Bottom-up Approaches to Implementing Driver Models based on Human Factors Experiments Chongkwan Rah *, Kihan Noh, Hyungjeen Choi Body & Chassis Engineering Center, Intelligent Vehicle Technology R&D Division, Korea Automotive Technology Institute (KATECH), 74 Yongjeong-ri, Pungse-myun, Chonan, Chungnam, 330-912, Korea ABSTRACT Objective: This paper introduces and contrasts two different modeling approaches of driver models to implement computer-based vehicle simulation environments for evaluating intelligent driving support systems (IDSS): (1) modular-designed driver models; (2) performance-oriented driver models. Background: As the IDSSs become more intelligent and provide active safety controls, more comprehensive understanding of each facet of a driver s cognitive processes is required as well as vehicle dynamics. That means, the driver-centered evaluation tools implementing the driver's cognitive characteristics and limits are needed for assessing such intelligent subsystems beyond traditional function-oriented vehicle simulations with optimized computational driver models. Method: Two kinds of human factors experiments were conducted to differentiate the conventional driver models into driver-centered models in accordance with the specified driver s generic human factors: (1) As for the modular-designed driver model, one humdred thirty drivers were participated in a psychophysical experiment of velocity estimation at 71 checkpoints on a road section to implement a part of drivers cognitive architecture, i.e., the sensory-perception processes of drivers were statistically modeled. (2) Double lane change maneuvers were conducted to implement differentiated performance-oriented driver models based on the efficiency and driving performance of different groups of drivers. Conclusion: The needs for developing a set of driver-centered simulation environments were clearly identified and the fundamental bottom-up approaches of developing such computational driver models were exhibited for diverse vehicle simulations. Application: The results from this study may ultimately provide meaningful data for early stages of the advanced safety vehicle design. Keywords: human factors in driving, computer-based driver model, vehicle simulations, intelligent driving support systems 1. Introduction 운전자및동승자의안전과안락함추구를목적으로차량에적용되는지능형운전보조시스템 (intelligent driving support system: IDSS) 의발전은전자 / 기계적센서기술과신호처리기술그리고차량동역학적해석기술등을통해눈부시게발전하고있다 (e.g., Bishop, 2005; Cacciabue and Martinetto, 2006; Trivedi, et al., 2007; Miyaji and Oguri, 2010, etc.). 이와같은운전보조시스템은기존차량에적용되던수동적안전장치또는편의장치 (i.e., feedback 신호기반의대응장치 : 안전벨트, 에어백, 충격흡수장치등 ) 뿐만아니라, 주행중인차량의다양한동역학적상황과운전자의제 어입력등으로부터유추된운전의도를분석하여능동적제어를수행하는지능형시스템을의미한다. 이러한시스템의평가과정, 특히개발초기단계에서는, 컴퓨터기반의동역학적해석프로그램을이용하여특정주행환경 (e.g., double lane change [DLC], J-turn, slalom, etc.) 에대한차량의동적특성과한계그리고그에따른운전보조시스템의동작특성을관찰하고설계목적에부합하도록수정해나가는방법을널리활용하고있다. 이러한컴퓨터기반시뮬레이션환경은크게두개의부분으로나누어볼수있으며, 실제차량의기능적각부분 (e.g., 조향장치, 현가장치, 타이어, 등 ) 이가지는정 / 동적특성을해석적으로설명할수있는 (1) 차량동역학모델과, 이러한차량모델이미리정의된시나리오에따라주행할수있도록제어입
력 (i.e., lateral and longitudinal control input) 을계산하고제공 하는 (2) 운전자모델로구분할수있다 (Rah and Park, 2011: Figure 1). Scenario Driving objectives Load information Driving constraints Driver Model Driving environments Lateral control model Longitudinal control model Driving environments Control inputs Vehicle dynamic information Figure 1. Computer-based vehicle simulation environment 그중본연구에서는가상의차량을제어하는운전 자모델개발과정에초점을맞추어, 두가지서로다 른목적과개념으로구성된상향식 (i.e., bottom-up) 접 근방법을소개하고자한다. 두가지방법은기본적으 로점차지능화그리고능동화되어운전자의정보처 리과정전반에대한이해와운전자의인간공학요인 별다양성에대한정보가필수적으로요구되는차별화 된운전자모델개발을목적으로한다. 즉, 인간공학 실험과통계적분석결과를기반으로성별, 연령, 운전 경력등의수준별로구분된운전자모델을구현하기 위한두가지접근방법을소개하고여러가지측면에 서비교분석하였다. 특히, 개발된인간공학요인별 운전자모델이나타내는특정주행환경 (i.e., double lane change) 에서의주행결과와실제운전자의동일조건 하차량주행데이터를체계적으로비교분석하여서 로다른이론적인바탕으로개발된운전자모델이가 지는장단점을여러측면에서설명하였다. 2. Computer-based driver models 지능형운전보조시스템에대한컴퓨터기반적합성 평가과정에포함되어가상차량제어를수행하는운 전자모델구현은궁극적으로차량제어라는 특정상 황 이전제된인간정보처리과정 (Wickens and Hollands, 2000) 모델링으로이해할수있다. 즉, 운전상황에서 발생할수있으며, 차량제어와직 / 간접적인상관을갖 는정보에대한운전자의감각수용및지각과정, 고차 원적인지과정, 반응선택과정, 실행과정등으로나타 낼수있는기능적인지구조를적절한방법으로모델 링하는일련의과정을의미한다. Vehicle Model with IDSS e.g., CarSim 다양한구현방법중본연구에서는다음 2.1, 2.2 절을통해대표적인상향식접근방법두가지를소개하고자한다 : (1) 운전자정보처리과정을수개의인지적기능단위로객체화하여정의하고, 각단계에적합한개별적도구 (e.g., 통계적모델링, 알고리즘기반기능적모델, 의사결정모델, 등 ) 를활용하여통합적인지모델을구현하는 인지적정보처리단계기반운전자모델 (i.e., modular-designed driver model) 구현방법과 ; (2) 특정주행환경에대해실제운전자가도출하는데이터를바탕으로, 인간공학요인수준조합별운전자주행특성을설명할수있는수개의인간공학적지표 (e.g., 조향각, 조향률, 등 ) 또는동역학적평가지표 (e.g., roll, yaw rate, 등 ) 를측정한후, 컴퓨터기반운전자모델이동일한주행시나리오에대해각특성지표를유사하게출력하며인간공학요인별로차별화될수있되도록구현하는 목적성능 또는 목적효율 추구형운전자모델 (i.e., performance-oriented driver model) 을설명하였다. 2.1 Modular-designed driver model 본연구에서소개하는 인지적정보처리단계기반운전자모델 구현방법으로는저자의논문인 Rah and Park (2011) 의연구내용일부에해당한다. 기존연구에서저자는다양한기존연구를바탕으로복합적인지정보처리과정의결과로이해할수있는차량의제어가운전자인간공학요인별로유의한수준이상차별화될수있음 (see Park and Rah, 2005) 을지적하고, 이러한요인별변동은또다시운전자의인지적정보처리과정각단계별효율또는수행도차이에기인하는것임을설명하였다. 또한, 지능형능동제어를수행하는운전보조시스템의평가과정에는이러한단계별차별성을구현할수있는 정보처리과정기반운전자모델 개발이필요함을주장하였다. 이러한맥락에서, 다음과같은연구를진행하였다. (1) 운전자의인지적정보처리과정을독립적기능단위들 (i.e., 감각수용-지각, 인지-예측, 반응선택, 제어출력 ) 로정의하고서로를연동하는개념의운전자모델구조를정의하였다. (2) 그중첫번째단계에해당하는감각수용및지각과정일부, i.e., 실제차량주행속도에대한운전자의체감속도가가지는인지적실증구조 (i.e., cognitive empirical structure: 운전자체감속도모델 ) 를일련의인간공학적실험및통계적모델링과정방법으로정의하고운전자인간공학요인별로차별화하였다. 즉, 미리정의된도로구간의 71개측정지점에서운전자에게구두로체감속도에대한정신물리학적평가과정을진행한후, 운전자체감속도를종속변수로, 운전자인간공학요인 (i.e., 연령, 성별, 운전경력 ) 을정성적독립변수로정의하였으며, 실제차량의
주행속도, 도로곡률반경, 차량실내소음수준을정량 적공변인으로모델에포함시켰다 (Table 1). Table 1. Coefficients of the perceived velocity models Human Factors b 0 b 1 b 2 b 3 MYN -1.281 0.935-0.575 0.692 MON 1.774 0.876-0.057 0.093 FYN 1.727 0.725-0.575 0.692 FON 3.200 0.666-0.057 0.093 MYE 0.647 0.939-1.109 0.414 MOE 1.071 0.880-0.591 0.371 FYE 0.992 0.972-1.109 0.414 FOE -0.166 0.913-0.591 0.371 Note. M: male; F: female; Y: young; O: old; N: novice; E: expert; The general form of the perceived velocity models was PV GAD=b 0+b 1v+b 2ρ+b 3λ, (v, ρ, λ: vehicle velocity, radius of road, level of interior noise; the subscript GAD represents the treatment levels of the human factors, i.e., gender, age (under/over 40), and driving experience (under/over 30,000 Km), respectively. (3) 본연구에서는가상차량에대한제어성능 / 기능중심의운전자모델중운전자횡적제어를실현하는최적예견제어모델 (i.e., optimal finite preview control [OFPC] model: Tomizuka, 1975; Tomizuka and Whitney, 1975) 을운전자중심모델과의비교를위한기준모델로선정하여, Matlab/Simulink (Mathworks Inc., Natick, MA, USA) 를통해구현하였다. 이를기반으로심물리학적실험및통계적모델링과정을통해정의된운전자인간공학요인별 체감속도모델 을수정 / 적용하였으며, 결과적으로운전자감각수용및지각과정효율이차별화된요인별운전자모델집합을정의할수있었다. (4) 특정차량 (i.e., New EF Sonata, Hyundai Motors Co., Ltd., Ulsan, Korea) 의물리적제원을적용하여운용할수있는컴퓨터기반차량동역학모델 (i.e., CarSim : Mechanical Simulation Corporation, Ann Arbor, MI, USA) 과본연구에서개발한가상운전자모델을연동하여통합차량시뮬레이션환경을완성하였다. 마지막으로, (5) 운전자의감각수용및지각과정효율을달리하는인간공학요인별통합차량시뮬레이션환경이나타내는차별화된효과를확인하기위해, DLC 환경 (International Standard Organization [ISO] 3888-1:1999, 1999) 에서의컴퓨터기반시뮬레이션을실시하고, 그결과일부 (i.e., 주행경력을달리하는 40 세이상여성운전자모델의주행특성 ) 를기존기능중심운전자모델의주행결과와비교하면다음 Figure 2 ( 주행궤적 ), Figure 3 ( 조향각 ) 과같다. Figure 2, 3 공통으로붉은색실선으로나타낸결과가본연구의기준운전자모델로선정한 OFPC 모델의주행궤적과상응하는조향각에해당한다. 운전자인간공학요인별차별화를위해수행된기존 OFPC 모델수정이가지는물리적의미는크게두가 지경우로나눌수있다. 운전자체감속도가실제주 행속도보다높게형성된경우와그반대의경우가될 수있으며, 전자의경우운전자모델의제어시점이 OFPC 모델보다앞당겨지고, 후자의경우, 제어시점이그차이만큼뒤로이동하게된다. 이는조향각을나타낸 Figure 3에서보다뚜렷하게이해할수있다. 즉, FOE의경우, 첫번째차선변경구간에서발생한도로곡률반경변화에대해상대적으로 FON보다민감하게체감속도가상승했음을간접적으로알수있다. Figure 2. Vehicle trajectories of female old driver models. Figure 3. Steering angles of female old driver models. 2.2 Performance-oriented driver model 목적성능 추구형운전자모델개발방법은또하나의상향식운전자모델링접근법으로서, 실제차량을이용한주행실험을통해미리정의된특정경로에대한실제운전자제어특성을수개의물리적또는동역학적변수로정의하여수집하고, 이를유사하게실현할수있는컴퓨터기반운전자모델을구현하는과정을의미한다. 이러한실현과정을위해본연구에서는 2.1절에서설명한 DLC 환경에대해 OFPC 모델기반 목적성능 추구형횡적제어모델을다음과같은단계로구현하였다. (1) 앞서설명 (see 2.1절의 (3) 항 ) 한 OFPC 모델을동일한방법으로모델링하여 DLC 환경에서의성능중심최적제어모델을정의하였다. (2) 서로다른인간공
학요인 (i.e., 성별, 연령 (over/under 40), 운전경력 (over/under 30,000 Km)) 을갖는운전자 66명을모집하여실제차량을이용한 DLC 환경주행실험을실시하고, 인간공학요인별주행특성을수집하였다. 인간공학적특성으로는운전자가출력하는조향각 (degree) 과조향률 (steering angular rate: degree/sec) 이며, 해당제어입력으로부터출력되는동역학적요인으로는동일한데이터수집시점에서출력된 roll angle (degree), yaw rate (degree/sec), 그리고 lateral acceleration (m/sec 2 ) 등이해당된다. (3) (1) 항에서설명한 OFPC 모델에포함되어제어특성을조율할수있는변수 (i.e., tuning parameter: see Kang, et al., 2008) 를선정하여의미를부여하였다. 즉, OFPC 모델의핵심적인 feedforward 제어결정요인에해당하는운전자전방주시거리 (preview sight distance: Peng and Tomizuka, 1993; Tung and Tomizuka, 1993) 를비롯하여, 운전자인간공학요인별차별화가예상되는근골격계효율관련제어지연, 조향제어입력에대한다양한제어민감도등의조정가능한변수를선정하여, 각변수의조정이초래할수있는운전자모델특성변화의물리적의미를정의하였다. (4) (2) 항에서정의한운전자인간공학요인별주행특성을구현할수있는차별화된운전자모델개발을위해 (3) 항의조정가능변수들을조율하여각운전자모델의실험결과를근사화한다. 효율적인조율과정을위해널리사용되는정형화된방법론은존재하지않으므로, 각조정가능변수가가지는물리적의미를바탕으로반복적조율과정을통해근사화할수있다. 하지만, 각조정가능변수들은서로독립적이지않으며, 각운전자모델의궁극적출력인횡적제어수준과각조정가능변수는매우복잡한비선형특성을나타내므로이러한조율과정은많은시간과노력을요구하는작업이된다. 그과정의결과일부를나타내면 Figure 4, 5와같다. 붉은색실선으로나타낸주행궤적및조향각은인간공학요인고려가배제된기존의성능위주운전자모델이출력하는이상적주행결과에해당하며, 파란색실선은실제운전자의 DLC 주행결과, 그리고파란색점선이위과정을통해실제주행결과를근사한 목적성능 추구형운전자모델의결과에해당된다. 시각적으로주행궤적이나조향각의경우실제운전자의주행결과와기대이상의차이가나타날수있으나, Figure 6에나타낸바와같이, 본연구의근사화과정에서선정된 목적성능 에해당하는여섯가지지표를남자운전자모델의경우와여성운전자모델의경우로나누어비교 / 분석하면차별화되는그차이와목적된근사성을이해할수있다. Figure 6에나타낸여섯가지목적성능은 Figure 2, 4 에나타난 DLC 구간을지나는차량과운전자가출력한해당물리량의제곱평균 (i.e., root mean square) 을 의미한다. 개별적물리량의제곱평균은다양한제어 특성또는동역학적특성을하나의대표척도로나타 내어많은정보손실을가져올수있으나, 서로유기 적으로연관된물리량을동시에고려한경우로서전체 적인제어특성및주행특성을많은부분설명할수 있을것으로기대된다. Yaw rate (degree) Figure 4. Vehicle trajectories of female drivers/driver models. Figure 5. Steering angles of female drivers/driver models. Angular rate (degree/sec) Roll angle (degree) Steering angle (degree) Lateral Acceleration (m/sec 2 ) Driver Model (Female) Real Driver (Female) Real Driver (Male) Lateral error (m) Figure 6. Six target parameters for developing a performance- oriented driver models (Female driver model)
3. Discussions and Conclusions 일련의운전자제어입력이나간단한생체신호변화등을기반으로능동적으로추정한운전자의주행의도를시스템운용핵심정보로활용하는지능형 / 능동형운전보조장치의발전은기존의기능중심컴퓨터기반시뮬레이션환경에서의시스템타당성평가과정을단순히기능적또는구조적해석과정일부로그위상을축소시킬것으로판단된다. 즉, 운전자의전반적인정보처리과정에대한이해와적용이없는컴퓨터기반차량시뮬레이션환경은현재활발하게발전되고있는 IDSS 평가에부적합하다할수있다. 이에본연구에서는, 운전자의정보처리과정에대한이해와적용을위한두가지서로다른상향식접근방법론을소개하였다. 인지정보처리단계기반의운전자모델과목적성능추구형운전자모델을의미하며, 모두인간공학요인수준조합별운전자모델집합을구성하고, IDSS 평가등에활용될수있다는점에서는동일한의미를가질수있다. 인지정보처리단계기반의운전자모델의경우, 각정보처리단계별특성화된구현방법으로객체화하여개발, 서로를연동함으로써, 보다자유도높은정교한운전자모델을완성할수있다는장점이있고, 특히, 통계적으로정의될수있는감각수용및지각과정등은해당인간공학실험의분석결과로얻을수있는확률적정보를적극적으로활용할수있으며, 이는차량시뮬레이션환경자체를결정적시스템이아닌확률적시스템으로운용할수있는가능성을제공한다. 즉, 동일한입력에대해결정적결과만을제공하던기존의시뮬레이션환경과달리, 반복적시뮬레이션과정을통해, 발생확률이낮은극한의상황에대한평가과정을가능하게할수있어불확실성이큰지능형시스템또는능동형안전보조시스템평가에매우유용하게활용될수있을것으로기대된다. 반면, 상대적으로일반적인운전자모델, 즉, 많은시나리오에대응할수있는운전자모델을구축하는과정에해당하므로, 완성도를높이기위해서는다양한측면의운전자특성을탐색하고적용하기위한끊임없는수정과정이요구된다. 목적성능추구형운전자모델의경우, 운전자의인간공학요인수준조합별로다르게나타나는특정시나리오주행에대한맞춤형모델링과정으로이해할수있으며, 평가시나리오와목적이제한적인경우효율적으로활용될수있다. 하지만, 이러한운전자모델의경우, 그확장성이제한될수있으며, 기존에활용되어온가상차량제어모델을인간공학요인별로차별화하기위해그일부를수정한블랙박스모델 (e.g., Chen and Ulsoy, 2001; Salvucci, 2006) 에그칠수있다. 즉, 인지정보처리단계기반의운전자모델과목적성능추구형모델의제어성능일부를나타낸 Figure 2, 3, 그리고 Figure 4, 5를비교할때, 실질적으로실제운전자제어특성에가까운모델은목적성능추구형모델의경우에해당하지만, 주행시나리오가조금만바뀌어도새로운변수조율과정 (i.e., parameter tuning) 이요구되며, 해당시나리오에대한실험과정이반복적으로요구될수있다. 게다가, 보다복잡한주행시나리오에대응하기위해서는본연구에서활용한 DLC 환경운전자모델구현의경우보다더많은조율과정이요구될것이며, 이또한하나의새로운연구과제로등장하게될것이다. 인지적정보처리단계기반의운전자모델경우도이러한시스템자유도문제에서자유로울수없다. 즉, 본연구에서소개한운전자모델은횡적제어입력 (e.g., 조향, 차선변경, 등 ) 을결정하기위한기준모델을활용한구현사례에해당하므로, 종적제어 (e.g., 가속, 감속, 등 ) 가요구되는경우에는새로운기준모델구현과정과새로운조율과정이요구된다. 또한, 운전자정보처리과정이해를바탕으로한운전자모델구현이라는측면에서는높이평가될수있으나, 그결과로나타난운전자모델의제어가과연얼마나실제운전자특성을반영할수있는지, 또는, 실제와상이한측면이발견되었다면정보처리단계의어느부분을수정하고보완해야하는가하는매우어려운과제가발생할것이며, 그결과이론적모델에그칠수있다. 따라서, 두가지모델링방법의장점을살릴수있는혼합적방법론개발의필요성을조심스럽게제안하는바이다. 본연구의가장커다란의의는, 운전자의인간공학요인별로제어특성이유의하게달라질수있음을이해하고, 이러한차별성을설명할수있는컴퓨터기반의운전자모델개발이필요하며, 일반적인인간공학실험혹은정신물리학적실험과통계적방법, 그리고동역학적제어이론을혼합할수있는방법을제안하고보다실질적인연구와도전을시작하는것이라할수있다. 향후, 인간공학요인별로다른특성을나타내는운전자제어특성에대한정보처리단계별기초연구와이러한차별성을실현하기위한제어기설계연구가끊임없이이어진다면, 보다지능화되고능동적인운전보조장치평가과정에유용한도구로서활용될것이며, 궁극적으로는운전자에대한근본적이해가높아져새로운지능형시스템개발에활용될수있을것으로기대된다.
References pp.10-14. Prentice Hall, Upper Saddle River, NJ, 3rd edition, 2000. Bishop, R., IV application areas. In Intelligent Vehicle Technology and Trends, pages 25-38. Artech House, Inc., Boston, 2005. Cacciabue, P., & Martinetto, M. (2006). A user-centred approach for designing driving support systems: the case of collision avoidance. Cognition, Technology & Work, 8(3), 201-214. Chen, L.-K. and Ulsoy, A. G., Identification of a driver steering model, and model uncertainty, from driving simulator data. Journal of Dynamic Systems, Measurement, and Control, 123(4):pp.623-629, 2001. International Standard Organization [ISO] 3888-1: 1999 (1999). Passenger cars: Test track for a severe lane-change manoeuvre - Part 1: Double lane-change. Kang, J., Hindiyeh, R. Y., Moon, S.-W., Gerdes, J. C., and Yi, K. (2008). Design and testing of a controller for autonomous vehicle path tracking using GPS/INS sensors, 17thWorld Congress of the International Federation of Automatic Control, Seoul, Korea (pp. 2093 2098). Miyaji, M., Kawanaka, H., & Oguri, K. (2010, 19-22 Sept. 2010). Effect of pattern recognition features on detection for driver's cognitive distraction. Paper presented at the Intelligent Transportation Systems (ITSC), 2010 13th International IEEE Conference on. Park, M. and Rah, C. Human factors considerations for Human-Vehicle Interactive Models to enhance safety of intelligent vehicles. Paper presented at the XIX Annual International Occupational Ergonomics and Safety Conference, 46~51, Las Vegas, Nevada, USA, 2005. Peng, H., & Tomizuka, M. (1993). Preview control for vehicle lateral guidance in highway automation. Journal of Dynamic Systems, Measurement, and Control, 115(4), 679 686. Rah, C. and Park, M., Statistical modeling of the sensory-perception process of computational driver models for vehicle simulations, Human Factors and Ergonomics in Manufacturing & Service Industries, 21(2), 188-206, 2011. Salvucci, D.D., Modeling driver behavior in a cognitive architecture. Human Factors: The Journal of the Human Factors and Ergonomics Society, 48(2):pp.362-380, 2006. Tomizuka, M. (1975). Optimal continuous finite preview problem. Automatic Control, IEEE Transactions on, 20(3), 362-365. Tomizuka, M., & Whitney, D. E. (1975). Optimal discrete finite preview problems (Why and how is future information important?). Journal of Dynamic Systems, Measurement, and Control, 97(4), 319-325. Trivedi, M. M., Gandhi, T., & McCall, J. (2007). Looking-In and Looking-Out of a Vehicle: Computer-Vision-Based Enhanced Vehicle Safety. IEEE transactions on Intelligent Transportation Systems, 8(1), 108-120. Tung, E. D., & Tomizuka, M. (1993). Feedforward tracking controller design based on the identification of low frequency dynamics. Journal of Dynamic Systems, Measurement, and Control, 115(3), 348 356. Wickens, C. D. and Hollands, J. G., A model of human information processing. In Engineering psychology and human performance, Author listings Chongkwan Rah: ckrah@katech.re.kr Highest degree: PhD, Department of Industrial Engineering, Hanyang University Position title: Researcher at Body & Chassis Engineering Center, Intelligent Vehicle Technology R&D Division, Korea Automotive Technology Institute (KATECH) Areas of interest: driver models, statistical modeling, pattern recognition Kihan Noh: khnoh@katech.re.kr Highest degree: PhD Candidate, Department of Mechanical Engineering, KAIST Position title: Team Leader/Senior Researcher at Body & Chassis Engineering Center, Intelligent Vehicle Technology R&D Division, Korea Automotive Technology Institute (KATECH) Areas of interest: Smart/Intelligent vehicle, HILS, driver models Hyungjeen Choi: hjchoi@katech.re.kr Highest degree: PhD Candidate, Department of Mechanical Engineering, KAIST Position title: Senior Researcher, Body & Chassis Engineering Center, Intelligent Vehicle Technology R&D Division, Korea Automotive Technology Institute (KATECH) Areas of interest: Smart/Intelligent vehicle, HILS, driver models, intelligent chassis systems