关 键 词：

大学生 方程式赛车 设计 整体 车架 标准 安全 系统 座椅 附件 cad 三维 中英文 翻译

资源描述：

大学生方程式赛车设计（整体车架、标准安全系统、座椅及附件设计）（有cad图+三维图+中英文翻译）,大学生,方程式赛车,设计,整体,车架,标准,安全,系统,座椅,附件,cad,三维,中英文,翻译

内容简介：

Design and evaluation of a unified chassis control system for rolloverprevention and vehicle stability improvement on a virtual test trackJangyeol Yoona, Wanki Choa, Juyong Kanga, Bongyeong Koob, Kyongsu Yia,naSchool of Mechanical and Aerospace Engineering, Seoul National University, 599 Gwanangno, Gwanak-Gu, Seoul 151-742, Republic of KoreabMando Corporation Central R&D Center, 413-5 Gomae-Ri, Giheung-Eub, Yongin-Si, Kyonggi-Do 449-901, Republic of Koreaa r t i c l e i n f oArticle history:Received 6 May 2009Accepted 23 February 2010Available online 23 March 2010Keywords:Full-scale driving simulatorHuman-in-the-loop evaluationRollover mitigation controlUnified chassis controlVehicle stabilityVirtual test tracka b s t r a c tThis paper describes the development of a unified chassis control (UCC) scheme and the evaluation ofthe control scheme on a virtual test track (VTT). The UCC scheme aims to prevent vehicle rollover, andto improve vehicle maneuverability and its lateral stability by integrating electronic stability control(ESC) and active front steering (AFS). The rollover prevention is achieved through speed control, and thevehicle stability is improved via yaw rate control. Since the UCC controller always works with thedriver, the overall vehicle performance depends not only on how well the controller works but also onits interactions with the human driver. Vehicle behavior and the interactions between the vehicle, thecontroller, and the human driver are investigated through a full-scale driving simulator on the VTTwhich consists of a real-time vehicle simulator, a visual animation engine, a visual display, and suitablehumanvehicle interfaces. The VTT has been developed and used for the evaluation of the UCC undervarious realistic conditions in the laboratory making it possible to evaluate the UCC controller in thelaboratory without risk of injury prior to field testing, and promises to significantly reduce the cost ofdevelopment as well as the overall cycle development time.& 2010 Elsevier Ltd. All rights reserved.1. IntroductionVehicle rollover is a serious problem in the area of groundtransportation and a report published by the National HighwayTraffic Safety Administration (NHTSA) has found that, eventhough rollover constitutes only a small percentage of allaccidents, it does, however constitute a disproportionately largeportion of severe and fatal injuries. Almost 11 million passengercars, SUVs, pickups, and vans crashed in 2002, yet only 2.6% ofthese involved a rollover. However, the percentage of fatal crashesthat involved the occurrence of rollover was about 21.1%, whichis significantly higher than the corresponding percentages forother types of crashes (NHTSA, 2003). In order to help consumersunderstandavehicleslikelihoodofrollover,therolloverresistance rating program was proposed by NHTSA which usesthe static stability factor (SSF), which is the ratio of half the trackwidth to the height of the center of gravity (CG), to determine therollover resistance rating. The SSF has been questioned by theautomotive industry as it does not consider the effects ofsuspension deflection, tire traction aspects, or the dynamics ofthevehiclecontrolsystem.Accordingly,in2002,NHTSApublished another announcement with regard to a tentativedynamical rollover test procedure (NHTSA, 2001).Most existing rollover prevention technologies can be classi-fied into two types, namely, (1) the type which directly controlsthe vehicle roll motion through an active suspension, an activeanti-roll bar, or an active stabilizer (Chen & Hsu, 2008) which canprevent rollover by raising the rollover threshold; and (2) the typewhich indirectly influences roll motions by controlling the yawmotions through differential braking and active front steering(Wielenga & Chace, 2000). Several studies have been undertakenon rollover detection and its prevention and Hac et al. haveproposed an algorithm that detects impending rollover and anestimator-based roll index (Hac, Brown, & Martens, 2004). Chenand Peng proposed an anti-rollover algorithm based on the time-to-rollover (TTR) metric (Chen & Peng, 2001). In this research,differential braking is selected as the actuation methodology.Ungoren and Peng evaluated a vehicle dynamics control (VDC)system for rollover prevention (Ungoren & Peng, 2004). Yang andLiu proposed a robust active suspension for rollover prevention(Yang & Liu, 2003) and Schofield and Hagglund proposed amethod for rollover prevention that employs an optimal tire forcedistribution (Schofield & Hagglund, 2008). Yoon and Yi proposed arollover index that indicates the danger of vehicle rollover as wellas an index-based rollover mitigation control system to reducethe rollover index through Electronic Stability Control (ESC)(Yoon, Kim, & Yi, 2007). Since the lateral acceleration is theARTICLE IN PRESSContents lists available at ScienceDirectjournal homepage: /locate/conengpracControl Engineering Practice0967-0661/$-see front matter & 2010 Elsevier Ltd. All rights reserved.doi:10.1016/j.conengprac.2010.02.012nCorresponding author. Tel.: +82 2 880 1941; fax: +82 2 882 0561.E-mail address: kyisnu.ac.kr (K. Yi).Control Engineering Practice 18 (2010) 585597ARTICLE IN PRESSdominant factor in vehicle rollover, much research into rolloverprevention has proposed the use yaw motion control to reducethe lateral acceleration. However, since these rollover preventionschemes only focus on reducing the lateral acceleration, vehiclemaneuverability and lateral stability cannot be guaranteed (Yoon,Cho, Koo, & Yi, 2009). For instance, when the rollover preventioncontroller works to reduce the lateral acceleration, this tends to bein the opposite direction to the intentions of the driver which maycause the vehicle to deviate from the road, thereby resulting in anaccident. Studies have been conducted to prevent rollover whilemaintaining good lateral stability. Jo et al. proposed a VDC systemfor rollover prevention and ensuring lateral stability (Jo, You,Jeong, Lee, & Yi, 2008). In this research, a VDC is designed andactivated in descending order of priority rollover prevention,excessive side-slip angle, and under-steering/over-steering of thevehicle.However,thismethodleadstoreductionofthemaneuverability or rollover prevention.For this reason, the unified chassis control (UCC) algorithm hasbeen designed to prevent vehicle rollover while, at the same time,ensuring good maneuverability and lateral stability by integratingindividual chassis control modules, such as ESC and active frontsteering (AFS). A vehicle speed control algorithm has beendesigned to prevent rollover and an algorithm for controllingthe yaw motion has been designed to improve the maneuver-ability and the lateral stability. The proposed UCC works toenhance the maneuverability and the lateral stability in normaldriving situations without danger of rollover. When the risk ofrollover increases, the proposed UCC works to prevent vehiclerollover and at the same time ensures the vehicle can continu-ously move in the path intended by the driver. In order to detectan impending vehicle rollover, the rollover index (RI), as proposedin a prior study (Yoon et al., 2007), is employed.Since the UCC controller always works with the driver, theoverall vehicle performance will depend not only on how well thecontroller works but also on its interactions with the humandriver. Therefore, a closed human-in-the-loop evaluation wouldbe a more effective way of designing the UCC controller thanperforming open-loop simulations that use the prescribed steer-ing and velocity profiles (Chung & Yi, 2006). Moreover, theevaluation of active safety systems, such as UCC, active cruisecontrol, collision warning, collision avoidance, etc., rely heavily onfield testing that entails time-consuming and expensive trials, andoften significant danger (Han & Yi, 2006a). A model-basedsimulation makes it possible to perform exhaustive design trialsand evaluations prior to field testing. For this reason, a full-scaledriving simulator on a virtual test track (VTT) has been developedand used in a human-in-the-loop evaluation of the UCC where theVTT, based on the concept of rapid control prototyping (RCP), hasbeen described in Lee (2004).In this paper, the control performance of the proposed UCCalgorithm has been investigated by a real-time human-in-the-loop simulation, using a vehicle simulator on a VTT. The tests,based on the VTT, are conducted by thirteen drivers and theresults have been analyzed in detail and summarized here.2. Unified chassis controller designIn this study, the UCC system is designed to prevent a vehiclerollover and to improve both the maneuverability and the lateralstability of the vehicle by integrating the individual chassiscontrol modules such as the ESC and AFS. There are three controlmodes, namely, ROM, ESC-c, and ESC-b, which stand for rolloverprevention, maneuverability and lateral stability, respectively.The proposed UCC works to enhance the maneuverability and thelateral stability in normal situations without danger of rollover.The improvement in maneuverability and lateral stability isachieved by reducing the yaw rate error between the actualyaw rate and the desired yaw rate, based on the drivers steeringinput and the vehicles side slip angle. When the risk of rollover ishigh, the proposed UCC works to reduce vehicle rollover and, atthe same time, improves the maneuverability and the lateralstability. As mentioned in the previous section, since priorresearch concerning rollover mitigation (ROM) control, i.e., anRI-based ROM control (Yoon et al., 2007), is only focused on theprevention of vehicle rollover, then vehicle maneuverability andlateral stability cannot be guaranteed. For instance, since vehiclerollover generally occurs at large lateral accelerations, prior RI-based ROM controllers operate to reduce the lateral acceleration.This control strategy tends to control the vehicle in the oppositedirection intended by the driver which may cause the vehicle todeviate from the road resulting in accidents. For this reason, an RI/vehicle stability (VS)-based UCC controller is designed to preventNomenclatureadistance from the center of gravity (CG) to the frontaxleaylateral acceleration of the vehicleay,desdesired lateral accelerationay,ccritical lateral accelerationay,msensor measurement of the lateral accelerationbdistance from CG to the rear axlemvehicle massttread (track width)vxlongitudinal velocity of the vehiclevx,desdesired longitudinal velocity of the vehiclevylateral velocity of the vehicleCfcornering stiffness of the front tireCrcornering stiffness of the rear tireFxlongitudinal tire forceFx,1longitudinal tire force of the front-left wheelFxflongitudinal tire force of the front sideFxylongitudinal tire force of the rear sideFyflateral tire force of the front sideFyrlateral tire force of the rear sideFy,1lateral tire force of the front-left wheelFzfvertical tire force of the front sideFzrvertical tire force of the rear sideFz,1vertical tire force of the front-left wheelFz,2vertical tire force of the front-right wheelFz,3vertical tire force of the rear-left wheelFz,4vertical tire force of the rear-right wheelIzmoment of inertia about the yaw axisMzdirect yaw momentbside slip angle of the vehicledftire steer anglefvehicle roll anglefthroll angle threshold_fvehicle roll rate_fthroll rate thresholdgyaw rategddesired yaw rateJ. Yoon et al. / Control Engineering Practice 18 (2010) 585597586ARTICLE IN PRESSvehicle rollover and at the same time ensuring that the vehiclecan continuously move in the intended path of the driver.Fig. 1 shows a schematic diagram of the RI/VS-based UCCstrategy where the proposed UCC system consists of upper andlower-levelcontrollerswheretheupper-levelcontrollerdetermines thecontrolmode,such asrollover prevention,maneuverability level, and lateral stability; it also calculates thedesired braking force and the desired yaw moment for itsobjectives. Each control mode generates a control yaw momentand a longitudinal tire force in line with its coherent objective.The lower-level controller calculates the longitudinal and lateraltire forces as inputs of the control modules, such as the ESC andthe AFS.2.1. The upper-level controller: decision, desired braking force, anddesired yaw momentThe upper-level controller consists of three control modes anda switching logic. A control yaw moment and the longitudinal tireforce are determined in line with its coherent control mode sothat the switching across control modes is performed on the basisof the threshold. Based on the drivers input and sensor signals,the upper-level controller determines which control mode is to beselected, as shown in Fig. 2.In this study, RI is used to detect an impending vehicle rolloverwhere the RI is a dimensionless number that can indicate the riskof vehicle rollover and it is calculated through: the measuredlateral acceleration, ay, the estimated roll angle,f, the estimatedroll rate,_f, and their critical values which depend on the vehiclegeometry in the following manner (Yoon et al., 2007):In (1), C1, C2, and k1are positive constants (0oC1o1,0oC2o1), C1and C2are weighting factors, which are related tothe roll states and the lateral acceleration of the vehicle, and k1isa design parameter which is determined by the roll angle-ratephase plane analysis. These parameters in (1) are determinedthrough a simulation study undertaken under various drivingsituations and tuned such that an RI of 1 indicates wheel-lift-off. Adetailed description for the determination of the RI is provided inprevious research (Yoon et al., 2007). The lateral acceleration caneasily be measured from sensors that already exist on a vehicleequipped with an ESC system. However, additional sensors areneeded to measure the roll angle and the roll rate, although it isdifficult and costly to directly measure these (Schubert, Nichols,Fig. 1. RI/VS-based UCC strategy.Fig. 2. Control modes for the proposed UCC system.RI C1ft? _fth_ft?fthfth_fth01AC2ay?ay,c?1?C1?C2ft?ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffift2_ft?2r0BB1CCA,f_f?k1f?40RI 0,f_f?k1f?r08:1J. Yoon et al. / Control Engineering Practice 18 (2010) 585597587ARTICLE IN PRESSWallner, Kong, & Schiffmann, 2004). For this reason, the roll angleand the roll rate are estimated by a model-based roll stateestimator (Park, Yoon, Yi, & Kim, 2008).The proposed RI is evaluated using vehicle test data obtainedfrom the MANDO Corporation. Note that the test data used in thisevaluation are not the outcome from the proposed UCC system. Inother words, the control algorithm of MANDO is different fromthe one described in this paper so that the test results show littledifference compared with the desired results. Fig. 4 shows thevehicle test data and the rollover index for the fishhook test whichhas been developed by NHTSA, as a dynamical test for theprediction of dynamic rollover propensity and the test results areused for vehicle evaluation. The fishhook test maneuver isdescribed in Fig. 3.Fig. 4(a) shows the time histories of the steering angle of twotest cases where the entrance speeds are 43.2 and 45.6 mph,respectively, but the vehicle stability control input is applied onlyfor the 45.6 mph case. In both cases, either one or two wheels arelifted off at about 4.2 s, and the rollover indices increase overunity. However, once the control input is selected, the roll angleand the lateral acceleration are decreased, and the rollover indexalso decreases below unity, as shown in Fig. 4(b)(d). In contrastwith the control case, the roll angle, the lateral acceleration, andthe rollover index increase over unity in the non-control case.Consequently, the vehicle is rolled over at about 6 s.If the RI exceeds a particular threshold, then the rolloverprevention mode, ROM, is activated, otherwise, the controller is ineither the maneuverability mode or in the lateral stabilitymode. Under a small side slip angle, the controller is in themaneuverability mode, ESC-c, if the error between the actual yawrate and the desired yaw rate exceeds a particular threshold.The condition of activation of the lateral stability mode isdetermined by the vehicle side slip angle. If the side slip angleexceeds the threshold value, the controller is in the lateralstability mode, ESC-band the side slip angle can be successfullyestimated in real time from already existing vehicle sensors(You, Hahn, & Lee, 2009).The maneuverability and the lateral stability are ensured bythe yaw moment control method and rollover prevention isachieved by the yaw moment/speed control. The upper-levelcontroller calculates the desired braking force,DFx, for rolloverprevention and the desired yaw moment, Mz, for maneuverabilityand lateral stability. The state-transition diagram for the requiredcontrol mode switching in the upper-level controller is given inFig. 5.The signals used for the state transitions are the yaw rate error,ge, the side slip angle,b, and the RI so that each event in Fig. 5represent a switching condition, and the conditions of itsactivation are described in Table 1. When the vehicle state iseither ESC-cor ESC-b, as shown in Fig. 5, the yaw moment controlis applied and generates the desired yaw moment to track a targetyaw rate. In ESC-c, a target yaw rate is generated on the basis ofthe drivers steering input for maneuverability and in ESC-b, atarget yaw rate is generated to reduce some excessive side slipangle,b, for achieving lateral stability. When the vehicle state isROM, the yaw moment and speed control are applied to generate Steering wheel angle 012345678Time secNo control 43.2mphControl 45.6mphRoll angle 012345678Time secNo control 43.2mphContro l45.6mph Lateral acceleration 012345678Time secNo control 43.2mphControl 45.6mphRollover index 012345678-15-10-5051015-15-10-505101500.511.5-200-1000100200Time secRoll angle deg/secay m/sRollover indexSWA degNo control 43.2mphControl 45.6mphFig. 4. Rollover index validation through vehicle test data (NHTSA fishhook test).Fig. 3. Fishhook maneuver developed by NHTSA (adopted from Corrsys-Datron).J. Yoon et al. / Control Engineering Practice 18 (2010) 585597588ARTICLE IN PRESSthe desired yaw moment for vehicle stability and the brakingforce for rollover prevention, respectively.In Table 1, The RI_threholdis set to 0.7, which is the critical valueat which all the wheels of the vehicle contact with the ground andtheb_thresholdis selected as 0.06 rad under the assumption ofm0.3 from the literature (Rajamani, 2006). The threshold for theyaw rate errorge,this set to 0.08 rad/s to give the largest yaw rateerror when the vehicle is performing a single lane change at60 km/h on dry asphalt.2.1.1. Desired yaw moment for maneuverability and lateral stability(ESC-g/ESC-bmodes)If the RI is small, the ESC-cor the ESC-bmode is activated forachieving the desired maneuverability or lateral stability, respec-tively.Inthiscontrolmode,thedesiredyawmomentisdetermined whose purpose is to reduce the yaw rate error byusing a bicycle model for computing the target vehicle response.This linear model can represent the vehicle dynamics in theregion of linear tire characteristics, and has been validated inmany publications in the literature (see for example, Nagai, Shino,& Gao, 2002). In addition, since the vehicle active safety controlshould be intervened before the vehicle enters any dangeroussituations in which the tires are near the limits of adhesion, thecharacteristic of the tire is beyond the linear region at that timewhen the control intervention is needed. Hence, the linear bicyclemodel is sufficient to design a controller to ensure vehiclestability.Adirectyawmomentcontrolmethodisemployedtodetermine the desired yaw moment and Fig. 6 shows the 2-Dbicycle model, including the direct yaw moment, Mz.The dynamic equations of the 2-D bicycle model are repre-sented as follows:_b_g#?2CfCrmvx2?aCfbCrmv2x?12?aCfbCrIz?2a2Cfb2CrIzvx2666437775bg#2Cfmvx2aCfIz2666437775Df01Iz2435Mz2In general, through Eq. (2), the desired yaw rate, based on thedrivers steering input, is theoretically determined in light of the2-D bicycle model with a linear tire force. The steady-state yawrate of the bicycle model is introduced and the maneuver of thevehicle is considered to reflect the drivers intentions and this isexpressed as a function of the vehicles longitudinal velocity andthe drivers steering input, as follows:gdes_yaw11?maCf?bCrv2x=2CfCrab2vxabDf3The desired yaw rate, which is represented in (3), is used as thereference yaw rate for the ESC-ccontrol mode.In general, the lateral stability cannot be guaranteed if the sideslip angle exceeds about 31 and excessive body side slip of avehicle causes its yaw motion to be insensitive to the driverssteering input and threatens the lateral stability. As the side slipangle of a vehicle increases, the stabilizing yaw moment due tothe steering input decreases, and thus, the lateral behavior of thevehicle becomes unstable. Therefore, a control intervention tomaintain the body side slip angle to lie within a reasonably smallrange, i.e., 31, is required to improve the lateral stability of thevehicle (Jo et al., 2008).Through a 2-D bicycle model, the lateral vehicle dynamics areexpressed as follows:m_vy ?mvxg2FyfcosDf2Fyr4From (4), assuming that_vx? 0, the side slip angle dynamicscan be expressed as follows:_b ?g2FyfcosDf2Fyrmvx5Let the desired yaw rate be defined asgdes_lateral K1b2FyfcosDf2Fyrmvx6Then, the dynamics of the body side slip angle are stable, asshown in (7), which implies that the body sideslip angleasymptotically converges to zero:_b ?K1b7In (7), K1is a design parameter, which is strictly positive.The desired yaw rate, which is represented in (6), is usedas the reference yaw rate for the ESC-bcontrol mode and thereference yaw rate,gdes, for determining the desired yaw momentis selected through either (3) or (6), depending on the controlmode.Fig. 6. A 2-D bicycle model including the direct yaw moment.Table 1Events and corresponding conditions of activation.EventsActivation conditione19ge9Zge_thresholde29ge9oge_thresholde3RIoRIthresholde4RIZRIthresholde5RIoRIthreshold, 9b9Zbthresholde6RIoRIthreshold, 9b9obthreshold, 9ge9Zge_thresholde7RIZRIthreshold, 9b9Zbthresholde8RIoRIthreshold, 9b9Zbthresholde9RIoRIthreshold, 9b9obthreshold, 9ge9oge_thresholdFig. 5. State-transition diagram for the control mode switching.J. Yoon et al. / Control Engineering Practice 18 (2010) 585597589ARTICLE IN PRESSThe desired yaw moment can be obtained through (2) and thereference yaw rate, which is one of (3) and (6). From (2), thedynamic equation concerning the yaw rate, including the directyaw moment, is presented as follows:_g2?aCfbCrIzb?2a2Cfb2CrIzvxg2aCfIzDf1IzMz8The sliding mode control method has also been used todetermine the desired yaw moment; in this the sliding surfaceand the sliding condition are defined as follows:s1g?gdes,12ddts12 s1_s1r?Z1s1jj9whereZ1is a positive constant. The equivalent control input thatwould achieve_s1 0 is calculated as follows:Mz,eq ?Iz2?aCfbCrIzb?2a2Cfb2CrIzvxg2aCfIzDf !10Finally, the desired yaw moment for satisfying the slidingcondition regardless of the model uncertainty is determined asfollows:Mz Mz,eq?K2satg?gdesF1?11whereF1is a control boundary, and the gain, K2, which satisfiesthe sliding condition, is calculated as follows:K2 IzFyfIz?ab?a2gaDf?FyrIzbb?b2g?_gdes?Z2?122.1.2. Desired braking force for rollover prevention (the ROM mode)IftheRIincreasestoapredefinedRIthresholdvalue,which can predict an impending rollover, the ROM controlinput should be applied to the vehicle in order to preventrollover. Rollover prevention control can be achieved throughvehicle speed control and the desired braking force is determinedin this section to control the speed. In addition, the desired yawmoment, as determined in the previous section, is also applied tothe vehicle to improve the maneuverability and the lateralstability.As mentioned previously, since vehicle rollovers occur at largelateral accelerations, the desired lateral acceleration should bedefined and can be determined from the RI (cf. Eq. (1) as follows:ay,des1C2RItar?C1ft? _fth_ft?fthfth_fth01A?1?C1?C2ft?ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffift2_ft?2r0BB1CCA8:9=;ay,c13In (2), the target RI value, RItar, is set to 0.6.The desired vehicle speed for obtaining the desired lateralacceleration is calculated from the lateral vehicle dynamics asfollows (Yoon et al., 2009):vx,des1gay,des? ay,m?vxg?14The desired braking force to yield the desired vehicle speed iscalculated through a planar model, as shown in Fig. 7, andthrough the sliding mode control law.Fig. 7 shows a planar vehicle model including the desiredbraking force,DFxand the dynamic equation for the x-axis isdescribed as follows:m_vx FxrFxfcosDf?FyfsinDfmvyg?DFx15By the assumption of having small steering angles, Eq. (15) canbe rewritten in terms of the derivative of the vehicle speed asfollows:_vx1mFxrFxf?FyfDfvyg?1mDFx16In order to obtain the desired braking force, the sliding modecontrol method is used. The sliding surface and the slidingcondition are defined as follows:s2 vx?vx,des,12ddts22 s2_s2r?Z2s2jj17whereZ2is a positive constant.Finally, the desired braking force for preventing a rollover isobtained as follows:DFxDFx,eq?K3satvx?vx,desF2?,K3 ?Z2m18whereDFx,eq FxfFxr?FyfDfmvyg?_vx,des. In (18),F2is acontrol boundary to eliminate high signal chattering due to highfrequency components in the control input. Further informationabout the desired braking force can be found in previous research(Yoon et al., 2009).2.2. The lower-level controllerThe lower-level controller distributes the desired braking forceand the yaw moment to the longitudinal and lateral tire forces asinputs of the ESC and AFS modules. In this paper, two schemes areused to distribute the desired braking force and the yaw moment.One is an optimized distribution scheme without any risk ofcausing rollover, and the other is a simple distribution schemethat has risk of rollover. The former is used in the ESC-cand ESC-bmodes, while the latter is used in the ROM mode. The optimizeddistribution scheme determines the differential braking input andactive front steering input for the ESC and AFS modules,respectively. This optimization problem focuses on minimizingthe use of braking because the ESC module has some negativeFig. 7. Planar model including the desired braking force.J. Yoon et al. / Control Engineering Practice 18 (2010) 585597590ARTICLE IN PRESSeffects as the simple distribution scheme determines only thedifferential braking input for the ESC module. These two schemesare switched in accordance with the protocol for switching acrosscontrol modes in the upper-level controller and the only ESCmodule is used in the ROM mode since the optimized distributionscheme for the AFS and ESC modules provides a very smallbraking to each wheel, which cannot decrease the vehicle speedwhich is essential for preventing rollover.Moreover, the slip angle of the tire is proportionally increasedwith the lateral acceleration as shown in Fig. 8. Since vehiclerollovers generally occurs at large lateral acceleration, the slipangle of the tire is also very large in the ROM mode situation.The AFS module cannot generate the lateral tire force in largeslip angle situations as shown in Fig. 9; therefore the AFS moduleis not used in the ROM mode, that is, the ESC is the most effectivefor the ROM mode. For this reason, only the ESC control module isused for the ROM mode.2.2.1. Tire-force distribution in vehicle stability situations (ESC-g/ESC-bmode)In vehicle stability situations that do not have risk of rollover,the control interventions for maneuverability, ESC-c, and forlateral stability, ESC-b, are activated. When the lateral accelera-tion is small enough so that the slip angle is small, thecharacteristics of the lateral tire force lie within the linear region,as shown in Fig. 9. In these situations, only the AFS controlmodule is applied and the AFS control input is determinedthrough the consideration of the 2-D bicycle model as follows:DdfMz2aCf19When the lateral acceleration increases greatly, the combinedcontrol inputs that are based on the ESC and AFS modules areapplied. Since the ESC module has some negative effects, such asthe degradation of ride comfort and the wear of tires and brakes,the optimized coordination of tire forces is focused on minimizingthe use of braking. An optimal coordination of the lateral andlongitudinal tire forces for the desired yaw moment is determinedthrough the KarushKuhnTucker (KKT) conditions (Cho, Yoon, &Yi, 2007). Fig. 10 shows the coordinate system corresponding to theresultant force when the desired yaw moment is positive. The signof the desired yaw moment determines what tire forces should beused for optimal coordination. If the desired yaw moment ispositive, four variables,DFx1,DFy1,DFy2, andDFx3, should becoordinated to generate the yaw moment as represented in Fig. 10.These optimal variables can be reduced by using somerelations which correspond to the vertical load of the vehicle.Since the active steering angles for both front tires are the same,the active lateral tire forces have a relation as follows:DFy2Fz2Fz1Fy120Moreover, the longitudinal tire forces at the front and rearhave a relation as follows:DFx3Fz3Fz1DFx121Slip angle degESCAFS+ESC AFSLateral tire force N036Fig. 9. Characteristics of the lateral tire force.+Fig. 10. Coordinate system corresponding to the resultant force (Mz40).00.70.8-8-6-4-20Lateral acceleration gSlip angle degVehicle StabilityRolloverPreventionFig. 8. Relation between the lateral acceleration and the slip angle.J. Yoon et al. / Control Engineering Practice 18 (2010) 585597591ARTICLE IN PRESSUsing (20) and (21), two variables,DFy2andDFx3, can beeliminated in the optimization problem so that the optimaldistribution problem for the longitudinal and lateral tire forcesinvolves only two variables, namely,DFx1andDFy1.Thecostfunctionoftheproposedoptimizationisthemagnitude of the additional longitudinal tire force by braking asfollows:LDFx DF2x122This optimization problem has the two variables,DFx1andDFy1, along with equality and inequality constraints; two of theseconstraints are determined as follows:fx ?t2D1DFx1aD2DFy1?MZ 023gx DFx1Fx12DFy1Fy12?m2Fz12r024In the above, D1 1Fz3=Fz1, D2 1Fz2=Fz1.The equality constraint in (23) means that the sum of theyaw moment generated by the longitudinal and the lateraltire forces should be equal to the desired yaw moment. Theinequality constraint in (24) means that the sum of the long-itudinal and the lateral tire forces should be less than the frictionforces on the tire.From (22)(24), the Hamiltonian is defined as follows:H DFx12l?t2D1DFx1aD2DFy1?MZ?rDFx1Fx12DFy1Fy12?m2UFz12c2?25wherelis the Lagrange multiplier, c the slack variable, andrthesemi-positive number.First-ordernecessaryconditionsabouttheHamiltonianare determined by the KarushKuhnTucker condition theoryas follows:HDFx1 2DFx1?t2D1l2rDFx1Fx1 026HDFy1 aD2l2rDFy1Fy1 027Hl ?t2D1DFx1aD2DFy1?DMZ 028rgx rDFx1Fx12DFy1Fy12?m2Fz12? 029From (29), two cases are derived with respect torand g(x) asfollows:Case 1.r 0, gxo0.Case 2.r40, g(x)0.Case 1 means that the sum of longitudinal and lateral tire forcesis smaller than the friction of the tire. On the other hand, Case 2means that the sum of the longitudinal and lateral tire forces isequal to the friction of the tire. The solutions of the optimizationproblem represented in (3.41) can be obtained for both cases.If the desire yaw moment is positive, Mz40, the solutions areobtained as follows:Case 1 :DFx1 0DFy1MZaD20B30Case 2 :DFx1?Fx1kzffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1k2m2Fz12?kFx1?z2q1k2 DFy1tD12aD2DFx11aD2MZ31wherek(tD1/2aD2) andz(1/aD2)MZ+Fy1.The brake pressure for the ESC module and the additionalsteering angle for the AFS module are determined from (32)FxF,maxxR,maxFFxFFxRFzFFzRFig. 11. Friction circles of the front and rear tires.Fig. 12. Hardware configuration of the driving simulator with a human in-the-loop.J. Yoon et al. / Control Engineering Practice 18 (2010) 585597592ARTICLE IN PRESSas follows:DDfDFyiCfPBirwfDFxiKBii 1,20BBB32In (32), KBiis the brake gain, and rwfthe radius of the wheel.When the desired yaw moment is negative, Mzo0, the tire forcescan be obtained in a manner similar to (30) and (31).2.2.2. Tire-force distribution in rollover situations (ROM mode)In the previous sections, the desired braking force, whichshould be subjected to the vehicle for rollover prevention, and thedesired yaw moment for reducing the error in the yaw rate havebeen determined. By utilizing the above two values, a braking-force distribution is accomplished simply to help prevent vehiclerollover, while ensuring that the vehicle follows the intended pathof the driver. The forces of the vehicle can be determinedkinematically, as follows:DFx,left12DFxMztDFx,right12DFx?Mzt8:33The braking forces of the left and right sides are obtained bysubstituting (18) and (11) into (33). Fig. 11 shows the frictioncircles of the front and rear tires and the traction force,determined through the shaft torque, is applied at the front tire,and the drag force is applied at the rear tire.The maximum braking forces of the front and rear tires can bedetermined as follows:DFxf,max Fxf?ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffimFzf2?Fyf2q34DFxr,max ?Fxr?ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffimFzr2?Fyr2q35The braking-force distributions of the front and rear tires areachieved by using equations from (33) through to (35) as follows:DFxr,leftDFxr,left,max?DFxf,left,max?DFxf,left36DFxr,rightDFxr,right,max?DFxf,right,max?DFxf,right37Intheabove,DFxf,leftDFxr,leftDFx,leftandDFxf,rightDFxr,rightDFx,right.024681012141618Time sec Yaw rate 024681012141618Time sec Lateral acceleration 024681012141618Time secVehicle testSimulator Roll angle 024681012141618-20-1001020-6-4-20246-4-2024-50050Time secYaw rate deg/sLateral acceleration m/s2Roll angle degSteering wheel angle degVehicle testSimulatorVehicle testSimulatorVehicle testSimulator Steering wheel angle Fig. 13. Comparison between actual vehicle test data and the driving simulator(for the slalom test).80km/hObstacleFig. 14. The test scenario: obstacle avoidance.J. Yoon et al. / Control Engineering Practice 18 (2010) 585597593ARTICLE IN PRESSThe braking pressure of the front-left wheel can be determined asfollows:PBf,leftrwfDFxf,leftKBfifDFxf,leftoDFxf,maxrwfDFxf,maxKBfifDFxf,leftZDFxf,max8:38The other tire forces can be obtained in a manner similar to(38).3. Full-scale driving simulatorThe configuration of the full-scale driving simulator for thehuman-in-the-loop system is shown in Fig. 12, consisting of fourparts: a real-time (RT) simulation hardware, a visual graphicalengine, a human-vehicle interface, and a motion platform. The hostcomputer in Fig. 12 is utilized to modify the vehicle simulationprogram and to display the current vehicle status. The RT simulationhardware calculates the variables of the vehicle model representedusing a CARSIM model controlled by the UCC controller withmeasured driver reactions. By the use of the vehicle-behaviorinformation obtained using RT simulation hardware, the visualgraphical engine projects a visual representation of the drivingconditions to the human driver via a beam projector with a 100-inscreen who interacts with the 3-D virtual simulation and thekinesthetic cues of the simulator body. The drivers responses areacquired through the steering wheel angle, brake pressure, andthrottle positioning sensors, as shown in Fig. 12.The motion platform provides kinesthetic cues, which arerelated to the behavior of the vehicle with regard to the humandriver. An actual full-sized braking system, including a vacuumbooster, master cylinder, calipers, etc., is implemented in thesimulator so that the feel of the braking action is similar to that ofan actual vehicular brake pedal. In the case of the steering wheel,a spring and damper are used to produce the reactive forces of thesteering wheel where the spring and damper characteristics areadjusted to make the feel of the steering wheel similar to that ofan actual vehicle being driven in the high-speed range.3.1. Configurations of the driving simulatorThe most important feature of the driving simulator is toguarantee real-time performance and so all the subsystems are0246810121416180-200-100100200020406080100120-10-50510Steering wheel angle degVelocity km/hRoll angle degw/o controlRI-based ROMRI/VS-based UCCw/o controlRI-based ROMRI/VS-based UCCw/o controlRI-based ROMRI/VS-based UCC024681012141618024681012141618Time secTime secSteering wheel angle. Lateral acceleration. 024681012141618Time secTime sec Velocity. Rollover index. 024681012141618024681012141618-1-0.500.5100.511.52-0.500.5Time secTime secLateral acceleration gRollover indexYawrate error deg/secw/o controlRI-based ROMRI/VS-based UCCw/o controlRI-based ROMRI/VS-based UCCw/o controlRI-based ROMRI/VS-based UCC Roll angle. Yaw rate error. Fig. 15. Driving tests results using the full-scale simulator based on the VTT.J. Yoon et al. / Control Engineering Practice 18 (2010) 585597594ARTICLE IN PRESSdesigned and integrated in light of achieving RT performances.The dSPACE Autobox environment is used as the RT simulationhardware in the driving simulator with the CARSIM, vehiclesimulation software as the vehicle model (Han, Yi, & Yi, 2006).The visual graphical engine in Fig. 12 provides a visualrepresentation of the driving situation with the human driverand is composed of prepared 3-D model components, such as theroad surface, street light, tree, guard rail, etc., in a 3-D model of areal-time simulation environment (Han & Yi, 2006b).The human drivers inputs are measured by the brake pressure,the throttle angle and the steering angle sensors, as shown inFig. 12. The steering angle sensor, which produces a gray codewith a synchronous serial interface (SSI), is an absolute-typeencoder, the throttle positioning sensor (TPS) and oil pressuresensor are installed to measure the throttle angle position and thebrake pressure.The vehicle cockpit is mounted on a 3 DOF-1000 kg electricmotion platform, which applies the behavior of the vehicle modelto the simulator body, as shown in Fig. 12 and the motionplatform allows displacements up to a maximum of about710 cm (heave) and 7101 (roll and pitch). The motion platformrenders the linear and angular accelerations of the simulatedvehicle model, as computed by the RT simulation hardware sothat the human driver gets an impression that s/he is driving anactual vehicle by means of the kinesthetic cues generated by themotion platform, and from the visual representation of the drivingsituation provided by the visual graphical engine.3.2. Validation of the vehicle simulatorThe driving simulator used in this paper is evaluated viaactual vehicle test data and Fig. 13 shows the results of a slalomtest in which the driver maintains an approximately constantvehicle speed of about 60 km/h. The cone width is 30 m.The magnitude and frequency of the drivers steering inputsare almost identical in both the vehicle test results and thedriving simulator, as shown in Fig. 13(a). The vehicle responses interms of the yaw rate, the lateral acceleration and the roll angleare also quite similar to the actual test results as shown inFig. 13(b)(d).The comparison between the driving simulator and actualvehicle test results shows that the proposed driving simulator isfeasible for describing actual vehicle dynamic behaviors. Thismeans that the driving simulator accurately reproduces actualdriving conditions.4. Evaluation of the proposed UCC based on a VTTTests using the full-scale driving simulator based on the VTThave been conducted to verify the proposed RI/VS-based UCCcontrol algorithm and its performance with that of the previousRI-based ROM control system are compared. The tests based onthe VTT have been conducted by thirteen drivers, and the resultsare analyzed and summarized here.The test scenario is set to the obstacle-avoidance situationshown in Fig. 14 so that when a driver follows the precedingvehicle moving at a constant speed of 90 km/h in a straight laneand an object is dropped suddenly from the preceding vehicle. Inthis situation, the driver abruptly steers the vehicle to avoid thedropped obstacle and the vehicle is placed in a dangeroussituation. Moreover, in this extreme situation, vehicle rollover ispossible and there may be a loss of maneuverability without anUCC control system.Tests have been conducted by thirteen drivers. Fig. 15 showsthe test results of the first driver, while Fig. 17 shows the vehiclestrajectories. If the UCC control input is not applied, the vehiclerolls over in this situation. It is clear from Fig. 15(e) that the RIincreases over unity in the absence of control. Further, the rollangle and lateral acceleration also increase to large values, asshown in Fig. 15(c) and (d). In addition, because this situation isvery severe, the vehicle deviates from the lane, as shown inFig. 17.It can be seen that the drivers detects the dropped obstacle atabout five seconds and immediately tries to avoid the obstacle bychanging lane. The vehicle velocities at about five seconds of threecases, viz., NON-control, RI-based ROM, and RI/VS-based UCC, aresimilar to each other, as shown in Fig. 15(b).When the UCC control is activated, two of the control systemsyield good resistance to rollover, as shown in Fig. 15(c) and (e). Asthe RI-based ROM system intends to control the vehicle in adirection that is opposite to the drivers intention, the yaw rate-505200300400500600700800XmYmw/o controlRI-based ROMRI/VS-based UCCFig. 17. Trajectories of the vehicle.024681012141618Time secBrake pressures MPaFront-leftFront-rightRear-leftRear-rightRI-based ROM system. 024681012141618051015200246810Time secBrake pressures MPaFront-leftFront-rightRear-leftRear-right RI/VS-based UCC system. Fig. 16. Brake pressures.J. Yoon et al. / Control Engineering Practice 18 (2010) 585597595ARTICLE IN PRESSerror increases significantly at about 7 s, as shown in Fig. 15(f). Atthis point, the steering wheel angle also increases to compensatefor the yaw rate error and to maintain the lane of travel, as shownin Fig. 15(a). It is observed that the yaw rate error and the steeringwheel angle of the RI/VS-based UCC system are maintained atsmaller values than under the RI-based ROM control system, asshown in Fig. 15(a) and (f). In the case of the RI/VS-based UCCsystem, the drivers steering effort for maintaining the lane isreduced by about 33.5% when compared with that under the RI-based ROM system. The roll angle and the yaw rate error are alsoreduced by about 40.7% and 67.3%, respectively.Fig. 16 shows the brake pressures for two cases, namely RI-based ROM, and RI/VS-based UCC. In the case of the RI-based ROMsystem, the brake pressure increases because the controllerintends to move the vehicle in a direction that is opposite to thedrivers steering. Compared with the RI-based ROM case, onlyslight brake pressure is needed to prevent vehicle rollover while,at the same time, improving the maneuverability and lateralstability in the RI/VS-based UCC system.The RI/VS-based UCC system shows the best tracking perfor-mance, as can be seen from Fig. 17. In the case of the RI-basedROM system, the vehicle deviates from the lane and secondaryaccidents may occur in spite of preventing vehicle rollover.The test results of the thirteen drivers are analyzed andsummarized in Table 2. The results of the proposed UCC systemare compared with those of the previous RI-based ROM controlscheme. Compared with the RI-based ROM control system, theproposed RI/VS-based UCC system reduces the drivers steeringeffort for maintaining the lane by up to about 63.9%. The roll angleand yaw rate error are reduced by up to about 55.9% and 67.3%,respectively.5. ConclusionThis paper has described the evaluation of a rollover index(RI)/vehiclestability(VS)-basedunifiedchassiscontrol(UCC) algorithm by using a full-scale simulator on a virtualtest track (VTT). The RI/VS-based UCC system has been proposedand compared with a prior RI-based ROM control system. A two-level control structure, i.e., with upper- and lower-level con-trollers, is adopted in this UCC system, which operates byswitching across three control modes in the upper-level controllerand switching across distribution schemes in the lower-levelcontroller.Real-time human-in-the-loop simulations have been con-ducted to verify the proposed RI/VS-based UCC control algorithmthrough the driving simulator based on the VTT which isdeveloped and used for the evaluation of the RI/VS-based UCCcontrol system under various realistic conditions in a laboratory.One virtual driving test, an obstacle avoidance situation at highspeed, is conducted to evaluate the performance of the rolloverresistance and the vehicle stability aspects. In addition, the testsfor thirteen drivers have also been conducted and the resultsanalyzed. From these test results, it is verified that the proposedUCC system shows good performance for rollover prevention andimproving the maneuverability and the lateral stability. Com-pared with the RI-based ROM system, it is shown that theproposed RI/VS-based UCC system reduces the drivers steeringeffort for maintaining the lane, and reducing roll angle and yawrate error. In particular, from the viewpoint of maneuverability,the RI/VS-based UCC system is shown to be potentially superior tothe RI-based ROM control system in terms of the yaw rate and thetracking error. This implies that the proposed RI/VS UCC systemcan prevent vehicle rollovers while, at the same time, improvingthe maneuverability of the vehicle.For more accurate results, further evaluations under morevariedconditionsarerequiredtosubstantiatetheresultspresented. In addition, the proposed human-in-the-loop evalua-tion can be a good solution for determining o