Human Pe o mance En elope as a se o in e depend-
en ac o s o eal- ime moni o ing o d i e s a e
Ricca do Ce iani1*[0009-0002-5191-7110], And ea Cas ellano2[0000-0002-3636-2228], Lo enzo Uc-
cello3[0009-0008-1080-3874], Gianpie o Mas inu3[0000-0001-5601-9059], Elisa Landini2, Claudio
Lan ie i1[0000-0003-2852-567X]
1 Depa men o Ci il, En i onmen al and Ma e ial (DICAM) Enginee ing, Uni e si y o Bolo-
gna, Bologna, I aly, 40136
2 Depa men o Physics, In o ma ics and Ma hema ics, Uni e si y o Modena and Reggio
Emilia, Modena, I aly, 41121
3 Depa men o Mechanical Enginee ing, Poli ecnico di Milano, Milan, I aly, 20156
CORRESPONDING [email p o ec ed]
Abs ac . D i ing is a complex pe cep ual-mo o skill ha combines cogni i e, psychomo-
o , and pe cep ual abili ies o ope a e a ehicle sa ely and e icien ly. I in ol es he d i e ’s
abili y o manage emo ions, such as s ess and us a ion o cogni i e s a es, like d owsiness,
dis ac ion and main ain ocus. In his con ex , he Human Pe o mance En elope (HPE) is
desc ibed as a cons uc use o moni o d i e s a us based on he combina ion o a se o in e -
dependen ac o s and ep esen ed wi h a spide web model. I s goal is o minimize human e o
by analysing beha iou and decision-making while op imizing sys ems and en i onmen s o
enhance pe o mance. The in luence o he ac i i y demand on he model needs o be consid-
e ed. Assuming ha an HPE can be cha ac e ized o a speci ic ac o , he e could be a numbe
o accep able HPEs o he same ac o , each depending on he ac i i ies pe o med. One possi-
ble app oach o add essing ac i i y-based HPEs is o isola e speci ic asks o si ua ions, assess
i s ac o s in hose con ex s, and de e mine he easibili y o de ining a speci ic HPE o each
ac i i y. Depending on he alue o each ac o , he esul ing model could e ol e om ully
accep able o no accep able. Based on an in-dep h bibliog aphic esea ch his wo k aims o
popula e he h ee main “HPE pa ame e s” (wo kload, s ess, and si ua ion awa eness). wi h
ehicula , beha iou al and physiological pa ame e s.
Keywo ds: Human Pe o mance En elope, isk mi iga ion, d i e moni o ing
2
1 In oduc ion
Nume ous oad a ic sa e y (RTS) s udies ca ied ou in di e en coun ies o e
he yea s ha e consis en ly shown ha , o he h ee main componen s o he RTS
sys em ( oad use , ehicle and en i onmen ) he human ac o is he main cause o
acciden s [1]. The schema iza ion o he h ee ac o s is shown in Figu e 1.
Fig. 1. Schema iza ion o he h ee ac o s is RTS, aken om [2]
S udies aimed a imp o ing oad sa e y a e he e o e mainly conce ned in analyz-
ing human d i e : i s pe sonali y, i s beha iou on he oad, i s a i udes, i s emo ional
ension, i s knowledge, i s expe ience and many o he ac o s. These s udies a e o a
cogni i e na u e and in simple e ms i can be said ha hey a e dedica ed o iden i y-
ing he easons and ci cums ances o ce ain beha iou s on he oad [3]. The ad en o
heigh ened oad sa e y equi emen s and he decline in he inancial bu den associa ed
wi h da a acquisi ion equipmen has ende ed he iden i ica ion o d i ing beha iou a
subjec o conside able schola ly in e es in ecen yea s [4]. This echnology has a
b oad spec um o applica ions, including d i e s a e moni o ing and d i ing s yle
analysis [5,6]. I is a commonly held assump ion ha single- ac o e alua ion is an
inadequa e means o assessing d i e beha iou [7,8,9]; acco ding o his assump ion,
he aim o his wo k is o p opose a se o pa ame e s o be applied o a se o in e de-
penden ac o s o moni o he s a e o he d i e , based on exis ing li e a u e. In his
pa icula case, he ac o s o be popula ed ela e o a model called Human Pe o -
mance En elope (HPE). O iginally de eloped in a ionics, as pa o he Fu u e Sky
Sa e y p ojec , i has been desc ibed as a cons uc combining a se o nine in e de-
penden ac o s, called “HPE ac o s” (wo kload (WL), s ess, a igue, si ua ion
awa eness (SA), a en ion, igilance, eamwo k, communica ion and us ) and g aph-
ically ep esen ed wi h a spide web model. As shown in Figu e 2, he ou come may
ange om ully accep able (ensu ing a nominal se o cogni i e esou ces) o no
accep able (o deg aded), whe e he en i onmen becomes p one o e o s bo h cogni-
i ely and physically. Al hough a single ac o , such as low igilance, may each an
unaccep able le el, i is p esumed ha in e ac ions among ac o s should allow o
compensa o y mechanisms o be iden i ied, he eby demons a ing he o e all accep -
abili y o he model [10]. I was de e mined ha he HPE should p io i ise h ee c i i-
cal aspec s o human pe o mance (wo kload, s ess, and si ua ion awa eness) as he
3
mos ele an measu es o conside . While addi ional ac o s, such as a igue and
a en ion, we e ecognised as being signi ican , challenges in ob aining eliable indi-
ca o s posed a po en ial limi a ion o he alidi y o he esul ing model. Based on he
exis ing li e a u e, his wo k aims o popula e he h ee main HPE ac o s wi h ehicu-
la , beha iou al and physiological pa ame e s. This wo k also aims o lay he ounda-
ions o he applica ion o he HPE concep o a con olled scena io in a simula ed
en i onmen , needed o moni o ing he e olu ion o he use 's d i ing pe o mance on
he basis o a p ede ined sequence o e en s.
Fig. 2. Rep esen a ion o HPE spide web model (le ) and example o e en - ela ed e olu ion
o HPE pa ame e s ( igh ). Adap ed om [9]
O he esea ches on he subjec manages o moni o indi idual “pa ame e s” in he
model using a usion o da a sou ces wi h excellen esul s. Fo example Liu (2020)
[11] manage o moni o and p e en a igue by moni o ing hea bea and d i ing pos-
u e; u he mo e Schneegass e al., [12] moni o d i e wo kload by measu ing skin
conduc ance, hea bea , empe a u e senso s and d i e came a; Lee a al., [13] es i-
ma ed d i e 's s ess employing he use o physiological signals and s ee ing wheel
mo ion analysis.
Despi e he e iden e icacy o moni o ing d i ing pe o mance using a soli a y
“pa ame e ”, i is he conside ed opinion o he au ho s ha a combina ion o hem
and a usion o ins umen s is equi ed in o de o achie e a mo e comp ehensi e and
comple e moni o ing o he e olu ion o d i ing beha iou me ics [14].
In his espec , he objec i e o his wo k is o emphasise he possibili y o ackling
he p oblem o moni o ing d i e s a us om he opposi e pe spec i e. Tha is o say,
i is p oposed ha he o al po en ial o senso s be exploi ed by using he same ins u-
men o moni o se e al pa ame e s.
In o de o explo e he concep o HPE in a simula ed en i onmen , i is necessa y
o employ ques ionnai es (p e and pos simula ion) in o de o moni o he subjec i e
pe cep ion o he use . The ools ha a e he subjec o his a icle a e illus a ed in
Figu e 3. These include he ollowing: ques ionnai es, eye acke , elec oencephalo-
g am (ECG), elec oca diog am (EEG), ins umen ed s ee ing wheel (ISW), and elec-
ode mal ac i i y (EDA).
4
Fig. 3. P oposed expe imen al se up
The applica ion o his app oach ype, has he po en ial o be pa icula ly in e es -
ing especially in he con ex o u ban d i ing scena ios. This is because, acco ding o
s a is ics, he occu ence o acciden phenomena is mo e equen in an u ban en i-
onmen and in gene al d i e s a e mo e s essed by ex e nal s imuli [15,16]. The
u ban en i onmen is cha ac e ized by a he e ogeneous a ic composi ion and oads
a e he public space in which his a ic lows. Acco ding o he I alian egula ion,
hey a e de ined as “ he a ea o public use in ended o he ci cula ion o pedes ians,
ehicles and animals”. This de ini ion implici ly implies he need o design he oad
acco ding o he dic a es and needs o di e en use s. To ensu e he sa e y o all o
hem, he NACTO guidelines [17] de ine “In an u ban con ex , s ee design mus
mee he needs o people walking, d i ing, cycling, and aking ansi , all in a con-
s ained space. The bes s ee design also adds o he alue o businesses, o ices, and
schools loca ed along he oadway”. Jus as he design o he s ee akes on unda-
men al alue in ensu ing sa e y, he use 's esponse o d i ing ask and su ounding
en i onmen akes on impo ance a he same le el and in he same way.
2 Li e a u e e iew
To suppo he heo e ical ounda ion o he HPE in he con ex o eal- ime d i e
moni o ing, a bibliome ic analysis was pe o med using VOS iewe [18]. The sea ch
e m " eal- ime d i e moni o ing" was applied o he PubMed da abase and a ne wo k
isualisa ion was gene a ed based on he co-occu ence o keywo ds in he ele an
li e a u e. As shown in Figu e 4, “human” esul as he cen e o he esea ch land-
scape and he majo keywo d adop ed in he speci ic esea ch ield unde analysis,
ac ing as nexus be ween he di e en hema ic clus e . As ma e o ac his is in ac
cohe en wi h he aim o he HPE concep . Fou kinds o clus e s can be no iced: ed,
yellow, g een and blue. The la ge clus e is he ed one and includes such as d i ing,
moni o ing, physiological, acciden s, a ic, and machine lea ning, e lec ing an ac-
5
i e body o esea ch ocused on physiological and cogni i e moni o ing o d i e s. I
is impo an o no e ha he HPE concep inds i s na u al colloca ion wi hin his pa -
icula clus e . This ed domain is in ac s ongly aligned wi h he co e p inciples o
HPE, whe e beha iou al and senso -based inpu s a e c i ical o de e mining d i e ’s
pe o mance s a us. Adjacen clus e s u he en ich he HPE cons uc . A g een clus-
e highligh s ac o s ela ed o demog aphics and clinical condi ions, such as age,
bioma ke s and ea men ou comes, sugges ing he ele ance o in e -indi idual di -
e ences. The concep o HPE, being based on he indi idual's d i ing pe o mance
and condi ion, is in insically linked o pe sonal condi ions. I can hus be concluded
ha he 'op imal' o 'deg aded' condi ion o pa ame e s is in luenced by hese pe sonal
di e ences. A blue clus e add esses me abolic and psychophysical a iables, such as
blood glucose and hypoglycaemia, which may in luence d i e beha iou unde ce -
ain condi ions. In addi ion, pe iphe al clus e s cap u e he b oade ecosys em, includ-
ing en i onmen al ac o s (ai pollu an s, ehicle emissions) and echnological dimen-
sions (algo i hms, eal- ime sys ems), highligh ing he po en ial need o in eg a e ex-
e nal in luences in o he HPE model. This isualisa ion con i ms ha he li e a u e
add esses many o he in e dependen ac o s en isaged in he HPE amewo k, albei
o en in isola ion. The iden i ica ion o missing links be ween di e en clus e s by
means o he g aphical ool VOXViewe indica es he possibili y o in e ening wi h
esea ch opics ha ill p ecisely hese gaps, he eby in oducing inno a ion o he
ma ke heme. By mapping hese elemen s, he analysis suppo s he a icle's goal o
popula ing he HPE dimensions wi h ehicula , beha iou al and physiological a ia-
bles and explo ing hei in e ela ionships. I also highligh s he agmen ed na u e o
cu en app oaches and ein o ces he need o a uni ying model such as he HPE o
in eg a e mul iple da a s eams o imp o ed d i e s a e moni o ing. This sec ion
p esen s also some o he con ibu ions in he li e a u e ha link he ins umen a ion
p oposed in Figu e 2 o he h ee HPE p io i y me ics (wo kload, s ess and si ua ion-
al awa eness).
Fig. 4. VOS iewe ep esen a ion o esea ch amewo k
6
2.1 Eye acke
Bi kina e al., [19] by means o a s udy conduc ed in 2022 shows ha ocula me -
ics such as ixa ion du a ion, gaze poin and pupil diame e a e s ongly co ela ed
wi h pe cei ed wo kload while d i ing ask, sugges ing he use ulness o such me ics
in wo kload’s le el classi ica ion and p edic ion. Fu he mo e, A ias-Po ela e al.,
(2023) [20] p o ided an a icula ed e iew on si ua ion awa eness assessmen by
means o eye acking me ics. In his e iew, 38 scien i ic pape s we e conside ed,
e ealing a wide ela ionship be ween he eye- acking me ics and si ua ion awa e-
ness.
2.2 Elec oca diog aph (ECG)
A ecen e iew ca ied ou by S i anga e al., (2023) [21] p o ided b oad e idenc-
es o how ECG me ics (speci ically hea a e a iabili y (HRV)) can be employed o
moni o d i e men al wo kload. In he ield o si ua ion awa eness, his s udy [22]
de eloped a pa adigm o measu ing si ua ional awa eness in he con ex o high-
speed ain d i e s and examined he ela ionships be ween SA le els and physiologi-
cal measu es, including ECG signals. Ras goo e al., (2021) [23] de eloped a sys em
o classi y s ess le els in eal ime employing ECG signals. Pa ame e s’ op imiza ion
signi ican ly imp o ed sys em’s accu acy, enabling i o each each 77.78% on a
da ase collec ed using ad anced d i ing simula o .
2.3 Elec oencephalog aph (EEG)
In esea ch conduc ed by Di Fulme i e al., (2018) [24] EEG-based index o men al
wo kload o assess he impac o a ious ac o s on d i e beha iou in eal d i ing
condi ions. Resul s showed ha a ic condi ion and he ype o oad had a signi ican
impac on d i e wo kload, wi h he EEG p o ing mo e sensi i e me ics han subjec-
i e measu es like ques ionnai es. The e iciency in he moni o ing o si ua ion
awa eness du ing d i ing asks ha e been explo ed in a s udy [25] which made use o
EEG and unc ional nea -in a ed spec oscopy ( NIRS); demons a ing i s po en ial.
2.4 Ins umen ed s ee ing wheel (ISW)
Those wo s udies [26,27] used an ins umen ed s ee ing wheel o analyse he o c-
es h ough load cells momen s applied by he d i e du ing d i ing in a simula o .
Resul s showed ha analysis o hese pa ame e s can p o ide de ailed in o ma ion
abou he d i e 's wo kload du ing complex manoeu es. In a con ibu ion p o ided
by Saha e al., 2021 [28] hey explo e he easibili y o employed g ip o ce on s ee -
ing wheel as a measu e o s ess le el. By showing s ong posi i e co ela ion be-
ween he pa ame e s, hese esul s p o ide ini ial e idence ha g ip o ce can be used
o measu e s ess in d i ing asks. A esea ch a icle p oposed by [29] analysed he
in e ac ion be ween he d i e and he s ee ing wheel in eme gency si ua ions using an
ins umen ed s ee ing wheel. The esul s showed ha he measu emen o applied
7
o ces and momen s can p o ide use ul in o ma ion o unde s anding he d i e 's
si ua ion awa eness du ing c i ical e en s.
2.5 Elec ode mal ac i i y (EDA)
In a con ibu ion ca ied ou by Dogan e al., (2019) [30] he applicabili y o EDA
in he moni o ing o he s ess le el by compa ing i wi h su ey. Resul s showed a
consis ence o 87.5% be ween he EDA senso s esul s and he answe s p o ided in
he ques ionnai es. Mo eo e , [31] based on esea ch conduc ed on a simula ed en i-
onmen , s a ed ha ECG and EDA signals a e sensi i e o a ia ions in wo kload.
2.6 Ques ionnai es
Fo wha conce n he sel -assessmen o wo kload le el he e a e di e en ypes o
ques ionnai es which can be employed. Among hese we can men ion NASA Task
Load Index (NASA-TLX), O e all Wo kload Scale (OW). The i s one is mul idi-
mensional a ing p ocedu e ha p o ides an o e all WL sco e based on a weigh ed
a e age o a ings on 6 subscales [32]. OW ins ead is a unidimensional measu e a ing
o he subjec 's o e all wo kload on a unidimensional scale o 0 o 100 [33].
Ano he aluable ques ionnai e which can be employed o measu e d i e condi-
ion is he A en ional Ne wo k Tes (ANT). I is a compu e -based es o measu e
use s’ pe o mance in h ee componen s o a en ion: ale ing, o ien ing, and execu-
i e con ol. F om i s i s de elopmen by Jin Fan e .al, [34] i has been widely em-
ployed in basic and applied esea ch s udies which ha e been conduc ed on a ious
opics including d i ing beha iou , speci ically o p edic d i ing es sco es [35].
Complemen a y o he o iginal e sion, a new e sion in which a measu e o igi-
lance is added is called A en ion Ne wo k Tes o In e ac ions and Vigilance
(ANTI-V) [36]. I will be especially bene icial in e alua ing speci ic hypo heses con-
ce ning igilance, in conjunc ion wi h o he a en ional unc ioning me ics, such as
phasic ale ness, o ien ing, and execu i e con ol.
The e olu ion o d i e s a e condi ion o e ime is p one o modi ica ion induc ed
by ex e nal ac o s; hose includes a ic condi ions, a ic ype and he d i ing con-
ex (e.g., mo o ways o u ban) [16]. Depending on he p e iously men ioned ac o s,
du ing he same ip unde aken by use s, speci ic e en s o ac ions may be obse ed.
These includes o ins ance ollow he leading ehicle (also known as leade ehicle),
o e aking, accele a ions / ha sh accele a ion and b aking / ha sh b aking. Each o
hese ac ions en ails a cha ac e is ic cogni i e load and has he capaci y o induce
luc ua ions in emo ional and a en ional s a es, especially when hey occu in apid
succession o in pa icula i y demanding s ess-le el scena ios.
The in e play o hese con ex ual and beha iou al ac o s is o pa amoun im-
po ance in he s udy o d i e s a e dynamics, pa icula ly in ela ion o he b oade
amewo k o HPE. I is impe a i e o comp ehend he manne in which hese com-
ponen s in e sec o in luence he d i e 's condi ion o he ad ancemen o sophis i-
8
ca ed d i e -assis ance sys ems (ADAS), human-machine in e aces (HMI), and au-
onomous d i ing echnologies ha a e a uned o eal-wo ld beha iou al a iabili y.
Based on he abo e-men ioned con ibu ions a schema iza ion o physiological,
ehicula and beha iou al- ela ed pa ame e s is epo ed in (Tab. 1).
F om a me hodological pe spec i e, he adop ion o simula ed d i ing en i on-
men s cons i u es a powe ul ins umen o scien i ic in es iga ion. Indeed, simula-
o s o e o esea che s a high-deg ee o expe imen al se up con ol, enabling he
sys ema ic manipula ion o a ic condi ions, en i onmen al complexi y, and ask
di icul y [37]. This, in u n, acili a es he gene a ion o bespoke scena ios ha simu-
la e a b oad spec um o use engagemen le els and s ess-inducing si ua ions.
Th ough he implemen a ion o such con olled expe imen a ion, i becomes possible
o isola e speci ic a iables and examine hei e ec s on d i e beha iou , pe o -
mance, and psychophysiological esponses wi h a deg ee o p ecision ha is o en
una ainable in na u alis ic s udies.
Consequen ly, simula ed en i onmen s acili a e he eplica ion o eal-wo ld d i -
ing condi ions in a sa e and epea able manne , as well as suppo ing he de elopmen
o obus p edic i e models o d i e s a e a ia ion.
Table 1. Di e en ia ion be ween physiological, ehicula and beha iou al- ela ed pa ame-
e s
Physiological pa ame e s
Vehicula and Beha iou al-Rela ed Pa ame e s
Eye T acke
Accele a ion (longi udinal/la e al)
EEG
Speed (longi udinal/la e al)
ECG
ISW
EDA
B aking ac i i y
3 Ma e ials and me hod
In his sec ion, based on exis ing li e a u e and on he undamen s highligh ed up o
his poin , au ho s p opose a combina ion measu e based on p e iously men ioned
senso s capable o moni o d i e ’s e olu ion o HPE undamen al pa ame e s, name-
ly: si ua ional awa eness, wo kload and s ess. In pa icula , au ho s p opose his se
o ools which can be applicable o e alua e he e olu ion o he pa ame e s in case o
con olled simula ed es ca ied ou h ough he usage o a dynamic simula o . The
combina ion o physiological, beha iou al, ehicula , and subjec i e measu es is ex-
pec ed o enable a ho ough e alua ion o :
1. Si ua ion Awa eness, ha in ol es pe cei ing, comp ehending, and an icipa -
ing changes in he d i ing en i onmen . I is in luenced by ehicle pe o -
mance pa ame e s such as ehicle longi udinal speed, ehicle la e al speed,
and ehicle o a ions (pi ch, oll, yaw). The posi ion o he cen e o mass in
ela ion o he global e e ence sys em can e lec how well he d i e pe -
9
cei es he dynamics o he ehicle, aiding in hei si ua ional unde s anding.
While speci ic s udies di ec ly linking hese ehicle dynamics o si ua ion
awa eness a e limi ed, ongoing esea ch in d i e beha io modeling con in-
ues o explo e hese associa ions. [38]
2. Wo kload, ha e lec s he cogni i e demands on he d i e and can be as-
sessed using pa ame e s such as hea a e a iabili y (HRV), ehicle accel-
e a ion, and b ake pedal posi ion. Inc eases in ehicle longi udinal accele a-
ion and b ake pedal o ce ha e been linked wi h highe wo kload, as hese
pa ame e s indica e he le el o con ol a d i e mus apply o d i e e ec-
i ely [39].
3. S ess, which is closely ela ed o physiological esponses like hea a e,
hea a e a iabili y, and g ip o ce. HRV is widely s udied as an indica o
o s ess, pa icula ly h ough he analysis o he LF/HF a io, whe e a shi
owa ds lowe equencies is associa ed wi h highe s ess le els. S ess e-
sponses a e also e lec ed in ehicle dynamics; o example, sudden changes
in ehicle longi udinal accele a ion and b ake pedal posi ion can indica e
momen s o high emo ional o cogni i e s ess.
Each da ase migh be mapped o key HPE Pa ame e s, allowing o a quan i a i e and
quali a i e assessmen o d i e pe o mance and he impac o human s a e on pe -
o mance in d i ing si ua ions ac oss di e en le els o au oma ion. By combining
hese pa ame e s, he p oposed da ase (Table 2) in ends o iden i y pa e ns and
h esholds ha de ine di e en d i e s a es, he eby acili a ing he de elopmen o
mul i-pa ame e moni o ing sys ems.
Table 2. Mapping o HPE pa ame e s and ela ed measu es
HPE Pa ame e
Senso
Measu e
Desc ip ion
Wo kload
EEG
𝜃𝑓𝑟𝑜𝑛𝑡 / 𝛼𝑝𝑎𝑟
Rela ionship5be ween5 on al5 he a5wa es5and5pa i-
e al5alpha5wa es5
𝜃𝑓𝑟𝑜𝑛𝑡
/
𝛽𝑓𝑟𝑜𝑛𝑡
Ra io5be ween5 on al5 he a5and5 on al5be a5wa es5
ISW
To al d i e g ip o ce
Measu emen o he o ce exe ed by he d i e on he
s ee ing wheel moni o ed by load cells
Eye acke
PERCLOS
Pe cen age o ime du ing which he eyes a e pa ially
o comple ely closed
Fixa ion du a ion
A e age eye ixa ion ime
Dynamic
d i ing
simula o
Vehicle longi udinal speed
Measu es needed o de ine he kinema ic o he ehicle
Vehicle la e al speed
Vehicle longi udinal accel-
e a ion
Vehicle la e al accele a ion
B ake pedal posi ion
S ess
ECG
Hea Ra e (HR)
The speed a which he hea bea s
Hea Ra e Va iabili y
Mean NN e e s o he a e age du a ion o he no mal-
