Development of an automated testing system.pdf

ORIGINAL ARTICLEDevelopment of an automated testing systemfor vehicle infotainment systemYingping Huang & Ross McMurran & Mark Amor-Segan & Gunwant Dhadyalla &R. Peter Jones & Peter Bennett & Alexandros Mouzakitis & Jan KielochReceived: 18 December 2009 /Accepted: 12 March 2010 /Published online: 15 April 2010# Springer-Verlag London Limited 2010Abstract A current premium vehicle is implemented witha variety of information, entertainment, and communicationfunctions, which are generally referred as an infotainmentsystem. During vehicle development, testing of the info-tainment system at an overall level is conventionally carriedout manually by an expert who can observe at a customerlevel. This approach has significant limitations with regardto test coverage and effectiveness due to the complexity ofthe system functions and humans capability. Hence, it ishighly demanded by car manufacturers for an automatedinfotainment testing system, which replicates a humanexpert encompassing relevant sensory modalities relatingto control (i.e., touch) and observation (i.e., sight andsound) of the system under test. This paper describes thedesign, development, and evaluation of such a system thatconsists of simulation of vehicle network, vision-basedinspection, automated navigation of features, randomcranking waveform generation, sound detection, and testautomation. The system developed is able to: stimulate avehicle system across a wide variety of initialisationconditions, exercise each function, check for systemresponses, and record failure situations for post-testinganalysis.Keywords Automatic testing.Infotainment.Image processing.Modeling and simulation.Hardware-in-the-loop.Robustness.Validation1 IntroductionAn infotainment system provides a variety of information,entertainment, and communication functions to a vehiclesdriver and passengers. Typical functions are route guidance,audio entertainment such as radio and CD playback, videoentertainment such as TVand interface to mobile phones, aswell as the related interface functions for the users tocontrol the system. There has been a large growth in thisarea driven by rapid developments in consumer electronicsand the customer expectations to have these functions intheir vehicles. Examples of this are surround sound, DVDentertainment systems, iPod connectivity, digital radio andtelevision, and voice activation.With this growth in features there has been acorresponding increase in the technical complexity ofsystems. In a current premium vehicle, the infotainmentsystem is typically implemented as a distributed systemconsistingofanumberofmodulescommunicatingviaahighspeed fiber optic network such as Media Orientated SystemsTransport (MOST). In this implementation the infotainmentsystem is in fact a System of Systems (SOS) with individualsystems having autonomy to achieve their function, butsharing resources such as the HumanMachine Interface(HMI), speakers, and communication channel 1. Typicalissues with such SOS are emergent behavior as systemsinteract in an unanticipated manner particularly duringsome initialisation conditions where it may be possible toget delays and failures in individual systems. These maynot be readily observable until the particular part of theY. Huang (*):R. McMurran:M. Amor-Segan:G. DhadyallaWarwick Manufacturing Group, University of Warwick,Coventry CV4 7AL, UKe-mail: yingping.huangwarwick.ac.ukR. P. JonesSchool of Engineering and IARC, University of Warwick,Coventry, UKP. Bennett:A. Mouzakitis:J. KielochJaguar Land Rover, Engineering Centre,Coventry, UKInt J Adv Manuf Technol (2010) 51:233246DOI 10.1007/s00170-010-2626-2system is exercised. During vehicle development, valida-tion of the infotainment system is extremely important andis conventionally carried out manually by engineers whocan observe at a customer level but this has limitations withregard to test coverage and effectiveness. The firstlimitation is the time available to do manual tests, whichis constrained by the development time scale and engineersworking hours. The second is in the repeatability of the test,which is subject to human error. Hence, there is arequirement for an automated infotainment test capability,which replicates a human expert encompassing relevantsensory modalities relating to control (i.e., touch and voice)and observation (i.e., sight and sound) of the system undertest. This test capability must be able to stimulate thesystem across a wide variety of initialisation conditionsincluding those seen under cranking, low battery or faultconditions, exercise each function, check for systemresponses, and record related data, e.g., MOST bus trace,in the case of a malfunction for subsequent analysis. Thispaper describes the design and development of such asystem as part of a UK academic and industrial collabora-tive project into the validation of complex systems.In the system, a Hardware-in-the-Loop (HIL) platformsupported by a model-based approach simulates the vehiclenetwork in real time and dynamically provides variousessential signals to the infotainment system under test. Sincethe responses of the system are majorly reflected in thedisplay of the touch screen, a machine vision system isemployed to monitor the screen for inspection of thecorrectness of the patterns, text, and warning lights/tell-tales.The majority of infotainment functions are accessed by theuser through an integrated touch screen. In order to achieve afully automated testing, a novel resistance simulationtechnique is designed to simulate the operation of the touchscreen. It is known that voltage transient processes, such asengine start where an instantaneous current inrush can reach800 A, may result in some failures on the system. To test thesystem robustness against low voltage transient conditions, atransient waveform generator is developed to mimic threespecific transient processes. A testing automation softwareintegrates and controls all devices to form a fully automatedtest process, which can be run continuously over days oreven weeks. The developed testing system not only makesvarious testing possible, repeatable, and robust, but alsogreatly improves testing efficiency and eases the task oftedious validation testing.Model-based testing of functionality of an ElectronicControl Unit (ECU) using HIL has been implemented byautomotive manufacturers over the last few years 25.Currently, Jaguar Land Rover (JLR) has adopted the HILtechnology for automated testing and validation of elec-tronic body systems, powertrain, and chassis controlsystems 6, 7. The benefits of this technology includeautomated testing, earlier testing before physical prototypevehicle build, ability to perform robustness and dynamictesting, and reduction of supplier software iterations.Machine vision systems have been used in many manufac-turing applications such as automotive 810, roboticguidance 11, and tracing soldering defects 12, 13. Theauthor also employed machine vision technology forobstacle detection in advanced driver assistant systems14, 15. However, no research has been reported using amachine vision system for design validation testing.Validation testing in the design stage is very much differentfrom testing in manufacturing. Firstly, design validationtesting requires diverse test cases covering a large numberof, rather than a restricted, set to prove proper design. Theonly way to generate the test cases when the car is in theearly development phases is using model-based testingtechniques, which simulate vehicle-operating conditions inreal time. Secondly, design validation testing requiresiterative and repeated tests for robustness evaluation,although it does not require a high volume of parts to betested. Thirdly, design validation testing needs frequentadaptation of the testing system for different types of carsor for different development stages of the same car. Onenovelty of this paper is the integration of the machinevision and HIL techniques for complex design validationtesting. In addition, the paper proposes a novel pseudo-random concept for generating three voltage transientwaveforms, which allows the testing to mimic the randomprocess as seen in real cases, and also enables the testing tobe regenerated for further investigations. Furthermore, acommon approach to mimic the operation of the touchscreen by a human is by using robot arms. In this design, acrafty resistance simulation approach replaces the robotarms to achieve the goal. The approach can be completelyimplemented in software by using the HIL simulator,therefore eliminating the need of complicated mechanicaldevices such as robot arms, pneumatic/hydraulic, andsolenoid actuators.2 System configurationsThe configuration of the system developed for testing theinfotainment system is shown in Fig. 1. The system consistsof six vital elements including the unit under test, HILtester, machine vision (camera), operation of the touchscreen, transient waveform generator, and test automation.The infotainment system under test consists of a numberof modules including the radio/CD player, amplifier(AMP), navigation system, blue tooth/telephone/USB,vehicle setup, auxiliary audio interface, and climate controlfunctions. The HMI is based primarily on a 7 TFT resistivetouch screen with additional hard keys on an Integrated234 Int J Adv Manuf Technol (2010) 51:233246Control Panel (ICP) in the center console and remotecontrols on the steering wheel. Audio output is via a DSPamplifier. Communication between the modules is througha MOST optical bus carrying control, data, and audioinformation. The infotainment system is connected to therest of the vehicle via a module called ICM acting as agateway between MOST and a vehicle Controller AreaNetwork (CAN) bus. It is worth noting that the ICP and theremote controls on the steering wheel reside in the vehicleCAN bus. In addition, a MOST analyzer was connected inthe MOST ring during the testing. The MOST analyzer wascontrolled by the HIL tester via digital outputs to trigger thelogging of the MOST traces when a failure occurs.Within the testing system, the HIL tester simulates thevehicle network and dynamically provides various essentialsignals to the infotainment system under test. It also acts as acontrol center to control other devices. For example, it sendscommands via a serial port to trigger the camera and receivethe inspection results from the camera. The machine visionsystem (camera) checks the responses of the system bymonitoring the display of the touch screen such as patternsand text. The operation of the touch screen is achieved byusing a resistance simulation approach, which is imple-mented in the HIL tester. By using this approach, the testingsystem can get access to the majority of infotainmentfunctions. The transient waveform generator producesvoltage signals and powers up the infotainment system viaa programmable power supplier. The waveform generator,mimicking three voltage transient processes, is used fortesting system robustness against low voltage events. Thetest automation is running in the host computer to integrateand control all devices to form a fully automated testprocess. In addition, the host PC has been linked with themachine vision system via a TCP/IP Ethernet communica-tion. This link allows the storage of time-stamped images inthe host PC so that the behavior of the unit under test can bereviewed offline in terms of the test results. The followingsections describe the individual elements of the automatedtesting system including the HIL tester, vision-basedinspection, automated touch screen operation, transientwaveform generator, and test experiments.3 HIL testerA dSPACE simulator 16 was used to form a hardware-in-the-loop simulation test system. The HIL test systemsimulates the vehicle CAN bus to provide power modesignals to the MOST Network via the MOST gateway. Italso simulates the ICP to operate the infotainment system.Test ScriptPrecondition.test.post conditionPrecondition.test.post conditionPrecondition.test.post conditionPrecondition.test.post conditionPrecondition.test.post condition.Test ResultsPrecondition. te s t.post condition Precondition. te s t.post condition Precondition. te s t.post condition Precondition. te s t.post condition Precondition. te s t.post condition.OpticalControl ParametersCapture DataHostPCCameraTest ResultsVision TestTriggerTest Automation scriptsImages (referenced to test)RS232 SerialEthernetTrigger Low Voltage test profileRS232Transient waveform generatorHIL TesterMOST RingCANAMPOptolyserAudio output monitoringResistive control of touch screenPowerSupplyClimate/Setup/InterfaceICM gatewayRadio/CDplayerBlue tooth/Phone/USBMOST analyzerMOST AnalyzertriggerTouchScreenNavigationICP & Remote Cont Fig. 1 System configurationInt J Adv Manuf Technol (2010) 51:233246 235In addition, the HIL tester also provides RS 232 serialinterfaces to communicate with the camera and transientwaveform generator, resistance simulation to operate thetouch screen, and an A/D interface for detecting sound andmeasuring sound frequency.ThedSPACESimulatorconsistsofsimulationmodelsandexpansion hardware as shown in Fig. 2. The expansion boxincludes one processor board DS1006 and one interfaceboard DS2211. The DSP board runs the simulation models,while the interface board provides various interface linkswith other devices, such as CAN, resistance outputs, A/Dconverters, analog/digital input and output, and RS 232serial communication to control the machine vision system.In the HIL system, simulation models are implementedin MATLAB/Simulink/Stateflows and compiled using theauto-C-code generation functions of Matlabs Real-TimeWorkshop for real-time execution.3.1 Simulation of power modeThe behavior of the components of the Infotainment systemis determined by a CAN signal known as Power mode,which indicates the operational state of the vehicle e.g.,ignition off,ignition on,engine cranking,enginerunning, etc. To test the performance of the infotainmentsystem under cranking conditions, the car under test must bein the engine-cranking state when applying crankingtransient voltages to the car. Moreover, any subsequentfunctional tests must be conducted in the engine-runningstate after the cranking. In a real car, power mode messagesare transmitted by the body ECU connected to the CAN.Since we were testing the infotainment system on a testplatform representing a real car sometimes, in order togenerate the correct power mode behavior, we utilized CANsimulation of the HIL tester to simulate the body ECU totransmit power mode messages to the MOST gateway.3.2 ICP simulationThe Integrated Control Panel of the infotainment systemprovides users with a number of hard keys for operating thesystem. The functions controlled by the ICP includeselection of the audio sources, loading and ejecting CDs,seeking up/down for radio stations and CD tracks, volumecontrols, and so on. To enable an automated testing of thesefunctions, the ICP must be controlled by the test center, thedSPACE real-time simulator.The ICP electronic control unit interfaces with a vehiclevia the vehicle CAN. Therefore, the ICP unit was simulatedby using the CAN simulation of the dSPACE simulator.The models of ICP simulation are shown in Fig. 3.3.3 Sound detectionSound detection contains two aspects i.e., detecting soundon or off and detecting the frequency (dominant) of thesound. The sound signal is sampled from the speaker end asshown in Fig. 1, and converted into digital signal by an A/Dconverter within the dSPACE simulator. The sound on/offis determined by checking the amplitude of the signal. Thefrequency of the sound is detected by the specific circuit ofthe simulator. The purpose of detecting sound frequency isto identify a sound source and active CD track. The modelis shown in Fig. 4.Power supply controlSimulation of touch Screen operationand On/Off Switches -resistance outputsSerial communications RS232Digital signal processorStandard I/OInterfaceSimulation ModelsExpansion boxCAN I/O RS232Real-time simulatorSimulation of Power Mode and integrated control panel - CANSound detection and measuring sound frequency A/D inputsFig. 2 dSPACE real-time simulator236 Int J Adv Manuf Technol (2010) 51:2332463.4 Simulation of serial communicationsThe RS232 serial communication is used to establishthe link between the HIL tester with the camera and thetransient waveform generator so that closed loop testingcan be performed. During the test, the HIL tester is thecontrol center to command the camera and the transientwaveform generator and to obtain the inspection resultsfrom them. For example, the camera needs to becommanded to select a specific image processing jobfile for specific testing. The checking results generatedbythecameraneedtobereturnedtotheHILtester.The transient waveform generator needs to be com-manded to generate a specific cranking waveform forspecific testing. The parameters of the waveformresulting in a failure need to be returned to the HILtester so that this specific testing can be duplicated inthe later analysis stages.A simplified version of the simulation models of theRS232 serial communication is shown in Fig. 5.Atransmitted message is ended with a carriage return andhas a maximum length of 10 bytes. A received message hasa fixed length of 8 bytes. The first 3 bytes gives the resultname while the following 5 bytes indicates the resultvalues. For example, the active track number is abbreviatedas the result name ATN.Fig. 3 Model of ICP simulationInt J Adv Manuf Technol (2010) 51:233246 2374 Vision-based inspection4.1 Machine vision systemThe machine vision system consists of a camera, lighting,optics, and image processing software. A Cognex In-sightcolor vision sensor 17 was selected for image acquisitionand processing, which offers a resolution of 640480 pixelsand a 32-MB flash m