control units are connected tothis network.

    Figure 2 The Volvo Truck Electronic System Architecture

    Module DescriptionTECU Transmission Electronic Control

    UnitGECU Gearlever Electronic Control UnitABS-B Anti-lock Braking SystemABS-F Anti-lock Braking SystemEBS Electronic Brake SystemESP Electronic Stabilizing ProgramVECU Vehicle Electronic Control UnitTachograph TachographEMS Engine Management SystemBBM Body Builder ModuleIMMO Immobilizer

  • Deliverable D9 Dissemination PU 9 (48)

    Module DescriptionSRS Supplemental Restrain SystemECC Electronic Climate ControlMCC Manual Climate ControlDynafleet DynafleetVCADS N/AECS Electronic Controlled SuspensionCluster Instrument ClusterAudio AudioSWM Steering Wheel ModuleButtons in wheel Buttons in wheelPhone PhoneLCM Light Control ModuleRECU Retarder Electronic Control UnitRAS Rear Axle Steerable

    Table 1 Description of ECU:s in the Volvo Truck Electronic System Architecture

    1.3 Vehicle Based Sensors

    The truck is of course equipped with a big number of vehicle based sensors that could beof interest for the FRICTION algorithm. The most important vehicle sensor signals andtheir CAN format are outlined in Table 2 below.

    sensor signal name transmitterinitialvalue

    minvalue

    maxvalue factor value

    offsetvalue unit c-type

    FrontAxleSpd ABS 0 0 251 0.00391 0 km/huint16

    RelSpdFrontLeft ABS 0-7.81257.8125 0.0625

    -7.8125km/huint8

    RelSpdFrontRight ABS 0-7.81257.8125 0.0625

    -7.8125km/huint8

    RelSpdRear1Left ABS 0-7.81257.8125 0.0625

    -7.8125km/huint8

    RelSpdRear1Right ABS 0-7.81257.8125 0.0625

    -7.8125km/huint8

    RelSpdRear2Left ABS 0-7.81257.8125 0.0625

    -7.8125km/huint8

    WheelSpeed

    RelSpdRear2Right ABS 0-7.81257.8125 0.0625

    -7.8125km/huint8

    YawRate ESP 0 -3.92 3.92 0.00012207 -3.92 rad/s uint16Yaw rateYawRate ABS 0 -3.92 3.92 0.00012207 -3.92 rad/s uint16

    LateralAcc ESP 0-15.68716.31240.00048828

    -15.687m/s2 uint16

    Lateralacceleration

    LateralAcc ABS 0-15.68715.687 0.00048828

    -15.687m/s2 uint16

  • Deliverable D9 Dissemination PU 10 (48)

    sensor signal name transmitterinitialvalue

    minvalue

    maxvalue factor value

    offsetvalue unit c-type

    Inclination Inclination TECU 0 -25 25 0.2 -25% uint8

    EngineTorqueModeInstrumentCluster 0 0 15 1 0 uint4

    Enginetorque

    EngineTorqueMode EECU 0 0 15 1 0 uint4

    SteeringWheelAngle ESP 0-

    31.374 31.374 0.000976563-

    31.374rad uint16Steeringangle

    SteeringWheelAngle ABS 0-

    31.374 31.374 0.000976563-

    31.374rad uint16BrkPressFrontLeft ABS 0 0 12.5 0.05 0bar uint8BrkPressFrontRight ABS 0 0 12.5 0.05 0bar uint8BrkPressRear1Left ABS 0 0 12.5 0.05 0bar uint8BrkPressRear1RightABS 0 0 12.5 0.05 0bar uint8BrkPressRear2Left ABS 0 0 12.5 0.05 0bar uint8BrkPressRear2RightABS 0 0 12.5 0.05 0bar uint8BrkPressRear3Left ABS 0 0 12.5 0.05 0bar uint8

    BrakePressure

    BrkPressRear3RightABS 0 0 12.5 0.05 0bar uint8

    Table 2 Truck Vehicle sensors – CAN format

  • Deliverable D9 Dissemination PU 11 (48)

    PART 2 – Demonstration Vehicle SystemDescription

    2 Demonstrator vehicle network architecture

    The demonstrator vehicle network architecture is the network architecture used forexperimental purposes (for projects other than FRICTION). Usually they are equippedwith dedicated CAN buses in order to avoid interferences with normal vehicle functions.Also in this case messages acting on the added experimental CAN buses might be neededfor the FRICTION algorithm operation.

    The truck demonstrator is a vehicle used for several experimental projects and thereforecontains several dedicated experimental CAN buses and rapid prototype units. Figure 3below outlines the already existing CAN buses and RPU:s. In addition to this aFRICTION Gateway RPU (xPC) was added as the processing platform for the VFF, EFFand TFF. Connected to the FRICTION Gateway RPU there is a dedicated CAN bus thatcontains the relevant sensors. The tyre sensor receiver ECU is connected to the secondaryRPU (PDU), which calculates the tyre forces in real time.

    Figure 3 Truck Demonstrator Experimental Electronic Architecture

    CANJ1939

    VehicleGateway / ICA

    (xPC)

    AIDEGateway

    (xPC)

    Lane Tracker

    Maps&Adas

    Decision

    ACC Radar

    ActiveSteering

    Gateway(xPC)

    Side Sensors(Lateral Safe)

    APALACIImage

    Processing

    APALACIRadars

    Blind SpotSensor (right)

    SAFELANE INSAFESGateway RPU

    (xPC)

    FRICTION

    Road eye

    IMU

    Tyre SensorReceiver

    2ndRPU

    IcOR

    Laser scanner

  • Deliverable D9 Dissemination PU 12 (48)

    Figure 4 below outlines the detailed connection scheme for the basic and additionalsensors that was integrated on the demonstrator truck. The xPC has 4 different CAN-ports denoted CAN1, CAN2, CAN3 and CAN4. CAN1 is connected to the Road eyesensor, the PDU and a Camera PC from VTT. CAN2 is connected to the J1939 vehicleCAN bus. CAN3 is connected to the IMU and the CAN4 is connected to the IBEO laserscanner. A laptop is also connected to the FRICTION xPC via a UDP link in order to logdata and download changes in the Simulink implementation of the VFF and EFF runningon the xPC.

    Figure 4 Integration of additional sensors

    IMU

    PDU

    1 Mbit/s CAN

    From tyre sensor receiver

    Road eye

    xPCJ1939

    Laser scanner

    Cameras

    IcORPC

    Laptop

    CAN3

    CAN1

    CAN4CAN2

    UDP

  • Deliverable D9 Dissemination PU 13 (48)

    3 Additional sensors

    3.1 Environmental Sensor – Road eye

    Different surfaces scatter and absorb electromagnetic radiation differently due to forexample differences in surface structure and the scattering medium. Illumination by thesame electromagnetic radiation the reflected response will alter for different surfaceswhich make a classification of the surface possible. This technique is implemented in theRoad eye sensor (see Figure 5) for road condition classification ahead of vehicles.

    Figure 5 Classification boundaries depicted in a 2-dimensional plane with the twowavelengths as axis and a picture of the Road eye sensor

    3.1.1 Road eyeThe environmental sensor Road eye is constructed of two laser diodes and a photo diodecombined with focusing optics. The two laser diodes, of wavelength 1320 and 1570 nm,are used for illumination of the road. Lenses in front of the diodes making the illuminatedspot on the asphalt about 1 cm in diameter at a distance of 1 m. The photo diodemeasuring the reflected intensities, the lens ensures measuring within the illuminatedspots.

    The Road eye sensor output is an intensity measurement for each wavelength,respectively (λ1 and λ2). By plotting the measurements in a plane with the twowavelengths as axis each measurement can be represented by a magnitude and anargument as:

  • Deliverable D9 Dissemination PU 14 (48)

    =+=

    2

    122

    21 arctan, λ

    λλλ ArgMag

    The difference in both magnitude and argument for different road condition makesclassification possible.

    In Figure 5 a classification is depicted for a measurement done on a test track with thefour road conditions dry asphalt, water, ice and snow, each measurement depicted by amarker, the coloured polygons representing the classification boundaries for each roadcondition. Note that the water and ice marker are gather close together, hence theclassification of water and ice is difficult. The measurement depicted in Figure 5 wasdone with an older version of the Road eye sensor which were suggested someimprovements, to get a more valid classification of the surfaces water and ice. Theimprovement was carried out and tested within the FRICTION project.

    3.1.2 Laser diodesLaser diodes are available in a wide range of wavelengths the ones used in the Road eyeis infrared wavelengths, this is due to the fact that within the infrared bandwidth water,ice and snow have distinguished absorptions spectra, i.e. the three surfaces absorbs thelight differently making a contribution for classification.

    Due to the construction of laser diodes the light emitted from the semi conductors islinearly polarized i.e. the electromagnetic waves only fluctuate in one directionperpendicular to the direction of propagation. When electromagnetic radiation (light) isreflected against a surface the polarization of the incident wave is effected due to thestructure and medium. Resent investigations show that to accomplish low probability ofwrong classification between water and ice the state of polarization should beperpendicular (S-) to the incident angle, i.e. S-polarized.

    The first version of the Road eye sensor had the laser diodes mounted with one diode S-polarized and one parallel (P-) polarized, the updated version was updated so both diodeswere S-polarized to improve the classification.

  • Deliverable D9 Dissemination PU 15 (48)

    3.1.3 Physical integration

    Figure 6 Installation of the Road eye

    The Road eye sensor has been tested in trucks in previous projects IVSS RFE, hence themounting was easy. The sensor was mounted in the wheel house in front of the left wheelas seen in Figure 6 This placement ensured measuring 1 m in font of the left wheel. Themounting resulted in an illumination spot of around 0.01 m in diameter on the road foreach laser diode, and the measuring spot of around 0.1 m. The length of the tube holdingthe sensor as depicted in Figure 6 was 0.35 m and the sensor was mounted about 0.6 mabove the road.Investigations within the IVSS RFE project suggested two small alteration of themounting of the Road eye to improve the classification, especially the classification ofwater and ice. These alterations were carried out within the Friction project. The firstalteration was to change the position of the illumination source and the receiver in respectto each other, previously they hade been place parallel with each other but the resultsfrom the IVSS investigation showed that they should be placed with the illuminationsource above the receiver. Hence such a mounting would increase the optical path thelights propagate through the covering medium leading to the covering mediumsscattering and absorption properties get more prominent on the reflected light, decreasingthe probability of wrong classification.The second alteration was the angle of inclination which should be as small as possibleaccording to IVSS investigations. Due to space limitation inside the truck see Figure 6Installation of the Road eye and the risk of pollution of dirt on the sensor, which will bediscussed in a following section, the inclination angle was changed from previous 60° to45°. This was also done to increase the optical path as described above.

  • Deliverable D9 Dissemination PU 16 (48)

    3.1.4 Encapsulation of the sensorThe encapsulation of the sensor is a crucial factor, hence the sensor is a “seeing” sensor itis vulnerable for dirt. To get a good classification the sensor needs to be close to the roadsurface which exposes the sensor for a lot of water and other dirt splashing from the roadon to the sensor.

    Figure 7 The three different dummy tubes mounted on the tractors for the winter expedition toArjeplog

    As shown in Figure 6 the sensor was mounted in a straight cut tube, this mounting havework in previous projects and protected the sensor from water and dirt. For theFRICTION project an investigation of the tubes cutting was implemented. Three tubestwo with the same straight cutting and one with a 45° slope of the cutting. One straightand the 45° tube where mounted at the opposite side on the tractor’s frame behind thecompartment and the front tire, the third tube was placed behind the 45° cut tube. Alltube where mounted with an angle of 45° to the ground. The mounting took place beforethe travel of 1500 km from Gothenburg to Arjeplog in order to investigate the pollutionfactor inside the tubes.

  • Deliverable D9 Dissemination PU 17 (48)

    3.2 Environmental Sensor - IcORThe IcOR application is a camera system which includes two detectors, one for horizontaland the second for capturing the vertical polarisation reflection. The system also analysesgraininess of the road surface in order to improve performance of distinguishing wet andicy road surfaces. The system provides as an output whether the road ahead is icy, wet,snowy or if the asphalt is dry.

    ROAD SURFACE CLASSIFICATION

    0,4

    0,45

    0,5

    0,55

    0,6

    0,65

    0,7

    0 2 4 6 8 10 12

    Polarisation difference

    Gra

    inin

    ess

    SnowWet/SlushSnow/Ice

    Figure 8 The classification results of the IcOR road surface monitoring application

    The IcOR system bases on cameras and is therefore capable of analysing the road area asfar as 50 - 100 m ahead of the vehicle. The system does not contain illuminationequipment itself but it is rather optimised to operate close to the near infrared band whichmakes possible to utilise the vehicle’s head lamps while driving in dark conditions.

    3.2.5 Equipment

    The IcOR hardware consists of camera unit, polarization filters and CPU. There are twoimage detectors in the camera which has been bought from Videre Design LLC, U.S.A.and its model is STH-DGSG-6cm. The system is originally designed for low cost stereovision applications but it is feasible solution for performing synchronized imagecapturing in automotive application since the design is compact and takes power directlyvia the Firewire cable. The camera cell is Micron MT9V022 which is particularlydedicated for automotive products.

  • Deliverable D9 Dissemination PU 18 (48)

    In addition to the camera hardware, polarization filters have been installed front of theoptics. The filters transmittance is approx. 30% in visible spectrum 400 – 700 nm andtheir polarization efficiently is 95 %. The visible spectrum is here utilized since thesystem is expected work also during night time when the vehicle’s head lamps are turnedon.

    The camera unit is connected with a Firewire cable to the computing unit. In thedemonstration vehicle CPU is a laptop with the MS Windows XP –operating system. TheIcOR software runs in the laptop and via the implemented CAN interface the road stateinformation is transmitted to the in-vehicle computing unit.

    3.2.6 Physical Integration

    Figure 9 Installation examples. On left side the truck and on right side the passenger car installation

    The IcOR system was installed behind of the windscreen since it offers water proof placefor the camera and the vehicles own wipers maintain the optical path of the camera clear.The installation when prepared to the truck test runs in Arjeplog, Sweden on March 2008was made to the dashboard of the vehicle. The image processing PC was installed to therack where the other vehicle-PCs exist as well.

    In the truck implementation the first generation camera setup was used which wasphysically quite big. To the IKA’s Audi demonstration the second generation camerainterface was elaborated and it enabled opportunity to install the camera side of theinternal mirror where for example the lane trackers normally locate in passenger cars.

  • Deliverable D9 Dissemination PU 19 (48)

    3.3 Environmental Sensor – IBEO Laser ScannerThe Ibeo LUX Laser scanner was developed under the requirement to simultaneouslysupport multiple ADAS applications like e.g. ACC Stop & Go, Automatic EmergencyBraking, Pre-Crash, Pedestrian Protection. These applications require accurate detectionand tracking of moving and static objects in the environment of vehicle.

    The acquisition of information about weather conditions is not a major requirement for thoseapplications; instead they demand robustness against adverse weather conditions like rain, snow andfog. Within the FRICTION project the sensor embedded algorithms have been enhanced so that it isalso possible to detect and classify measurements on precipitation related targets like raindrops or

    snowflakes.Figure 10 shows a visualization of a laser scanner measurement recorded during snowconditions. Visualized are the raw measurements. Each dot represents the distance of abackscattered laser pulse. The snow density, respectively the amount of precipitation canbe related to the number and the distribution of measurements classified as snow in adefined area of the laser scanners field of view. The classification of the singlemeasurements is based on the signal characteristic of the backscattered signal and theirspatial distribution:

    • Measurements on snow and rain are only detectable in a certain measurementrange of ~0.5 .. 12m.

    • Measurements on snow and rain are characterized by a significantly lower energyin the backscattered signal.

    Figure 10 detected snow in Laser scanner measurements

  • Deliverable D9 Dissemination PU 20 (48)

    The following figure shows the output of the precipitation estimation in three differentsnow conditions.

    Figure 11 Output of Laser scanner precipitation estimation

    The following figure shows the detection, and classification of spray during a motorwaytest drive with spray measurements marked blue, object measurements are marked red.

    Figure 12 Laser scanner spray detection and classification

  • Deliverable D9 Dissemination PU 21 (48)

    The main benefit of the laser scanner in the context of the FRICTION project is thecapability of detecting precipitation – relevant for friction determination, and objects -relevant for ADAS applications - simultaneously, and to distinguish between both typesof measurements.

    3.3.7 Physical Integration

    Figure 13 Physical integration of the IBEO Lux laser scanner on the truck

  • Deliverable D9 Dissemination PU 22 (48)

    3.4 IMUThe Inertial Measurement Unit (IMU) is developed by Continental AG (former SiemensVDO) to provide key data for vehicle dynamics control systems. The IMU applied in theEU-project contains three Yaw Rate Clusters (YRC). The three same YRCs are mutuallyperpendicularly mounted in the IMU housing (see Figure 14 below). The YRC isdesigned by using MEMS (Mirco Electromechanical sensors) technology to provide anangular velocity and two accelerations. It indicates that the combined IMU can providethe three angular velocities and three accelerations of vehicle body, i.e., yaw, roll andpitch rates, and longitudinal, lateral and vertical accelerations. By using the IMU, the 3-dimensional movement and 3-dimensional attitude of vehicle body can be preciselyestimated. The integrated IMU in a single unit has been applied in ECU in 2007.

    Figure 14 The IMU applied in the EU-project contains three Yaw Rate Clusters (YRC).

  • Deliverable D9 Dissemination PU 23 (48)

    Table 3 below outlines the IMU sensor signal definitions.

    Sensor Signalname

    Initialvalue

    Minvalue

    Maxvalue

    Factor value(in units perdigit)

    Offsetvalue

    Unit Size Signalage

    longitudinal acceleration ax 0 -32 +31.984 0.015625 -32 m.s-² 12 bits 10 mslateral acceleration ay 0 -32 +31.984 0.015625 -32 m.s-² 12 bits 10 msvertical acceleration az 0 -32 +31.984 0.015625 -32 m.s-² 12 bits 10 mspitch rate PitchRate 0 -128 +127.938 0.0625 -128 °.s-1 12 bits 10 msroll rate RollRate 0 -128 +127.938 0.0625 -128 °.s-1 12 bits 10 msYaw rate YawRate 0 -128 +127.938 0.0625 -128 °.s-1 12 bits 10 ms

    Table 3 – IMU Sensor Signal Definitions

    3.4.8 Physical Integration

    The IMU was pretty easy to install even though it could not be installed at the centre ofgravity of the truck.

    Figure 15 Installation of the IMU

  • Deliverable D9 Dissemination PU 24 (48)

    3.5 Tyre SensorFigure 16 below shows the truck tyre sensor system architecture with the antenna boxmounted in close proximity of the front left wheel hub centre, the Tyre Sensor receiverECU containing the receiver unit for the wireless link and the Secondary RPU (PDU) forreal time calculations of tyre forces.

    Figure 16 – Truck tyre sensor system architecture

    3.5.1 Tyre sensor systemTyre sensor system needs particular care due to the nature of data coming from it. Thesystem is made up by a receiver unit for each wheel. The outputs are available ondedicated point-to-point high speed (1 Mbps) CAN Buses. Such buses are completelysaturated and cannot be directly connected to the FRICTION Bus. For that reason aSecondary Rapid Prototyping Unit will be provided in order to pre-process signal andoutput synthetic tyre sensors data.

    The optical tyre sensor is well known from the APOLLO EU-project. An optical tyredeformation sensor was developed and tested in this EU founded project. This sensormeasures the displacement of the tire contact patch relative to the rim. Set up andmeasuring principle is discussed shortly in the following.

    The Sensor consists of a PSD (Position Sensitive Device), a lens and a light source on theinner liner of the tyre. The diode is glued to the inner liner. The PSD chip and the lens arelocated in a housing directly on the rim.

    1 Mb/s CAN

    Antennabox

    TyreSensor

    Receiver

    12V

    SecondaryRPU (PDU)

    12V

  • Deliverable D9 Dissemination PU 25 (48)

    IR-Diode

    LinsePSD

    IU

    ∆s

    Z

    Y

    X

    X

    Y

    Z

    I2

    I3

    I4

    I1

    IR-Diode

    LinsePSD

    IU

    ∆s∆s

    Z

    Y

    X

    X

    Y

    Z

    I2

    I3

    I4

    I1

    Figure 17 - Axial optical sensor on inner liner [source: APOLLO project]

    As visible in Figure 17 the light beam, which is emitted by the IR-diode, is focused onthe PSD chip. The centre of the light spot on the PSD lens is responsible for the 4currents that can be measured at the edges of the PSD chip. Vertical deflection of the tyrecauses a higher intensity of the light beam hitting the PSD. This results in an increasingoverall current I1 + I2 + I3 + I4 with the same ratio at all corners of the chip. Amovement of the light beam in lateral of longitudinal tyre direction changes the current atthe edges with different ratios.

    The relation between movement of the spot on the PSD chip and the current at the fourborders can be described mathematically and can be used to get the final displacement.Appropriate tyre models can be used to calculate global tyre forces from the displacementsignals.

    Each tyre sensor provides data on five channels:· Four currents (I1 to I4) from the PSD-chip (Position Sensitive Device)· Information on data transmission: defective data from PSD will be skipped

    As these signals represent pure raw data from the tyre sensor, a pre-processing is requiredIn a first step the deflection (longitudinal, lateral and vertical) of the tyre belt in relationto the rim is calculated using the four currents I1 to I4 from the PSD-chip. Potentially it ispossible to use the deflection signal of the tyre belt in order to gain information about theroad surface state.

    In a second step the deflection data of the tyre belt is used to calculate the tyre forcesacting in the tyre contact patch. The tyre force information can be used to very preciselyobserve the current state of vehicle motion. The exact knowledge of the tyre forcesenables the possibility of a wheel individual friction used and in some cases frictionavailable calculation.

  • Deliverable D9 Dissemination PU 26 (48)

    The data from the secondary RPU is read from the CAN-BUS (1Mbit/s) message 8(length 8 bytes). The values are between 0 and 4095 (12 bits). The update rate isapproximately 5200Hz (radio sends new message), but can be down sampled to ~3000Hz if needed. Message Composition is shown in Table 4. All signals are unsignedintegers.

    Table 4 - Message Composition

    3.5.2 ERROR-CORRECTIONIf CRC-value is non-zero, there is a disturbance in radio transmission. In this case, the “Ifaction”-subsystem keeps the existing value, until CRC is zero again. In the second phasein this subsystem, the signal values are compared to high and low parameters in order toevaluate if they are within reasonable range. If not, the previous value is kept. Note: thereare two CRC:s, one from the CAN and another one from the radio. Only CRC-information from the radio transmission is considered here, because radio transmissionerrors are more common than CAN-bus errors.

    3.5.3 CONTACT RECOGNITIONThe contact is not directly recognised. Instead the upright position of the sensor isdetected. The inductive sensor (magnetic pick-up) is installed into the rim with sameposition as the optical sensor. Secondly, a magnet is installed into the suspension of thevehicle (or to the test rig). When the sensor passes the magnet, a voltage output can bemeasured. The offset is removed from the signal (Figure 18). The zero crossing point hasto be detected, but only when the sensor is close to the magnet (excluding zero crossingsinduced by noise). This is realised with a relay-block in Simulink. When the certainthreshold (“ind_on”) is exceeded, the relay is switch on and the zero crossing is activated.After the certain threshold (“ind_off”) the relay is switch off. When the relay is on andsignal first time reaches zero or negative, the sensor is considered to be upright. Note:The pole of the magnet has influence for the signal. If the pole has changed, the signalgoes down first and the up to zero crossing. (This was the case with truck rim in Aachen).

  • Deliverable D9 Dissemination PU 27 (48)

    The problem can be avoided by changing sign of the parameters „ind_on , „ind_off , andsearch for first positive value (instead of negative) when relay is on.

    Figure 18 Contact recognition relay

    If the sensor position is not properly recognised, none of the further outputs are real andthe tyre force estimates are NOT reliable.

    3.5.4 FORCE CALCULATION-BLOCK

    3.5.4.1 ROTATIONAL VELOCITY-SUBSYSTEMRotational velocity is needed in force estimation algorithm. It can be also implemented tojudge the plausibility of the contact recognition, because rotational velocity from tyresensor can be compared to the ABS-sensor values. The time difference between twosequential “sensor up” peaks is calculated and rotational velocity follows as:

    3.5.4.2 LONGITUDINAL FORCESeveral different algorithms capable for longitudinal force estimation were studied. Manyof them were powerful in some simple conditions, but the following one performs quitewell even with some disturbances. The behaviour of longitudinal signal is shown inFigure 19. The peak values change, but on the other hand the signal level is raised aswell. Thus, the mean value of longitudinal signal for an independent rotation iscalculated. Thus the longitudinal force can be calculated:

  • Deliverable D9 Dissemination PU 28 (48)

    In the Simulink-model the “mean (x)” is realised by calculating sum of all x-signal valuesand dividing that with the number of values (for one rotation).

    Figure 19 Longitudinal signal when braking applied at 13s

    3.5.4.3 LATERAL FORCEThe optical sensor measures the movement of the light spot on sensor surface, not theactual movement of the LED (Figure 20). During the contact, the LED moves closer tothe sensor and the lateral position of the light spot (reflected from the lens) is moving,even though the LED stays still laterally (Figure 21).

  • Deliverable D9 Dissemination PU 29 (48)

    Figure 20 - Lateral signal before compensation

    Figure 21 - The influence of lateral LED position during contact

    This phenomenon is removed from the signal in “Compensate y”-subsystem.

  • Deliverable D9 Dissemination PU 30 (48)

    and the area of deflection is:

    where the influence of velocity is removed by multiplying with rotational velocity.Finally the equation force reads:

    where an additional compensation term from xgap is found to be reasonable (explained in“vertical force”-chapter. Note: It seems that the “Compensate y”- subsystem is notnecessary for the truck sensor, because of the greater distance of the LED to the sensor(in relation to lateral movement).

    3.5.4.4 VERTICAL FORCEThe intensity signal is natural starting point for vertical load calculation and itsperformance is really good when no other forces exist at the same time. The longitudinalsignal includes in-formation about the contact length, which correlates strongly withvertical load as well. It is proposed to be much more robust against the other disturbancesthan the intensity signal.

    Figure 22 - Longitudinal signal behaviour on wheel loads 5000, 3000, 1000NThe vertical force is mainly calculated from gap between the minimum and maximumvalue (longitudinal signal) of one rotation:

  • Deliverable D9 Dissemination PU 31 (48)

    In practise, the integrator is reset when the sensor is up, and the following triggeredsubsystem holds the existing value until new estimate is available. The cz equals verticalforce gain in the model and vertical force estimate reads:

    The longitudinal signal is influenced by lateral and longitudinal movements as well andcompensation terms are in parentheses. The yA and xmean are calculated in lateral andlongitudinal force blocks. Note: Because maximum and minimum values are used forxgap, the noise has quite remarkable influence for the estimation and parameters as well. Itis possible that some filtering would improve the performance of vertical force estimate.

    3.5.5 Implementation for Tyre Force CalculationsThis chapter gives an explanation on the implementation of the tyre force calculationsperformed in the Secondary RPU (PDU).

    3.5.5.1 HardwareThe PDU features a Star12-processor called MC9S12DG128 with two CAN controllers,labelled CAN0 and CAN4. The processor features 128KB flash memory and 8KB.

    Figure 23 – Secondary RPU (PDU) Hardware

  • Deliverable D9 Dissemination PU 32 (48)

    3.5.5.2 Software ToolsDevelopment studio used was Codewarrior Development Studio.

    3.5.5.3 The algorithm and its implementationFor the actual algorithm please refer to chapter 3.5.1.

    The algorithm was implemented using ANSI-C. The implementation uses the mainfunction only for initialization and startup code. The rest is implemented using interruptsof different priorities.

    The important interrupts and functions used are:void interrupt 7 RTI(void)This interrupt has the highest priority of all the interrupts and is called approximately 976times per second (976Hz). It’s used for measuring time via the counter1 global variable.The RTI interrupt is also used for outputting a CAN message once per second that reportsthe amount of CRC errors in the radio packets received from the tyre sensor.void interrupt 54 CAN4Rx(void)The CAN4 is set up to run at 1Mbps. This interrupt is called when a new CAN messagehas been received in the CAN4 controller. The CAN4Rx interrupt service routine readsthe desired parts of the packet and then calls the TyreForceCalc(). Look furtherdown for its description.void interrupt 39 CAN0Tx(void)The CAN0 is set up to run at 250Kbps. This interrupt is triggered everytime at least oneof the three hardware slots in the CAN0 controller is empty and therefore ready foraccepting a CAN message for transmission. Sending CAN messages is implementedusing a circular buffer. The CAN0Tx service routine reads the CAN message data fromthe circular buffer at a position indicated by the CANQReadPos global variable, andwrites the data to one of the free hardware slots of the CAN0 controller. It then marks theslot as full, which automatically starts transmission. If no CAN messages were availablein the circular buffer, then transmission interrupts is turned off in the CAN0 controller toprevent further interrupts. This is later turned on again by the CAN0Transmit()function.int CAN0Transmit()This function is called to equeue a CAN message for transmission. The contents of themessage are copied to a free slot in the circular buffer pointed to by the variableCANQWritePos, which is then increased to point to a new position. Before exiting thefunction enables transmission interrupts, so that the CAN0Tx interrupt gets called. If nofree slot was available in the circular buffer the message is not copied and an error codeis returned.void TyreForceCalc(unsigned char *data)This is the function that implements the actual Tyre Force Calculation algorithm. It isstructured as a state machine and expects a new CAN message from the tyre sensor eachtime it is called. Note that TyreForceCalc() is “fall-through”, so it features no busy-wait loops anywhere. Since the tyre sensor sends up to 5200 messages per second, the

  • Deliverable D9 Dissemination PU 33 (48)

    total time available for the CAN4RX interrupt to execute, including calling theTyreForceCalc(), is limited to 1/5200 = 180us before the next CAN message isexpected. This puts some great limits on the tyre force calculations in the last states of thestate machine, but it is solved by “pipelining” (splitting up) the calculations in severalstates. In that way every state is guaranteed to execute within 180us.

    The algorithm uses a mixture of fixed point and floating point calculations. Since theStar12 processor doesn’t have native floating point instructions, these have to besimulated which takes a lot of time for the processor. Fixed point is used for allcalculations during the earlier states of the state machine to make sure the processor getsit done within the 180us. During the final states where the friction forces are calculatedfloating point is used because both precision and ability to store large values is required.

    First the algorithm retrieves the required data from the CAN message according to Table4 and stores it in the variables CRC, x, y, z and ind. Then the data is compared tocertain limits. If any limit is exceeded or if the CRC is not zero, the data from the lastCAN message is used instead. This is done every time the tyre force calculationalgorithm is called regardless of the state it is in.

    The rest of the TyreForceCalc() consists of the actual state machine.

    The first state CALC_MEANIND is used for calculating a trustworthy mean value of theinductive voltage. After 5200 received CAN messages and with enough fluctuations inthe inductive value the mean inductive value is considered useful. The mean value isrequired later to detect the peaks in the inductive voltage, which indicate that the tyre hasrotated 360 degrees.

    The second state, FINDFIRSTPEAKS_RELAYOFF, looks for the first pair of voltagepeaks to be received. We need this because most of the friction force calculations aredone using data collected over an entire rotation of the tyre. When the inductive voltagereaches above or below a certain threshold the algorithm assumes that a pair of voltagepeaks is coming. It’s possible to change whether the algorithm should look for a voltageincrease or decrease by using the two defines called POSNEG and NEGPOS. Which one touse depends on how the magnet was mounted on the rim of the tyre.

    The third state, FINDFIRSTPEAKS_RELAYON, detects the transition from positive tonegative (or vice versa) in the inductive voltage. This indicates that the tyre has rotatedone complete turn and that from now on we should collect the necessary data for x, y, andz to do the tyre force calculations in the later states.

    The fourth state, RELAYOFF, is almost a copy of FINDFIRSTPEAKS_RELAYOFF,since it also looks for the inductive voltage to reach above or below the same threshold.But meanwhile it also performs some calculations like x_mean, x_min, x_max andy_mean on top of the same old inductive mean value calculation.

    The fifth state, RELAYON, is where all the action starts. Here the algorithm looks for thenegative-to-positive transition (or vice versa) like in FINDFIRSTPEAKS_RELAYON.When this is found some of the previously gathered information, like x_mean, x_min,x_max and y_mean is used for calculating a few floating point values that are used in thelater states.

  • Deliverable D9 Dissemination PU 34 (48)

    The three last states called RELAYON_pipe1, RELAYON_pipe2 andRELAYON_pipe3 is just a way to spread out the intensive tyre force calculations overtime, because they use floating point. Since no new pair of peaks is expected in the nextthree CAN messages from the tyre sensor (this would require the truck to travel at supersonic speed), this should be a safe way to do it.

    3.5.6 Physical IntegrationFor the optical tyre sensor a new rim design needed to be developed because all previousoptical tyres sensor systems had been adapted for car rim and tyres. One obstacle was thatfor truck rims it was not possible to use dividable rims as was the case for e.g. theAPOLLO project. Another problem was the physical size of a truck tyre compared to acar tyre. In a truck tyre, the PSD will physically be at a grater distance from the LED,which will lead to a less optical power and a weaker signal.

    In order to overcome those hurdles it was decided to design a manufacture a new rim forthe truck. By making a hole in the rim and weld a flange in to it, it was possible to makethe installation of the LED through that hole, after the tyre was mounted in the rim. Italso made it possible to make a sensor module in order to decrease the distance betweenthe LED and the PSD.

    A special sensor module was designed to interface the rim .The sensor module can bequickly removed and reinstalled without special tools.

    Figure 24 Sensor module installation to the test rim (1. rim, 2 flange joint welded to the rim, 3. sensormodule 5. sealing O-ring)

    The tyre is installed before the sensor module and the LED is glued into the inner liner ofthe tyre (Error! Reference source not found.). The LED is powered with wires fromthe sensor module. In addition to the sensor module, a magnetic pick-up sensor isinstalled on the inner edge of rim (Error! Reference source not found.). The magnet isinstalled on the suspension to indicate the upright position of the sensor. This enablesvery accurate information on the sensor rotation angle and the data is certainly on thesame time axis as the actual tyre sensor data, due to same signal path.

  • Deliverable D9 Dissemination PU 35 (48)

    Figure 25 Cross-section of optical tyre sensor for truck

    Figure 26Optical tyre sensor components, magnet and magnetic pick-up sensor

  • Deliverable D9 Dissemination PU 36 (48)

    Figure 27 Making of the hole in the rim and flange welded in to the hole.

    Figure 28 The sensor module mounted on the truck rim

    Figure 29 The LED mounted on the inner liner and the lid mounted along with inductive sensor

  • Deliverable D9 Dissemination PU 37 (48)

    Additionally, the antenna box with the receiver also needed to be installed on the truck.This was a bit cumbersome because it was difficult to find the sweet spot, where aminimum number of CRC errors occurred regardless of the position of the wheel.

    Figure 30 Installation of the antenna box

    4 HMI HardwareFor the commercial vehicle demonstrator the dashboard of the existing vehicle is used forvisualization.

    Figure 31 The configurable instrument cluster in the truck demonstrator

  • Deliverable D9 Dissemination PU 38 (48)

    5 FRICTION estimation systemThe FRICTION estimation system implemented in the Volvo FH12 is based on in-vehicle sensor together with one tyre-sensor, an IMU and an environmental sensor (Roadeye). It executes in the FRICTION xPC gateway according to Figure 3. Informationabout the system is listed in the following table.

    Input sensor Correspondingpreprocessed signal

    Unit Comment

    steering wheelangle sensor

    SteeringWheelAngle rad

    lateral accelerationsensor

    LateralAcc m.s-2

    Siemens IMU LateralAccLongitudAccVerticalAcc

    YawRatePitchRateRollRate

    m.s-2

    rad/s

    yaw rate sensor YawRate rad/swheel speed sensors

    (one for eachwheel)

    WheelSpeed km/h

    brake pressuresensors

    (one for eachwheel)

    BrkPressinj bar

    tyre sensor - - See chapter 3.5environment sensor

    (Road eye)- - See chapter 3.1

  • Deliverable D9 Dissemination PU 39 (48)

    PART 2 – Safety Application Description6 PReVENT Applications within the FH12 truckAs mentioned before a Volvo FH12 is used as a truck demonstrator. From production thetruck is equipped with ABS, ESP and ACC systems. It also supports a number ofapplications developed in different PReVENT subprojects. These applications are:

    • Active Lane Keeping Support• Collision Mitigation by Braking• Start Inhibit• Curve Speed Warning• Lane Change Assistance• All Around Warning

    The truck has been equipped with a number of environmental sensors in order to supportthe applications mentioned above:

    • 1 long range forward looking radar• 2 short range forward looking radars• 3 short range side-looking radars• 1 camera for the blind spot in front of the truck• 1 camera for the blind spot to the right of the truck• 1 lane tracker camera• 1 laser scanner (225 deg FOV)

    7 Truck Demonstrator Safety Application

    The chosen safety application for demonstration of the benefits of having informationabout the road friction is the CMbB safety function, developed within the PReVENT sub-project COMPOSE [COM08].

    The COMPOSE project developed safety functions for the protection of road users andfor mitigating the consequences of traffic accidents. The COMPOSE demonstrator truckof VTEC focussed on rear-end collisions in highway scenarios.

    For the COMPOSE VTEC demonstrator platform, a laser scanner located at the frontlower left corner is employed as perception system.

  • Deliverable D9 Dissemination PU 40 (48)

    Figure 32 VTEC Demonstrator with sensor locations. Red is COMPOSE-specific and greenAPALACI-specific.

    The perception system provides the decision system with the objects detected ahead ofthe own vehicle. Based on this information, an assessment is performed whether acollision has become unavoidable. In this case, the demonstrator vehicle brakes fullyautomatically to reduce impact speed and thus crash energy. For this purpose, both thebrake system as well as the engine is controlled electronically by the assessment unit.The demonstrator truck is also shared with other sub-projects within the IP PReVENT.

    For the FRICTION commercial vehicle demonstrator, it was not possible to use the laserscanner. Instead, the ACC radar was employed as perception system for the CMbBfunction.

    Figure 33 FH12 ACC radar

    Kamera

    Radar

    LaserScanner

  • Deliverable D9 Dissemination PU 41 (48)

    With friction information available, the CMbB braking distance can be calculated moreprecisely. Figure 34 below describes the relation between the braking distance and thefriction coefficient.

    Prediction of braking distance for stationary obstacle

    v

    Braking distance:

    Influence ofmax. friction available:

    Max. vehicle velocity(range of radar system: 160 m)

    0

    50

    100

    150

    200

    250

    0,1 0,3 0,5 0,7 0,9 1,1Max. friction available [ ]

    Vel

    ocity

    [km

    /h]

    Incline of 6%

    gµvs

    x

    B

    ⋅⋅=

    µ1

    Figure 34 Prediction of minimum braking distance

    The braking distance (S) of a vehicle can be calculated by the velocity (v) and themaximum friction available (µ). Therefore besides the speed, the most influencingparameter on the braking distance is the maximum friction available (µ).

    For simplicity, the COMPOSE CMbB application was running on the same xPC as thefriction estimation algorithm. The friction estimation application forwarded the frictioninformation to a modified version of the CMbB application, that could recalculate thebraking distance according to Figure 34 above.

  • Deliverable D9 Dissemination PU 42 (48)

    8 HMI Application

    8.1 Previous studies

    8.1.1 Focus on slipperiness. Attitudes towards slippery road informationsystems (VTI-2008)

    A group of 4 truck drivers and 4 bus drivers took part in a focus group in December2007. Some of the questions asked were:

    How big is the problem with slippery roads, compared to other traffic problems and howdifficult is it to drive on a slippery surface?

    A common view amongst the drivers is that the “black ice” (slippery asphalt) is the mostannoying one. If there’s no ice or snow on the road you don’t slow down because of whatyou see. There are a lot of variations in the surface over a long stretch; therefore the gripcan surprise you (this depends on the road administrations work which differs in differentcounties). The most unmanageable ice appears around 0 degrees Celsius. That is the mostuncontrollable slipperiness.

    A slippery surface is perceived very differently depending on what vehicle you’redriving. Many of the drivers are also worried about driving downhill and the fact thatthey can cause a lot of damage if something happens. The loading of the truck is alsovery important. You have to even out the weight on all the tires to get good grip.

    How do you receive information about the road conditions today?

    The most common way for the drivers to receive information about the road conditions isby calling colleagues. They also use the information from the radio but feel that the co-worker’s information is more precise and up to date.

    Many drivers know the temperature outside as a very good help, especially when theweather changes fast. They rely on the road temperature and make their own adjustmentsaccording to their knowledge and experience.

    What kind of demands would you have on a friction detection system, and how should theinformation be presented to the driver?

    Many drivers are afraid to be smothered with information and it can easily becomeannoying. When it comes to presentation they all prefer a big simple symbol, and maybea voice message. Sometimes it is hard to read, that’s why a symbol is easier to interpretfast. It’s important that the system doesn’t lead do an increased risk taking.

  • Deliverable D9 Dissemination PU 43 (48)

    Would you trust a friction system?

    “A driver never trusts any systems” they say. They will never stop calling each otherbecause that’s the most reliable information but a warning might be a good complement.

    8.1.2 Effects of weather-controlled variable message signing on driver behaviour(VTT -2001)

    The aim of the study was to investigate the effects of local and frequently updatedinformation about adverse weather and road conditions on driver’s behaviour. Theinformation was transmitted by several types of variable message signs (VMS).

    Operators manually controlled the signs. They classified the road surface conditions inthree categories. Good, possibly slippery and verified slippery. One of the evaluationsused road condition sign (shown below) that were turned off, in steady mode or inflashing mode together with other information like recommended headway depended onvehicle length, driving speed and road surface condition. Speed limit at the site in Finlandwas 80 km/h and the study was conducted during two winter periods.

    8.1.2.1 ResultThe system proved most effective when adverse weather and road conditions were noteasy to detect. Most drivers accepted lowered speed limits and found variable speedlimits useful.

    The flashing mode of the slippery road condition signs affected behaviour more than thesteady mode. The effect was greater and lasted longer.

    89 % of 114 interviewed drivers indicated that the sign had effect on their behaviour. Inproportion to driving on “black ice”, which was 99 %. Drivers reported that slippery roadcondition sign influenced driving speed, particular in curves. Other effects where morefrequent monitoring of oncoming vehicles than usual, concentrating more on one’sdriving and monitoring road surface.

    The test was emphasized especially in black ice conditions.The slippery road condition sign and minimum headway sign decreased the mean speedof cars travelling in free flow traffic, by 1-2 km/h.

  • Deliverable D9 Dissemination PU 44 (48)

    The main human error leading to increased risk in winter is driver’s poor ability torecognize slipperiness and to adapt their behaviour to weather conditions. On slipperyroad surface, only 14 % of Finnish drivers estimate the road to be slippery and more thanhalf considered the friction normal. The average speed on slippery road surface are about4 km/h lower than in good winter conditions but that may not enough.

    8.1.2.2 Decision makingOne important fact to have in mind is that driving is not the primary task at all times. Thedrivers have several parallel goals while driving and perceiving slipperiness is difficultsince there may be minimal visual cues to indicate the hazard. An operator revising ahypothesis is generally conservative and not extracting as much information as necessary.The concept of anchoring refers to difficulty for human operators in changing an initialhypothesis in line with subsequent sources of evidence, rather the opinion shifts onlyslightly. This makes it understandable why behavioural adaptation to slippery conditionsis difficult for drivers. Another aspect is the speed. Drivers get used to the speed, and thedecrease of speed feels subjectively greater than it objectively is.

    8.2 Problems concerning friction value and HMI presentationThere are a few issues that affect the outcome of the HMI. First of all, the friction valuewill not be available at all, most of the time. This is probably the biggest issue to address.One way of solving this is to separate the Road Eye information and the information fromthe other sensors. Since the driver have to accelerate or brake for the sensors to registersomething you can drive on ice, without any sensors detecting this, apart from the RoadEye (at least if you drive from asphalt to ice). The Road Eye may recognize differences inthe surface, but that doesn’t mean it’s slippery. Maybe the Road Eye should be used as avalidation of values from the other sensors.

    To visualize this, the HMI could consist of two icons (see below). The “Informationicon” (info from Road Eye) should indicate that “if you make any sudden moves, brake oraccelerate the vehicle might slip” and the “Warning icon” (from other sensors) indicatesthat “You are slipping”.

    Figure 35 ISO symbols

  • Deliverable D9 Dissemination PU 45 (48)

    There are different kinds of icons available. The ISO standard is shown (Figure 35) in thetwo modes, information and warning. Another is the one shown on traffic signs (Figure36). The warning triangle is a well known indicator of danger which is preferable in astressed situation when the driver just glimpse shortly on the dashboard. Also here, theuse of different levels of warning is shown. The three (or two) steps gives the driver agood reference of the warning level which makes it easier to rate its graveness in a shortamount of time.

    Figure 36 Traffic signs

    As a complement to the warning triangle, one could use an exclamation point for theinformative warning (Figure 37).

    Figure 37 Informative with exclamation

    A more illustrative way to warn a driver is tire mark behind the truck icon (Figure 38).

  • Deliverable D9 Dissemination PU 46 (48)

    Figure 38 Tire marks

    The Road eye detects a bit further ahead on the road than the other sensors. What Roadeye sees can, by the vehicles sensors, later be confirmed when the tires reach the samearea. If this confirmation is made, a warning appears. Then the HMI continue to warn aslong as these two parameters correlate. But then again, if you drive on a snowy road theRoad eye will detect something all the time, and a “Information icon” might loose itssignificance after a while and make the driver inattentive to variations in that value. Toavoid this, a sound may be a good complement to an icon shifting from information towarning. But if this is the case every two minutes it gets quite annoying.

    8.3 HMI simulationThe simulation of HMI for test at Hällered test site includes two icons. An informative(orange-yellow) and a warning (red with triangle).

    Figure 39 Final version

  • Deliverable D9 Dissemination PU 47 (48)

    8.3.3 ConcernsFirst and foremost these icons need to be tested by users to verify their appearance. Sincethe icons are shown in the display behind the steering wheel and is quite small, a soundcould be a good complement to catch the drivers attention even more.

    A driver’s sight in daylight is better than the Road eye, so an “Information icon” may beunnecessary. At night, on the other hand, it can be a useful complement. Maybe thereshould there be a Day- and a Night Mode.

    8.4 Other areas of useThe value of friction might also be interesting for the road administration. If they wherepresented the value as well as the position of the vehicle they might provide a moreefficient and sufficient maintenance of the roads. They could also have an economic gaindue to the more precise information that the value gives, unlike the information they relyon today in ways of predicting weather forecasts and act on experience. On the otherhand the road administration needs information in advance to apply the right amount ofwork. The information from friction could instead be a measure on how well they haveaddressed the problems. If they get indications that a newly salted road is slippery,something is wrong. All efforts can be evaluated afterwards.

    The information could also be presented to drivers around the area, or people in general.Those who are planning a day on the roads might consider their route or speed if theyknew someone before them have detected a significantly low value of friction.

    A problem though is that the road friction value for one vehicle is not directlycorresponding to the road friction value of another vehicle which means that theinformation needs to be translated in to some sort of road condition information in aformat that could be useful for other vehicles.

  • Deliverable D9 Dissemination PU 48 (48)

    PART 4 – Appendices

    References

    [KIR08] Kircher, K. et al (2008). Focus on slipperiness. Attitudes towards slipperyroad information systems (VTI notat 8- 2008)

    [RÄM01] Räme, P (2001). Effects of weather-controlled variable message signing ondriver behavior (VTT Publications).

    [KLI95] Klinkner, W.: Electronic Stability Program: The new Active Safety System ofMercedes-Benz. EuroMotor: Vehicle-Vehicle and Vehicle-RoadsideInteraction, Institut für Kraftfahrwesen Aachen (ika), Aachen, 1995.

    [PAR00] Park, K.; Heo, S.-J.: Design of a Control Logic for Improving VehicleDynamic Stability. 5th Intern. Symp. on Advanced Vehicle Control (AVEC),Ann Arbor, USA, 2000.

    [PAU98] Paul, J.: ESP – Elektronisches Stabilitäts-Programm – Ein aktivesFahrsicherheitssystem für einachsig- und allradgetriebene Mercedes-Fahrzeuge. Vekehrsunfall und Fahrzeugtechnik, Heft 4, 1998.

    [SFE01] Sferco, R.; Page, Y.; le Coz, J.-Y.; Fay, P. A.: Potential Effectiveness ofElectronic Stability Program (ESP) – What European Field Studies tell us.17th Intern. Techn. Conf. on Enhanced Safety of Vehicles (ESV), Amsterdam,Niederlande, 2001.

    [COM08] D51.82 COMPOSE deliverable: “Functional samples of the safetyapplications”