Motion sickness evaluation and comparison for a static driving simulator and a dynamic driving simulator

HAL Id: hal-01017498

https://hal.archives-ouvertes.fr/hal-01017498

Submitted on 2 Jul 2014

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

Motion sickness evaluation and comparison for a static driving simulator and a dynamic driving simulator

Baris Aykent, Frédéric Merienne, Christophe Guillet, Damien Paillot, Andras Kemeny

To cite this version:

Baris Aykent, Frédéric Merienne, Christophe Guillet, Damien Paillot, Andras Kemeny. Motion sick-

ness evaluation and comparison for a static driving simulator and a dynamic driving simulator. Pro-

ceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, SAGE

Publications, 2014, 228 (7), pp.818-829. �hal-01017498�

Science Arts & Métiers (SAM)

is an open access repository that collects the work of Arts et Métiers ParisTech researchers and makes it freely available over the web where possible.

This is an author-deposited version published in: http://sam.ensam.eu Handle ID: .http://hdl.handle.net/10985/8314

To cite this version :

Baris AYKENT, Frédéric MERIENNE, Christophe GUILLET, Damien PAILLOT, Andras KEMENY - Motion sickness evaluation and comparison for a static driving simulator and a dynamic driving simulator - Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering - Vol. 228, n°7, p.818-829 - 2014

Any correspondence concerning this service should be sent to the repository

Administrator : archiveouverte@ensam.eu

Mot ion s ickness eva luat ion and compar ison for a stat ic dr iv ing s imu lator and a dynam ic dr iv ing s imu lator

Bar is Aykent ¹ , Freder ic Mer ienne ¹ , Chr istophe Gu i l let ¹ , Dam ien Pa i l lot ¹ and Andras Kemeny ^{1 ,2}

Abstract

Th is paper dea ls w ith dr iv ing s imu lat ion and in part icu lar w ith the important issue o f mot ion s ickness . The paper pro- poses a methodo logy to eva luate the ob ject ive i l lness rat ing metr ics deduced from the mot ion s ickness dose va lue and quest ionna ires for both a stat ic s imu lator and a dynam ic s imu lator . Acce lerat ions o f the vest ibu lar cues (head move- ments) o f the sub jects were recorded w ith and w ithout mot ion p lat form act ivat ion . In order to compare user exper i- ences in both cases , the head-dynam ics-re lated i l lness rat ings were computed from the obta ined acce lerat ions and the mot ion s ickness dose va lues . For the sub ject ive ana lys is , the pr inc ipa l component ana lys is method was used to deter - m ine the con f l ict between the sub ject ive assessment in the stat ic cond it ion and that in the dynam ic cond it ion . The pr inc i- pa l component ana lys is method used for the sub ject ive eva luat ion showed a cons istent d i f ference between the answers g iven in the s ickness quest ionna ire for the stat ic p lat form case from those for the dynam ic p lat form case . The two-ta i led Mann–Wh itney Utest shows the s ign i f icance in the d i f ferences between the se l f-reports to the ind iv idua l quest ions . Accord ing to the two-ta i led Mann–Wh itney U test , exper ienc ing nausea (p= 0 .019\ 0 .05) and d izz iness (p = 0 .018\ 0 .05) decreased s ign i f icant ly from the stat ic case to the dynam ic case . A lso , eye stra in (p=0 .047\ 0 .05) and t iredness (p=0 .047\ 0 .05) were reduced s ign i f icant ly from the stat ic case to the dynam ic case . For the percept ion f ide- l ity ana lys is , the Pearson corre lat ion w ith a con f idence interva l o f 95% was used to study the corre lat ions o f each ques- t ion w ith thexi l lness rat ing component IR

x

, theyi l lness rat ing component IR

y

, thezi l lness rat ing component IR

z

and the compound i l lness rat ing IR

tot

. The resu lts showed that the long itud ina l head dynam ics were the ma in e lement that induced d iscom fort for the stat ic p lat form , whereas vert ica l head movements were the ma in factor to provoke d iscom- fort for the dynam ic p lat form case . A lso , for the dynam ic p lat form , latera l vest ibu lar - leve l dynam ics were the ma jor e le- ment wh ich caused a fee l ing o f fear .

Keywords

Dr iv ing s imu lator , dr ivers percept ion , inert ia l cue , vest ibu lar - leve l dynam ics , mot ion s ickness

Date rece ived : 20 January 2013 ; accepted : 18 November 2013

Introduct ion

Sensor ia l cues (v isua l , aud itory , hapt ic , inert ia l , vest ibu- lar and neuromuscu lar) p lay important ro les in repre- sent ing a proper percept ion in dr iv ing s imu lators .

¹

A dr iv ing s imu lator a ims to g ive the sensat ion of dr iv ing as in a rea l case . To fu lf i l th is ob ject ive , the dr iv ing s imu lator must enhance the v irtua l immers ion of the sub ject in a dr iv ing s ituat ion . For that reason , rest itut- ing the inert ia l cues on dr iv ing s imu lators is important in order to ach ieve certa in goa ls by s imu lat ions , recog- n iz ing that not a l l s imu lators can ach ieve th is .

²

Because of the restr icted workspace , it is not poss ib le to

represent the veh ic le dynam ics cont inuous ly w ith a one- to-one sca le on the mot ion p latform . Neverthe less , the most des ired a im is to m in im ize the dev iat ion between

1

Le Laborato ire dE lectron ique , dIn format ique et de lImage (Un ite ´M ixte de Recherche 6306 du CNRS) , Arts et Me ´t iers Par isTech , Cha lon-sur - Sao ˆne , France

2

Techn ica l Centre for S imu lat ion , Renau lt , Guyancourt , France Correspond ing author :

Bar is Aykent , Le Laborato ire dE lectron ique , dIn format ique et de lImage (Un ite ´M ixte de Recherche 6306 du CNRS) , Arts et Me´t iers Par isTech , 2 Rue T Dumorey , 71100 Cha lon-sur -Sao ˆne , France .

Ema i l : b .aykent@gma i l .com

the acce lerat ions from the v isua l ly (veh ic le mode l) rep- resented dynam ics and those from the inert ia l ly (mot ion p latform) represented dynam ics as rea l ist ica l ly as poss i- b le depend ing on the dr iv ing task .

³

Dr iv ing s imu lat ion s ickness has been assessed us ing dynam ic and stat ic s imu lators in some stud ies .

^4–6

For a brak ing manoeuvre , S ieg ler et a l .

⁷

stated that , if the mot ion p latform is act ivated , the b ias in reach ing increased leve ls of dece lerat ions is reduced in compar i- son w ith the case when the mot ion p latform is inact i- vated . Th is ind icates a more rea l ist ic representat ion of the veh ic le dynam ics , wh ich can be interpreted as a reduc ing effect on mot ion s ickness . In order to reduce the s imu lator s ickness , the d ifference between the acce l- erat ions through the v isua l and the vest ibu lar cues has to be m in im ized .

^3,6,7

The vest ibu lar cues correspond to the head movements of the sub jects dur ing the dr iv ing s imu lator exper iments . The dynam ics of the vest ibu lar cues are def ined by the long itud ina l , latera l , vert ica l and vector ia l compound dynam ics (see F igure 1 for the mu lt i-sensory integrat ion and mot ion s ickness mechan- isms) . Here , the i l lness rat ings are def ined as fo l lows : the component i l lness rat ings refer to the i l lness rat ings induced by the long itud ina l dynam ics , the latera l dynam ics and the vert ica l dynam ics separate ly , whereas the compound i l lness rat ing corresponds to the vector- ia l compound of the long itud ina l , latera l and vert ica l i l lness rat ings . The a im of th is paper is to address s imu- lator mot ion s ickness as a corre lated funct ion of the component i l lness rat ings and the compound i l lness rat- ings for vest ibu lar cues (head movements) us ing s ick- ness quest ionna ires . Th is research work was performed under stat ic and dynam ic operat ions us ing the dr iv ing

s imu lator ca l led the S imu lateur Automob i le Arts et Met iers (SAAM) .

The ob ject ives of th is work are to propose a b imoda l measurement approach to obta in ob ject ive and sub ject ive data , and the ir corre lat ion w ith each other to mon itor the s ickness leve l of the test sub jects dur ing the dr iv ing ses- s ion , f irst , as a method to determ ine the mot ion s ickness components ( long itud ina l , latera l , vert ica l or tota l) wh ich are benef ic ia l to represent ing the mot ion s ickness of the test sub jects as an ob ject ive measure (name ly the i l lness rat ing) , second , as a techn ique to measure sub ject ive impress ions of the test part ic ipants through a mot ion s ickness quest ionna ire and , th ird , for corre lat ion between the sub ject ive and the ob ject ive data .

Th is paper a ims to invest igate the s ign if icance leve l of the fo l low ing hypotheses . Is there any s ign if icant corre lat ion between the vest ibu lar- leve l i l lness rat ing (ob ject ive measure) and the percept ion of s ickness (sub- ject ive measure) in the case of the stat ic s imu lator and in the case of the dynam ic s imu lator? Is there any per- ceptua l d ifference between the stat ic s imu lator and the dynam ic s imu lator in terms of mot ion s ickness? (Here , the perceptua l d ifference refers to the sub ject ive eva lua- t ion d ifference for the stat ic operat ion and for the dynam ic operat ion of the s imu lator mot ion p latform regard ing our s imu lator s ickness quest ionna ire (see Tab les 3 and 4 later) .)

Potent ia l factors induc ing the s ickness in v irtua l rea l- ity systems can be sp l it into three ma jor groups :

⁹

ind i- v idua l factors , s imu lator factors and s imu lated task factors .

Ind iv idua l factors refer to the sens it iv ity to the s imu-

lator cond it ions (adaptat ion) , the postura l stab i l ity , etc .

F igure 1 . Structure o f the c losed- loop contro l o f the dynam ic dr iv ing s imu lator .

⁸

In genera l , they address the dr iver and h is or her beha- v iour dur ing the dr iv ing sess ion (F igure 1) . S imu lator factors refer to whether the s imu lator has a mot ion p latform or not (a dynam ic dr iv ing s imu lator or stat ic dr iv ing s imu lator , seen as the dr iv ing s imu lator techno l- og ies in F igure 1) . S imu lated task factors refer to the head movements (the vest ibu lar react ion of the dr iver) w ith respect to the road scenar io , the veh ic le dynam ics and the dr iv ing s imu lator techno log ies used for the exper imenta l tr ia ls . Two types of dr iv ing s imu lator com- mon ly used are dynam ic s imu lators (mov ing-base s imu- lators) and stat ic s imu lators (f ixed-base s imu lators) . It is reported that s imu lator s ickness is less l ike ly to occur in mov ing-base dr iv ing s imu lators .

^{4–6,10–14}

Prev ious ly , when dynam ic s imu lators and stat ic s imu lators have been compared , it has been suggested that there is a re la- t ion between the i l lness and the head movements of the p i lot .

¹⁴

The mot ion s ickness mechan isms in a dr iv ing s it- uat ion can be summar ized by F igure 1 . Accord ing to F igure 1 , the road scenar io and the veh ic le dynam ics mode l are formed in the v isua l system of the dr iv ing s imu lator . Compared w ith the stat ic p latform case , the dynam ic p latform cond it ion a lso has a hexapod mot ion p latform and mot ion cue ing a lgor ithms (see F igure 1) wh ich convert the veh ic le dynam ics to restr icted p lat- form dynam ics . In the study that we are current ly pre- sent ing here , the contr ibut ion of the mot ion p latform to mot ion s ickness is invest igated . Apart from the dynam ic systems , other cues (aud itory cues for eng ine and traff ic sounds , v is ion system rest itut ion to feed back the dr iven env ironment , and force feedback steer ing whee l as hapt ic cues to prov ide the steer ing fee l ing c loser to rea l-wor ld cond it ions) were added to dr iver- in-the- loop system to make the dr iv ing s imu lat ion more rea l ist ic (F igure 1) .

One phenomenon c lose ly invo lved w ith s imu lator s ickness is i l lusory se lf-mot ion due to a v isua l input , known as vect ion wh ich is inc luded in the s imu lated task factors .

²

If the i l lusory se lf-mot ion is due to an inert ia l input , it is ca l led somatograv ic i l lus ion.

Kennedy et a l .

¹⁵

stated that v isua l representat ions of mot ion affected the vest ibu lar system . Thus , they con- c luded that the mot ion patterns represented in the v isua l d isp lays of s imu lators may exert inf luences on the vest ibu lar system . The research l iterature from var i- ous types of vect ion study , inc lud ing those invo lv ing exposure to v irtua l env ironment systems , has shown that mot ion s ickness is a common s ide effect of v iew ing v isua l scenes of se lf-mot ion w ithout phys ica l move- ment .

^16–19

However , wh i le vect ion is corre lated w ith v isua l ly induced s ickness , Lawson et a l .

²⁰

ma inta ined that vect ion is not a necessary tr igger of symptoms . Peop le who exper ience vect ion are more l ike ly to expe- r ience s ickness .

^17,18

The inc lus ion of a mot ion base is a s imu lator-re lated factor that has been shown to affect s imu lator s ickness .

²

The p latform of a s imu lator is e ither a f ixed base or a mot ion base . In a f ixed-base s imu lator , informat ion regard ing se lf-mot ion is pro- v ided mere ly by the v isua l system . In contrast , a mot ion-base s imu lator prov ides a subset of the inert ia l

forces that wou ld be present dur ing rea l movement in the veh ic le be ing s imu lated .

^10,21

In part icu lar , a mot ion- base s imu lator can prov ide two types of inert ia l cues : acce lerat ion and t i lt .

^13–15

H igh-f ide l ity mot ion-base sys- tems are extreme ly expens ive , but they are used in spe- c if ic app l icat ions (e .g . f l ight s imu lators) to enhance the sense of se lf-mot ion prov ided by the v isua l d isp lay .

¹⁰

However , a mot ion-base s imu lator can prov ide mot ion cues compat ib le w ith the in it ia l , but not susta ined , acce lerat ion .

¹⁰

For examp le , forward acce lerat ion can be s imu lated by p itch ing the base backwards (t i lt coor- d inat ion) wh i le a lso trans lat ing it forwards (onset acce l- erat ion) s l ight ly .

¹⁰

V isua l movement through a s imu lated env ironment that is not accompan ied by the norma l inert ia l cues (e .g . forces and acce lerat ions) assoc iated w ith move- ment through the rea l env ironment m ight induce mot ion s ickness ,

^10,11,13

part icu lar ly nausea .

^22,23

For examp le , mot ion s ickness has been def ined in terms of metr ics re lat ing vom it ing inc idents .

^22,23

Consequent ly , the overa l l inc idence of s imu lator s ickness is typ ica l ly lower in s imu lators w ith a mot ion base than in those w ith a f ixed base .

¹²

Kennedy et a l .

¹⁴

suggested that one of the reasons that s imu lator s ickness inc idence was lower in mot ion-base s imu lators than in f ixed-base s imu lators was because of the d ifferences in the p i lots head movements dur ing exposure . They exp la ined that , in a mov ing-base s imu lator , the p i lots head movements were s im i lar to those in the actua l veh ic le whereas the head movements in f ixed-base s imu lators were often in conf l ict w ith the inert ia l st imu lus , wh ich augmented the contrad ict ion of the s imu lat ion . There have been , how- ever , a few reports that contrad ict the genera l f ind ings of a d ifference between the s ickness inc idence in a f ixed-base s imu lator and that in a mot ion-base s imu la- tors . For examp le , a study by McCau ley and Sharkey

¹³

obta ined a re lat ive ly s im i lar inc idence of s imu lator s ickness in a mot ion-base he l icopter s imu lator to that in a f ixed-base s imu lator .

The proposed approach cons ists of eva luat ing the mot ion s ickness at the dr iver leve l . It can be exp la ined as vest ibu lar dynam ics ( long itud ina l , latera l and vert i- ca l acce lerat ions) measurement from the r ight-ear leve l of the sub jects v ia the XSens mot ion tracker , wh ich is connected to a headphone . The long itud ina l , latera l and vert ica l acce lerat ions of the head movements were measured for th is approach and converted into the i l l- ness rat ing by us ing the mot ion s ickness dose va lue (MSDV) approach .

²²

In the l iterature , to be ab le to rate the s imu lator s ickness , the s imu lator s ickness quest ion- na ire ,

⁹

the mot ion s ickness quest ionna ire

²⁴

and the b io- feedback method

²⁵

are common ly app l ied w ith some other approaches such as the MSDV .

^22,23

A lthough the dr iv ing s imu lators ma in ut i l izat ion

doma in is tra in ing , it can be used to eva luate the

mot ion and s imu lator s ickness . We focused on the rep-

resented dynam ics f ide l ity ( i .e . how c lose the dynam ics

are to the rea l veh ic le dynam ics) for two p latform con-

d it ions and the ir effects on the part ic ipants vest ibu lar-

leve l-sensed i l lness rat ings (v ia the ob ject ive measures shown later in F igure 3) , the sub ject ive reports and the corre lat ion of sub ject ive and ob ject ive data .

Kennedy et a l .

²⁶

stud ied the v isua l ly induced mot ion s ickness . They reported that the contr ibut ion of the e le- ments to s imu lator s ickness is l ined up from max imum to m in imum in the v isua l ly induced mot ion s ickness as O.N.D (where O represents ocu lomotor fat igue , N represents nausea and D represents d isor ientat ion) .

Another study , wh ich was conducted by Drex ler ,

²⁷

revea led that the contr ibut ion order obta ined from the s imu lator s ickness quest ionna ire for dr iv ing s imu lators is O.D.N . In our study , the h ighest pr inc ipa l com- ponent was pr inc ipa l component 1 (d isor ientat ion-asso- c iated s ickness , i .e . nausea + d izz iness) and the second- h ighest pr inc ipa l component was pr inc ipa l component 2 (ocu lomotor fat igue , i .e . eye stra in + t iredness) .

Mot ion s ickness was invest igated by G ianaros et a l .

²⁸

v ia a mu lt i-d imens iona l method by us ing a mot ion s ickness assessment quest ionna ire that was pre- sented to part ic ipants who were exposed to a rotat ing optok inet ic drum . The resu lts from the mot ion s ickness assessment quest ionna ire were corre lated strong ly w ith the overa l l scores from the Pensaco la d iagnost ic index (r= 0 .81 ; p\ 0 .001) and the nausea prof i le (r= 0 .92 ; p\ 0 .001) .

²⁸

It was found that not on ly is the mot ion s ickness assessment quest ionna ire a va l id eva luat ion too l but a lso it is advantageous to use th is mu lt i- d imens iona l quest ionna ire rather than the one- d imens iona l form .

²⁸

The advantage of the pr inc ipa l component ana lys is (PCA) method in our study seems to be its capab i l ity to s imp l ify the interpretat ion of the mu lt i-d imens iona l corre lat ions compared w ith the factor ana lys is method . Depend ing on the PCA method , it was revea led that the sub ject ive se lf-report ana lys is was negat ive ly corre- lated between the stat ic p latform case and the dynam ic p latform case for pr inc ipa l component 1 (the h ighest pr inc ipa l component wh ich was tota l ly made up of the d isor ientat ion-re lated s ickness regard ing nausea and d izz iness) . The study by G ianaros et a l .

²⁸

shows a s im i- lar ity to our invest igat ion in terms of the s ickness pro- f i le in wh ich nausea is a pr inc ipa l factor .

Dur ing the dr iv ing s imu lator exper iments , the same dr iv ing scenar io of a doub le- lane-change manoeuvre w ith a constant ve loc ity of 60 km/h w ith the same cond i- t ions (name ly the same veh ic le mode l , long itud ina l ve lo- c ity and terra in) was used on a stat ic p latform and on a dynam ic p latform w ith the software SCANeRstud io vers ion 1 .1 from OKTAL .

Mo t ion s ickness dose va lue

The MSDV is one of the methods used to ob ject ify the mot ion s ickness rat ings and was def ined in accor- dance w ith ISO 2631-1 :1997 .

²²

In that work , an i l lness rat ing method , der ived from the MSDV , was ut i l ized . The mathemat ica l express ion for the MSDV is g iven

later in equat ion (5) . Accord ing to ISO 2631-1 :1997 , the r .m .s . acce lerat ion va lues on a l l axes are def ined to ref lect more c lose ly the hea lth hazard to wh ich the human body is exposed . The coeff ic ients are descr ibed by ISO 2631-1 :1997 on the bas is of the frequency and the d irect ion of v ibrat ion to wh ich the body is exposed . The coeff ic ientsw

k

= 0 .426 (cepha locauda l ax is) andw

d

= 0 .067 (anteroposter ior and med io lat- era l axes) were used to obta in the frequency-we ighted r .m .s . acce lerat ion on a l l the axes (see equat ion (1)) . For eva luat ion of the hea lth effects ,k

x

=1 .4 , k

y

= 1 .4 andk

z

= 1 (for the long itud ina l d irect ion , the lat- era l d irect ion and the vert ica l d irect ion respect ive ly) are chosen from the work by Abercromby et a l .

²⁹

The r .m .s . i l lness rat ing va lues at a vest ibu lar (sub jects head movements) leve l were computed by subst itut ing thea

x

,a

y

anda

z

va lues for the vest ibu lar leve l

^23,29,30

accord ing to

a RMS = ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi k x w k a x

ð Þ ² + k y w d a y 2 + k ð z w d a z Þ ²

q ð1Þ

a xRMS = ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi k x w k a x

ð Þ ²

q ð2Þ

a yRMS = ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi k y w d a y 2

q ð3Þ

a zRMS = ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi ffi k z w d a z

ð Þ ²

q ð4Þ

MSDV tot = ð ^t

0

a ² _RMS ðÞdt t 2

4

3 5

0: 5

m=s ³⁼² ð5Þ

IR x = 1 50MSDV ^x

= ð ^t

0

a ² _xRMS ðÞdt t 2

4

3 5

0: 5

m=s ³⁼² ð6Þ where MSDV

tot

(m/s

^3/2

) is the tota l mot ion s ickness dose va lue

²³

anda RMS (m/s

²

) is the r .m .s . acce lerat ion . The long itud ina l i l lness rat ing is ca lcu lated from equa- t ion (6) . The latera l and the vert ica l i l lness rat ings are ca lcu lated us ing the same process as in equat ion (6) by ut i l iz ing the correspond ing acce lerat ions . The tota l i l l- ness rat ing , denoted IR

tot

, deduced from the MSDV is g iven by

IR tot = 1 50MSDV ^tot ð7Þ Accord ing to the resu lts obta ined for the i l lness rat ing (m/s

^3/2

) , the i l lness rat ing scores are ranked as 0 , 1 , 2 , 3 and greater than 3 as fo l lows .

^22,23

I l lness rat ing = 0 : I fe lt good . I l lness rat ing = 1 : I fe lt a m i ld i l lness . I l lness rat ing = 2 : I fe lt very bad .

I l lness rat ing5 3 : I fe lt abso lute ly terr ib le .

Methods and mater ia ls Proposed approach

Th is study was carr ied out to understand and rate the effect of p latform act ivat ion in terms of ob ject ive assessments as we l l as sub ject ive assessments .

F igure 2 shows the approach that we used for enab l ing the corre lat ion (the Pearson corre lat ion w ith a conf idence interva l of 95%) between ob ject ive data and sub ject ive data in th is research to be made . Accord ing to th is f igure , we computed the i l lness rat ings on a ves- t ibu lar leve l dur ing the execut ion of the dr iv ing exper i- ments on the SAAM .

The SCANeRstud io and X-Sens measurements are separate measures obta ined as from a veh ic le leve l (a veh ic le mode l wh ich moves in the v isua l env ironment) and as from a vest ibu lar leve l (head dynam ics from the r ight-ear a l ignment) respect ive ly . Veh ic le- leve l

dynam ics from the v isua l env ironment a lso affect the vest ibu lar- leve l dynam ics . The vest ibu lar- leve l dynam ics of the dr ivers are inf luenced on ly by the veh ic le- leve l dynam ics for the stat ic p latform case whereas the dr ivers vest ibu lar- leve l dynam ics are affected by both the veh ic le- leve l dynam ics and the inert ia l- leve l dynam ics (hexapod mot ion p latform) for the dynam ic p latform cond it ion .

Exper imen ta l se t -up

In th is study , the percept ion of mot ion s ickness was eva luated ob ject ive ly and sub ject ive ly in the SAAM dur ing operat ing the p latform stat ica l ly and dynam i- ca l ly (see F igure 1 for the mu lt i-sensory integrat ion and mot ion s ickness mechan isms) . Tab le 1 g ives the cap- ab i l it ies of the SAAM used for the exper iments for the dynam ic operat ions .

F igure 2 . Procedure to corre late the i l lness rat ing w ith the percept ion o f the dr ivers .

The SAAM s imu lator was des igned and deve loped by Arts et Met iers Par is Tech and Renau lt . The ob jec- t ive rat ing refers to the methods performed through measurements of those parameters wh ich do not con- ta in persona l assessments . In the case of the sub ject ive assessment of s imu lator s ickness , we proposed a method ca l led percept ion due to psychophys ics.

For the ob ject ive eva luat ion , the perce ived dynam ics by the sub jects dur ing the rea l-t ime sess ions were observed regard ing the vest ibu lar leve l . The advantage of the data acqu is it ion modu le from the SCANeRstud io vers ion 1 .1 was that the data re lated to commands (steer ing-whee l ang le , acce lerator , brake peda l force , etc .) , dynam ics (veh ic le dynam ics data) , eng ine and fre- quency ana lys is (data co l lect ion in rea l t ime v ia SCANeRstud io) can be saved , whereas the record of the sensed dynam ics on the vest ibu lar leve l of the dr iver dur ing the on l ine tests was obta ined by us ing a three- d imens iona l acce lerat ion sensor (XSens) attached to a headphone a l igned to the dr ivers r ight ear (F igure 3) . Pro toco l

F igure 4 i l lustrates the b irds-eye v iew of the tra jectory wh ich was dr iven dur ing the dr iv ing s imu lator

exper iment . In F igure 4 , the start ing locat ion of the veh ic le tested on the dr iv ing s imu lator exper iments as we l l as the d istances between the py lons are dep icted . W andLare the w idth and the length respect ive ly of the veh ic le . HereW = 1 m and L= 1 .5 m .

The sub jects were asked to dr ive a NATO ch icane manoeuvre (see F igure 4) .

^31,32

D ifferent cond it ions were eva luated w ith or w ithout the mot ion p latform . After each attempt , the sub jects were asked to f i l l in the ava i l- ab le quest ionna ire for the sub ject ive rat ing of the per- cept ion due to psychophys ics . A lso , dur ing each tr ia l , the data were recorded w ith a samp l ing per iod of 0 .05 s . The sub jects drove the same scenar io tw ice to become fam i l iar ized w ith it before the eva luat ion phase . The tests were accomp l ished in d ifferent cond it ions w ith the NATO ch icane scenar io at a constant dr iv ing ve loc ity V

x

= 60 km/h for a durat ion of 37–50 s .

Tab le 2 shows the parameters of the mot ion cue ing a lgor ithm used in the dynam ic p latform s imu lator con- d it ion w ith in these exper iments .

The second-order low-pass (LP) cut-off frequency and the second-order LP damp ing factor are the onset cues for the t i lt ing (p itch and ro l l) as the susta ined part of the movements . The f irst-order LP t ime constant i l lustrates the t ime de lays for the susta ined acce lerat ions of the mot ion dr ive a lgor ithm . The susta ined part of the mot ion is assoc iated w ith the cue conf l icts between the v isua l (cab in- or veh ic le- leve l) env ironment and the inert ia l (mot ion-p latform- leve l) env ironment for low- frequency mot ion (t i lt ing , i .e . p itch and ro l l) .

The second-order h igh-pass (HP) cut-off frequency and the second-order HP damp ing factor are the onset cues for the yaw mot ion and the long itud ina l and latera l trans lat iona l acce lerat ions . The f irst-order HP t ime con- stant g ives the t ime de lays for the onset cues . The onset cues of the mot ion are re lated to the cue conf l icts between the v isua l (cab in- or veh ic le- leve l) env ironment and the inert ia l (mot ion-p latform- leve l) env ironment for h igh- frequency mot ion (trans lat iona l acce lerat ions and yaw) . Sub jec ts

20 sub jects (18 ma les and 2 fema les) w ith an age d istr i- but ion of 28 .66 5 .97 years o ld and a dr iv ing l icence exper ience d istr ibut ion of 9 .46 6 .11 years part ic ipated in the exper iments . The l im itat ion of the current inves- t igat ion is that no gender effect was taken into account . Tab le 1 . L im its for each degree o f freedom o f the SAAM .

^6,8

Degree o f freedom D isp lacement Ve loc ity Acce lerat ion P itch 622 630 deg /s 6500 deg /s

²

Ro l l 621 630 deg /s 6500 deg /s

²

Yaw 622 640 deg /s 6400 deg /s

²

Heave 60 .18 m 60 .3 m /s 60 .5g Surge 60 .25 m 60 .5 m /s 60 .6g Sway 60 .25 m 60 .5 m /s 60 .6g

F igure 3 . Vest ibu lar - leve l data acqu is it ion .

F igure 4 . NATO ch icane r ight turn used on the dr iv ing

s imu lator sess ion .

³¹

The sub jects were researchers , eng ineers and students from the laboratory . The quest ionna ires were adm in is- tered to the sub jects in the French language . They had a fam i l iar izat ion dr ive on the dr iv ing s imu lator before each sess ion to check for any pred ispos it ion regard ing mot ion s ickness , to avo id m iseva luat ion and to he lp them to assess the procedure as ob ject ive ly as poss ib le . Ob jec t ive measures

Ob ject ive eva luat ion refers to an assessment method for dr iv ing s imu lator app l icat ions in wh ich var ious data are measured . In th is invest igat ion , accord ing to the pro- posed approach , the measurements were rea l ized for vest ibu lar cues (measurements of the head movements of the sub jects) to obta in the i l lness rat ing wh ich is der ived from the MSDV for chang ing the mot ion base act ivat ion of the SAAM . Vest ibu lar- leve l data acqu is i- t ion (head movements of the part ic ipants) is performed v ia a sensor wh ich measures the ro l l , the p itch , the yaw ang les and the acce lerat ions a long thex , yandzaxes . The data are ca l ibrated w ith respect to the three- d imens iona l quatern ion or ientat ion. The samp l ing per iod for the data reg istrat ion dur ing the sensor mea- surements is 0 .05 s . For the ca l ibrated data acqu is it ion , the a l ignment reset was chosen , wh ich s imp ly comb ines the ob ject reset and the head ing reset at a s ing le instant in t ime . Th is has the advantage that a l l coord inate sys- tems can be a l igned w ith a s ing le act ion . In th is study , we ut i l ized the long itud ina l , latera l and vert ica l acce l- erat ions from the head (vest ibu lar) leve ls of the part ic i- pants (see F igure 3) . The r .m .s . acce lerat ions were computed from equat ions (2) , (3) and (4) . Then , the component i l lness rat ing metr ics were obta ined for the long itud ina l d irect ionx, the latera l d irect ionyand the vert ica l d irect ionz. For examp le , equat ion (6) g ives the i l lness rat ing of the long itud ina l acce lerat ions measured by the sub jects heads (vest ibu lar , F igure 3) leve ls as a component i l lness rat ing for long itud ina l dynam ics . We expanded the component i l lness rat ing ana lys is for the latera l and vert ica l acce lerat ions by subst itut ing the re lated r .m .s . acce lerat ions .

We def ined the compound i l lness rat ing as IR

tot

, and the compound r .m .s . acce lerat ion was obta ined by sub- st itut ing the measured vest ibu lar leve l long itud ina l (x) , latera l ( y) and vert ica l (z) acce lerat ions in equat ion (1) .

In order to determ ine the compound i l lness rat ing IR

tot

wh ich is g iven in equat ion (7) , equat ion (1) was subst ituted in equat ion (5) , and f ina l ly equat ion (5) was rep laced in equat ion (7) .

Sub jec t ive measures

The quest ions regard ing the percept ion due to psycho- phys ics that were asked just after the comp let ion of each tr ia l are the fo l low ing (Tab le 3) .

⁸

In order to assess the dr iv ing s imu lator tests sub jec- t ive ly on beha lf of the percept ion due to psychophys ics , a s imu lator s ickness and psychophys ica l percept ion quest ionna ire wh ich cons ists of 12 quest ions (graded from 1 : very l itt le! 10 : very strong) was g iven to each part ic ipant to obta in a sub ject ive measure of the dr iv- ing s imu lat ion . We named it percept ion due to psycho- phys ics , wh ich is rather d ifferent from the ex ist ing s imu lator s ickness quest ionna ire

⁹

and the mot ion s ick- ness h istory quest ionna ire

²⁴

, not on ly because of the d isor ientat ion-re lated s ickness cr iter ia (nausea and d iz- z iness) but a lso because we a imed to assess the sub jec- t ive impress ions of the part ic ipants in terms of perce iv ing the st imu l i and the ir react ions to them (psy- chophys ics) such as fear , menta l pressure , fat igue and anx iety (Tab le 3) .

In the s imu lator s ickness quest ionna ire , quest ions about nausea , d isor ientat ion and ocu lomotor fat igue were used , and the ir eva luat ion rat ing sca les were d iv ided into four .

⁹

In our quest ionna ire set , we used a grad ing sca le between 1 and 10 .

Data ana lys is

The PCA method is a stat ist ica l techn ique used for d imens iona l ity reduct ion and representat ion of mu lt i- var iate data sets represented by anN3pmatr ixXw ith N observat ions andpvar iab les .

³³

Th is method trans- forms a mu lt i-var iate data set of intercorre lated var i- ab les into a data set of new uncorre lated l inear comb inat ions of the or ig ina l var iab les . Before the ana l- ys is , the co lumns ofXare centred if var iab les have very heterogeneous var iances . In add it ion to centr ing , when the var iab les are measured w ith d ifferent un its , it is cus- tomary to standard ize each var iab le to the un it norm . Th is is obta ined by d iv id ing each var iab le by its norm . Tab le 2 . C lass ica l mot ion cue ing a lgor ithm parameters .

^6,8

Parameter (un its) Long itud ina l Latera l Ro l l P itch Yaw Second-order LP cut-o f f frequency (Hz) 0 .3 0 .7

Second-order LP damp ing factor 0 .3 0 .7

F irst-order LP t ime constant (s) 0 .1 0 .1 0 .1

Second-order HP cut-o f f frequency (Hz) 0 .5 0 .5 2

Second-order HP damp ing factor 1 1 1

F irst-order HP t ime constant (s) 2 2 2

LP : low-pass ; HP : h igh-pass .

In th is case , the ana lys is is referred to as a corre lat ion PCA .

For a g ivenp-d imens iona l data set , the PCA method app l ies an orthogona l transformat ion on anm-d imen- s iona l subspace (1\ m\ p) spanned by orthogona l axes ca l led pr inc ipa l axes a long wh ich the var iance of the data is max im ized . The pr inc ipa l axes can be g iven by the lead ing e igenvectors assoc iated w ith themlarg- est e igenva lues of the corre lat ion matr ix of the data set . The va lues of the data a long these axes are ca l led pr in- c ipa l components . In fact , a sma l l number of pr inc ipa l components can descr ibe most of the var iance in the or ig ina l data set and are therefore used as features to rep lace the or ig ina l data representat ion . The use of these pr inc ipa l components as features can reduce the d imens ions of the data representat ion w ithout much loss of the var iance in the or ig ina l data set .

As the var iab les are standard ized to the un it norm , they can be represented by vectors ly ing on anm- d imens iona l un it sphere , wh ich we can pro ject on the f irst two pr inc ipa l axes . These pro ject ions are vectors whose coord inates are corre lat ions between var iab les w ith the f irst two pr inc ipa l components .

Therefore the var iab les w i l l be pos it ioned ins ide a c irc le ca l led the c irc le of corre lat ions, wh ich is usefu l to eva luate the aff in it ies and antagon isms between var i- ab les . The c loser a var iab le is to the c irc le of corre la- t ions , the better we can reconstruct th is var iab le from the f irst two components . In th is case , if the var iab les are c lose to each other , they are s ign if icant ly pos it ive ly corre lated (rc lose to 1) ; if they are orthogona l , they are not corre lated (rc lose to 0) ; if they are on the oppos ite

s ide of the centre , then they are s ign if icant ly negat ive ly corre lated (rc lose to21) . When a var iab le is c loser to the centre of the c irc le , th is means that it is less impor- tant for the f irst two components .

In th is study , the percept ion of mot ion s ickness (psy- chophys ica l sub ject ive assessments) was compared by us ing the PCA method for the stat ic s imu lator cond i- t ion and the dynam ic s imu lator cond it ion . The ques- t ionna ires on ly about psychophys ica l percept ion for the two cond it ions (the stat ic case and the dynam ic p lat- form case) were eva luated by us ing a two-ta i led Mann–

Wh itney Utest and PCA . The inputs to the PCA meth- ods were the perceptua l quest ions g iven in Tab le 3 .

The same pattern or symptomato logy factor struc- tures were used in both cond it ions of the s imu lator p lat- form operat ions dur ing the exper iments .

Afterwards , perceptua l convergence to the phys ica l s ituat ion was checked by us ing the Pearson corre lat ion between the ob ject ive data and the sub ject ive measure data . In other words , the Pearson corre lat ion method was emp loyed to eva luate the effects of the ob ject ive measures (IR

x

,IR

y

,IR

z

and IR

tot

va lues computed from the vest ibu lar- leve l long itud ina l , latera l and vert i- ca l acce lerat ion measurements g iven in F igure 3) on the se lf-reports (the sub ject ive measures v ia the quest ion- na ire in Tab le 3) .

Resu lts and d iscuss ion

Tab le 4 i l lustrates the se lf-report resu lts of us ing two cond it ions of a dr iv ing s imu lator , name ly w ith a stat ic p latform or w ith a dynam ic p latform . Stat ist ica l ly s ig- n if icant mean d ifferences ex ist between the treatments for nausea , d izz iness , eye stra in and fat igue (t iredness) in the stat ic cond it ion and the dynam ic cond it ion .

The two-ta i led Mann–Wh itneyUtest was chosen to compare the s ign if icance in the d ifferences between the ind iv idua l quest ions rather than the PCA .

As seen from the two-ta i led Mann–Wh itneyUtest , exper ienc ing nausea (p =0 .019\ 0 .05) and d izz iness (p =0 .018\ 0 .05) were decreased s ign if icant ly from the stat ic case to the dynam ic case .

A lso , as found from the two-ta i led Mann–Wh itney Utest , eye stra in (p =0 .047\ 0 .05) and t iredness (p = 0 .047\ 0 .05) were reduced s ign if icant ly from the stat ic case to the dynam ic case .

In our context , the stat ist ica l ana lys is was accom- p l ished by us ing the language R for both ob ject ive assessments and sub ject ive assessments . On the r ight- hand s ide of F igure 5 , we see that the f irst two e igenva- lues (b lack bars) of the corre lat ion matr ix correspond to 63 .42% and 13 .38% respect ive ly of the var iance (see Tab le 5) . It appears that most of the informat ion car- r ied by the var iab les can be extracted from on ly the f irst two pr inc ipa l axes ow ing to the PCA carr ied out by R stat ist ica l ana lys is software . On the left-hand s ide of F igure 5 , the sub ject ive eva luat ion data were pro jected on the corre lat ion c irc le on the f irst two pr inc ipa l axes Tab le 3 . Quest ionna ire on the percept ion due to

psychophys ics .

Quest ion number Quest ion (quest ion rat ing) Q1 Were you prone to vom it?

(1 : very l itt le! 10 : very strong) Q2 D id you fee l nauseous?

(1 : very l itt le! 10 : very strong) Q3 D id you have a co ld sweat?

(1 : very l itt le! 10 : very strong) Q4 D id you fee l d izzy?

(1 : very l itt le! 10 : very strong) Q5 D id you fee l eye stra in?

(1 : very l itt le! 10 : very strong) Q6 D id you have troub le w ith your eyes?

(1 : very l itt le! 10 : very strong) Q7 D id you have a headache?

(1 : very l itt le! 10 : very strong) Q8 D id you fee l menta l pressure?

(1 : very l itt le! 10 : very strong) Q9 D id you fee l fear?

(1 : very l itt le! 10 : very strong)

Q10 Were you bored?

(1 : very l itt le! 10 : very strong) Q11 Were you t ired?

(1 : very l itt le! 10 : very strong) Q12 D id you fee l anx iety (uneas iness)?

(1 : very l itt le! 10 : very strong)

assoc iated w ith pr inc ipa l component 1 (hor izonta l ax is) ( labe l led Comp1 in F igure 5) and pr inc ipa l component 2 (vert ica l ax is) ( labe l led Comp2 in F igure 5) respec- t ive ly . Most of the or ig ina l pro jected var iab les had a good representat ion on the pr inc ipa l p lane generated by the f irst two pr inc ipa l axes . Moreover , accord ing to the pr inc ipa l ax is assoc iated w ith pr inc ipa l component 1 , it was demonstrated that the sub ject ive data were pos it ive ly corre lated w ith pr inc ipa l component 1 for the dynam ic p latform , whereas these data were negat ive ly corre lated w ith pr inc ipa l component 1 in the case of the stat ic p latform . Th is means that the s imu lator s ickness percept ion revea led from the quest ionna ire on the per- cept ion due to psychophys ics (Tab le 3) showed conf l ict between the stat ic p latform and the dynam ic p latform . Pr inc ipa l component 1 refers to the d isor ientat ion- re lated s ickness (nausea and d izz iness ; see Tab le 3) , whereas pr inc ipa l component 2 is ded icated to ocu lomo- tor fat igue and i l lness (eye stra in and t ired ; see Tab le 3) .

Ob ject ive data were computed from the i l lness rat- ings (equat ion (7)) on the vest ibu lar leve l dur ing the execut ion of the dr iv ing exper iments on the SAAM , and sub ject ive data were extracted from responses to a quest ionna ire on both the stat ic p latform cond it ion and the dynam ic p latform cond it ion . The approach

that we used to prove the corre lat ion between ob ject ive data and sub ject ive data in th is research is the Pearson corre lat ion .

The subscr iptsSTAandDYNs ign ify the stat ic con- d it ion and the dynam ic cond it ion respect ive ly for the answers g iven by the sub jects to the percept ion ques- t ionna ire in Tab les 6 and 7 .

So as to measure the dr ivers percept ion , a percep- t ion due to psychophys ics quest ionna ire was introduced to obta in an op in ion on the sub ject ive eva luat ion of s imu lator s ickness . Reason and Brand

³⁴

suggested that a s ign if icant reduct ion in mot ion s ickness occurs when an ind iv idua l adopts a postura l pos it ion . The postura l pos it ion refers to the sp ina l stab i l ity when the part ic i- pants are stand ing up just after the exper imenta l ses- s ions or at the change between pre- and post- exper imenta l sess ions . R icc io and Stoffregen

³⁵

intro- duced the postura l instab i l ity theory a lso to def ine the re lat ions between percept ion and the contro l of act ion (head movements in our study) . Th is approach cons id- ers the behav iour of the ind iv idua l as fundamenta l in mot ion s ickness aet io logy . The postura l instab i l ity the- ory of mot ion s ickness presumes that mot ion s ickness resu lts from and can be est imated by instab i l it ies in contro l l ing the sp ine . Th is was attr ibuted to constra ints in mot ion of the head . Kennedy et a l .

¹⁴

and Reason and Brand

³⁴

descr ibed the re lat ions between the head mot ions and mot ion s ickness through the Cor io l is mechan isms (w ith actua l inert ia l cues , i .e . the mot ion p latform) and pseudo-Cor io l is (through v isua l cues , i .e . the stat ic operat ion of the s imu lator) st imu lat ion . Cor io l is st imu lat ion occurs when the head is t i lted out of the ax is of rotat ion dur ing actua l body rotat ion .

^36–40

Pseudo-Cor io l is st imu lat ion occurs when the head is t i lted as perce ived se lf-rotat ion that is induced by v isua l st imu l i .

³⁸

Accord ing to Kennedy et a l . ,

¹⁴

the sub jects head movements in a mov ing-base s imu lator were s im i lar to those in the actua l veh ic le ,

^36–40

whereas the head move- ments in f ixed-base s imu lators were often in conf l ict Tab le 4 . Se l f-report resu lts .

Se l f-report (quest ion number) Va lue for the se l f-report for the fo l low ing p lat forms pva lue from a two-ta i led Mann–Wh itneyUtest Stat ic p lat form

(mean6standard dev iat ion) Dynam ic p lat form

(mean6standard dev iat ion)

Propens ity to vom it (Q1) 2 .50061 .225 1 .42960 .787 p=0 .067. 0 .05

Nausea (Q2) 3 .00061 .265 1 .57160 .535 p=0 .019\ 0 .05

Co ld sweat (Q3) 2 .00060 .632 1 .42960 .787 p=0 .120. 0 .05 D izz iness (Q4) 2 .66761 .211 1 .42960 .535 p=0 .018\ 0 .05 Eye stra in (Q5) 2 .16760 .983 1 .28660 .488 p=0 .047\ 0 .05 T roub le w ith eyes (Q6) 2 .00060 .894 1 .42960 .535 p=0 .214. 0 .05 Headache (Q7) 2 .00060 .894 1 .28660 .488 p=0 .114. 0 .05 Menta l pressure (Q8) 1 .66760 .816 1 .42960 .535 p=0 .630. 0 .05

Fear (Q9) 1 .50060 .548 1 .42960 .535 p=0 .805. 0 .05

Boredom (Q10) 2 .33361 .862 1 .28660 .488 p=0 .148. 0 .05 T iredness (Q11) 2 .00060 .632 1 .28660 .488 p=0 .047\ 0 .05 Anx iety (uneas iness) (Q12) 2 .16760 .753 1 .42960 .535 p=0 .074. 0 .05

Tab le 5 . Inert ias o f the e igenva lues and the percentages in the var iance .

Pr in ic ipa l

component Inert ia o f the

e igenva lues Percentage in the var iance(%) Component 1 15 .22 63 .42

Component 2 3 .21 13 .38 Component 3 2 .54 10 .58

Component 4 1 .29 5 .38

Component 5 0 .9 3 .75

Component 6 0 .48 2

Component 7 0 .23 0 .96

Component 8 0 .096 0 .53

w ith the inert ia l st imu lus , wh ich increased the d iscre- pancy of the s imu lat ion .

³⁸

Those f ind ings are para l le l to ours (Tab le 6 and Tab le 7) .

Th is work enab led us to answer some of the research quest ions asked .

Q3 (hav ing a co ld sweat) was s ign if icant ly corre- lated w ith the IR

y

for the dynam ic p latform cond it ion (p \ 0 .05) , whereas it was s ign if icant ly corre lated w ith IR

x

and IR

tot

in the stat ic cond it ion ( p\ 0 .05) . Q4 (fee l ing d izzy) was s ign if icant ly corre lated w ith IR

z

for the dynam ic p latform , whereas it was s ign if i- cant ly corre lated w ith IR

x

and IR

tot

for the stat ic case (p\ 0 .05) . Q5 (fee l ing eye stra in) was s ign if icant ly corre lated w ith IR

z

and IR

tot

at the dynam ic p lat- form , wh i le it was s ign if icant ly corre lated w ith IR

x

and IR

tot

at the stat ic p latform (p \ 0 .05) . Q7

(hav ing a headache) was s ign if icant ly corre lated w ith IR

z

and IR

tot

for the dynam ic s ituat ion (p\ 0 .05) . The other component and compound i l lness rat ings had no s ign if icant corre lat ion w ith Q7 (p. 0 .05) for the stat ic case . Q10 (fee l ing bored) was s ign if icant ly corre lated w ith the IR

z

and IR

tot

for the dynam ic s it- uat ion , wh i le it was s ign if icant ly corre lated w ith IR

x

and IR

tot

for the stat ic case (p\ 0 .05) . Q11 (fee l ing t ired) was s ign if icant lycorre lated w ith the IR

x

and IR

tot

for the stat ic s ituat ion and it was s ign if icant ly corre lated w ith IR

z

and IR

tot

for the dynam ic case (p\ 0 .05) .

The two-ta i led Mann–Wh itneyUtest ind icated that nausea (Q2 :U(20) ,p =0 .019\ 0 .05) and d izz iness (Q4 :U(20) ,p =0 .018\ 0 .05) abated s ign if icant ly from the stat ic cond it ion to the dynam ic cond it ion .

F igure 5 . Corre lat ion c irc le and e igenva lues o f the PCA for the stat ic cond it ion and the dynam ic cond it ion . Comp1 : pr inc ipa l component 1 (hor izonta l ax is) ; Comp2 : pr inc ipa l component 2 (vert ica l ax is) .

Tab le 6 . Corre lat ion coe f f ic ients between the ob ject ive data and the sub ject ive data for the stat ic p lat form .

Var iab le IR

x

IR

y

IR

z

IR

tot

q

1STA

0 .625 0 .108 –0 .302 0 .538 q

2STA

0 .453 –0 .145 –0 .211 0 .385 q

3STA

0 .836 0 .213 0 .367 0 .855 q

4STA

0 .922 0 .006 0 .602 0 .970 q

5STA

0 .948 0 .175 0 .363 0 .958 q

6STA

0 .661 0 .054 –0 .275 0 .575 q

7STA

0 .135 –0 .012 –0 .561 0 .029 q

8STA

–0 .308 0 .174 –0 .617 –0 .391 q

9STA

–0 .335 0 .300 –0 .681 –0 .424 q

10STA

0 .984 0 .187 0 .134 0 .953 q

11STA

0 .836 0 .213 0 .367 0 .855 q

12STA

0 .613 0 .149 0 .134 0 .603 The va lues in bo ld are d i f ferent from 0 w ith a s ign i f icance leve la= 0 .05 .

Tab le 7 . Corre lat ion coe f f ic ients between the ob ject ive data and the sub ject ive data for the dynam ic p lat form .

Var iab le IR

x

IR

y

IR

z

IR

tot

q

1DYN

–0 .291 –0 .262 0 .112 –0 .292

q

2DYN

0 .496 –0 .250 0 .694 0 .498

q

3DYN

–0 .283 0 .902 –0 .114 –0 .221

q

4DYN

0 .409 –0 .259 0 .765 0 .416

q

5DYN

0 .730 0 .118 0 .906 0 .758

q

6DYN

0 .415 0 .597 0 .599 0 .468

q

7DYN

0 .730 0 .118 0 .906 0 .758

q

8DYN

0 .415 0 .597 0 .599 0 .468

q

9DYN

0 .415 0 .597 0 .599 0 .468

q

10DYN

0 .730 0 .118 0 .906 0 .758

q

11DYN

0 .730 0 .118 0 .906 0 .758

q

12DYN

0 .415 0 .597 0 .599 0 .468

The va lues in bo ld are d i f ferent from 0 w ith a s ign i f icance leve la= 0 .05 .

S im i lar ly the two-ta i led Mann–Wh itneyU test sug- gests that eye stra in (Q5 :U(20) ,p =0 .047\ 0 .05) and fat igue (Q11 :U(20) ,p =0 .047\ 0 .05) were reduced s ign if icant ly by us ing the dynam ic p latform rather than the stat ic p latform .

Conc lus ion

The sub ject ive eva luat ion compar ison showed that the answers to the quest ionna ires for the stat ic case were negat ive ly corre lated w ith the answers g iven to the same quest ionna ires for the dynam ic p latform for the h ighest pr inc ipa l component (PCA , pr inc ipa l component 1 , i .e . nausea + d izz iness) (the percentage in the var iance is 63 .42%) . Th is a lready ind icates a d ifference for the con- tr ibut ion of the dynam ic p latform . In order to eva luate the percept ion f ide l ity , the Pearson corre lat ion between the sub ject ive data and the ob ject ive data was checked .

Hav ing a co ld sweat (fear ind icator) , fee l ing d izzy (d isor ientat ion) , fee l ing eye stra in (v isua l s ickness) , fee l- ing bored (psychophys ica l s ituat ion) and fee l ing t ired (psychophys ica l s ituat ion) were s ign if icant ly pos it ive corre lated w ith IR

x

and IR

tot

for the stat ic p latform . These resu lts for the stat ic case showed that the ma in factor wh ich induced the d iscomfort was the long itud i- na l head (vest ibu lar- leve l) dynam ics . Wh i le the latera l and the vert ica l head dynam ics were not s ign if icant in the percept ion of the mot ion s ickness , the tota l com- pound head dynam ics were a lso s ign if icant ly corre lated w ith the fear ind icator , the d isor ientat ion fee l ing , the v isua l s ickness and the psychophys ica l s ituat ion .

Hav ing a co ld sweat was s ign if icant ly corre lated w ith IR

y

for the dynam ic case . Th is showed that the se lf-report f ind ings are suggest ive . Moreover , fee l ing d izzy , fee l ing eye stra in , fee l ing bored and fee l ing t ired were pos it ive ly corre lated w ith IR

z

and IR

tot

. These f ind ings for the dynam ic cond it ion dep icted that the ma in factor wh ich provoked the d isor ienta- t ion , the v isua l s ickness and the psychophys ica l s itua- t ion was vert ica l head dynam ics of the sub jects . Because the long itud ina l and the latera l head dynam ics d id not have s ign if icant ro les in the percep- t ion of the mot ion s ickness , the tota l compound ves- t ibu lar leve l dynam ics werea lso corre lated w ith the d isor ientat ion , the v isua l s ickness and the psychophy- s ica l s ituat ion .

Acknow ledgements

Arts et Met iers Par isTech bu i lt the SAAM dr iv ing s imu lator in co l laborat ion w ith Renau lt and Grand Cha lon . Th is research study was rea l ized in the frame- work of the geDr iver pro ject .

Fund ing

Th is research rece ived no spec if ic grant from any fund- ing agency in the pub l ic , commerc ia l or not-for-prof it sectors .

Dec larat ion of conf l ict of interest

The authors dec lare that there is no conf l ict of interest . References

1 . Kemeny A and Panera i F . Eva luat ing percept ion in dr iv- ing s imu lat ion exper iments .Trends Cogn it ive Sc i 2003 ; 7(1) : 31–37 .

2 . Ko las insk i EM . S imu lator s ickness in v irtua l env iron- ments . Report ARI-TR-1027 , Army Research Inst itute for the Behav ioura l and Soc ia l Sc iences , A lexandr ia , V ir- g in ia , USA , 1995 .

3 . Berger DR , Schu lte-Pe lkum J and Bu ¨ lthoff HH . S imu lat- ing be l ievab le forward acce lerat ions on a Stewart mot ion p latform .ACM Trans App l Percept ion 2010 ; 7(1) : 5 :1–

5 :27 .

4 . Curry R , Artz B , Cathey L , et a l . Kennedy SSQ resu lts : f ixed-vs mot ion-based Ford s imu lators . In :Dr iv ing s imu- lat ion conference , INRETS : Par is , France , 11–13 Septem- ber 2002 , pp . 289–300 .

5 . Watson G . A synthes is of s imu lator s ickness stud ies con- ducted in a h igh-f ide l ity dr iv ing s imu lator . In :Dr iv ing s imu lat ion conference, INRETS : Par is , France , 6–8 Sep- tember 2000 , pp . 69–78 .

6 . Aykent B , Mer ienne F , Pa i l lot D and Kemeny A . Inf lu- ence of inert ia l st imu lus on v isuo-vest ibu lar cues conf l ict for latera l dynam ics at dr iv ing s imu lators .J Ergonom ics 2013 ; 3 : 1–7 .

7 . S ieg ler I , Reymond G , Kemeny A , et a l . Sensor imotor integrat ion in a dr iv ing s imu lator : contr ibut ions of mot ion cue ing in e lementary dr iv ing tasks . In : Dr iv ing s imu lat ion conference, INRETS : N ice , France , 5–7 Sep- tember 2001 , pp . 21–32 .

8 . Aykent B , Pa i l lot D , Mer ienne F , et a l . Study of the inf luence of d ifferent washout a lgor ithms on s imu lator s ickness for a dr iv ing s imu lat ion task . In :ASME wor ld conference on innovat ive v irtua l rea l ity, M i lan , Ita ly , 27–

29 June 2011 , paper WINVR2011-5545 , pp . 331–341 . New York : ASME .

9 . Kennedy RS , Lane NE , Berbaum KS , et a l . S imu lator s ickness quest ionna ire : an enhanced method for quant i- fy ing s imu lator s ickness .Int J Av iat Psycho l1993 : 3(3) : 203–220 .

10 . Dur lach NI and Mavor AS .V irtua l rea l ity:sc ient if ic and techno log ica l cha l lenges. Wash ington , DC : Nat iona l Aca- dem ies Press , 1995 .

11 . May J and Badcock D . V is ion and v irtua l env ironments . In : Stanney KM (ed)Handbook of v irtua l env ironments:

des ign , imp lementat ion , and app l icat ions. Mahwah , New Jersey : Lawrence Er lbaum Assoc iates , 2002 , pp . 29–64 . 12 . McCau ley M (ed) .Research issues in s imu lator s ickness ,

Proceed ings of a workshop. Wash ington , DC : Nat iona l Academ ies Press , 1984 .

13 . McCau ley M and Sharkey T . Cybers ickness-percept ion of se lf-mot ion in v irtua l env ironments .Presence: Te leo- perators V irtua l Env iron1992 ; 1(3) : 311–318 .

14 . Kennedy RS , Berbaum KS , L i l ientha l MG , et a l . Gu ide- l ines for a l lev iat ion of s imu lator s ickness symptomato l- ogy . Report NAVTRASYSCEN-TR-87-007 , Nava l Tra in ing Systems Center Or lando , F lor ida , USA , 1987 . 15 . Kennedy R , Berbaum K , A l lgood G , et a l . Et io log ica l

s ign if icance of equ ipment features and p i lot h istory in

s imu lator s ickness . In :Mot ion cues in f l ight s imu lat ion

and s imu lator induced s ickness,AGARD Conference Pro- ceed ings 443, Brusse ls , Be lg ium , 29 September –1 October 1988 , 22 pp . Neu i l ly sur Se ine : AGARD .

16 . Hett inger LJ . I l lusory se lf-mot ion in v irtua l env iron- ments . In : Stanney KM (ed)Handbook of v irtua l env iron- ments: des ign , imp lementat ion , and app l icat ions. Mahwah , New Jersey : Lawrence Er lbaum Assoc iates , 2002 , pp . 471–492 .

17 . Hett inger LJ , Berbaum KS , Kennedy RS , et a l . Vect ion and s imu lator s ickness .M i l it Psycho l 1990 ; 2(3) : 171–181 . 18 . Hett inger LJ and R icco G . V irtua l ly induced mot ion s ick- ness in v irtua l env ironments .Presence: Te leoperators V ir- tua l Env iron1992 ; 1(3) : 306–310 .

19 . Yard ley L . Mot ion s ickness and percept ion : a reappra isa l of the sensory conf l ict approach .Br J Psycho l 1992 ; 83(4) : 449–471 .

20 . Lawson B , Graeber D , Mead A , et a l . S igns and symp- toms of human syndromes assoc iated w ith synthet ic exper iences . In : Stanney KM (ed)Handbook of v irtua l env ironments: des ign , imp lementat ion , and app l icat ions.

Mahwah , New Jersey : Lawrence Er lbaum Assoc iates , 2002 , pp .589–618 .

21 . D iz io P and Lackner J . Spat ia l or ientat ion , adaptat ion , and mot ion s ickness in rea l and v irtua l env ironments . Presence1992 ; 1(3) : 319–328 .

22 . Gr iff in MJ .Handbook of human v ibrat ion. Amsterdam : E lsev ier , 1996 .

23 . ISO 2631-1 : 1997 .Mechan ica l v ibrat ion and shock: eva- luat ion of human exposure to who le-body v ibrat ion . Part 1 , Genera l requ irements. Geneva : Internat iona l Organ iza- t ion for Standard izat ion , 1997 .

24 . Kennedy RS , Fow lkes JE , Berbaum KS , et a l . Use of a mot ion s ickness h istory quest ionna ire for pred ict ion of s imu lator s ickness .Av iat , Space , Env iron Med 1992 ; 63(7) : 588–593 .

25 . K im MS , Moon YG , K im GD , et a l . Part ia l range sca l- ing method based washout a lgor ithm for a veh ic le dr iv ing s imu lator and its eva luat ion .Int J Automot Techno l2010 ; 11(2) : 269–275 .

26 . Kennedy RS , Drex ler J and Kennedy RC . Research in v isua l ly induced mot ion s ickness .App l Ergonom ics2010 ; 41 : 494–503 .

27 . Drex ler JM .Ident if icat ion of system des ign features that affect s ickness in v irtua l env ironments. Doctora l D isserta- t ion , Un ivers ity of Centra l F lor ida , Or lando , For ida , USA , 2006

28 . G ianaros PJ , Muth ER , Mordkoff T , et a l . A quest ion- na ire for the assessment of the mu lt ip le d imens ions of mot ion s ickness . Av iat Space Env iron Med 2001 ; 72(2) 115–119 .

29 . Abercromby A . , Amonette W , Layne C , et a l . V ibrat ion exposure and b iodynam ic responses dur ing who le-body v ibrat ion tra in ing .Med Sc i Sports Exerc ise 2007 ; 39(10) : 1794–1800 .

30 . Gr iff in MJ . M in imum hea lth and safety requ irements for workers exposed to hand-transm itted v ibrat ion and who le-body v ibrat ion in the European Un ion ; a rev iew . Occup Env iron Med2004 ; 61(5) : 387–397 .

31 . Gunter DD , By lsma WW , Edgar K , et a l . Us ing mode l- ing and s imu lat ion to eva luate stab i l ity and tract ion per- formance of a track- lay ing robot ic veh ic le . In : Trev isan i DA and S ist i F (eds) Enab l ing techno log ies for s imu lat ion sc ience IX,Proceed ings of the SPIE, Vo l 5805 , pp . 66–73 . Be l l ingham , Wash ington : SPIE , 2005 .

32 . Yang X and Gander J . Dr ivers prev iew strategy and its impact on NATO doub le lane change maneuver . SAE paper 2011-01-980 , 2011 .

33 . Husson F , Le ˆ S and Page ` sJ .Ana lyse de donne´ es avec R.

Rennes :Presses Un ivers ita ires de Rennes , 2009 .

34 . Reason J and Brand J .Mot ion s ickness. New York : Aca- dem ic Press , 1975 .

35 . R icc io G and Stoffregen T . An eco log ica l theory of mot ion s ickness and postura l instab i l ity . Eco l Psycho l 1991 ; 3(3) : 195–240 .

36 . D ichgans J and Brandt T . Optok inet ic mot ion s ickness and pseudo-Cor io l is effects induced by mov ing v isua l st i- mu l i .Acta Oto-Laryngo l1973 ; 76(1–6) : 339–348 . 37 . D iz io P and Lackner J . The effects of grav ito inert ia l force

leve l and head movements on post-rotat iona l nystagmus and i l lusory after-rotat ion .Exp l Bra in Res1988 ; 70(3) : 485–495 .

38 . D iz io P and Lackner J . Perce ived se lf-mot ion e l ic ited by postrotary head t i lts in a vary ing grav ito inert ia l force background . Attent ion , Percept ion , Psychophys 1989 ; 46(2) : 114–118 .

39 . Guedry FE . V isua l contro l of hab ituat ion to comp lex vest ibu lar st imu lat ion in man .Acta Oto-Laryngo l1964 ; 58(1–6) : 377–389 .

40 . Guedry FE and Montague EK . Quant itat ive eva luat ion

of the vest ibu lar Cor io l is react ion .Aerospace Med1961 ;

32(6) : 487–500 .