Gamiﬁcation of EcoDriving Behaviours through

Intelligent Management of Dynamic Car and Driver

Information

Nousias Stavros

, Aris S. Lalos

, Christos Tselios

, Dimitris Bitzas

Dimitris Amaxilatis

, Ioannis Chatzigiannakis

, Arvanitis Gerasimos

and Konstantinos Moustakas

University of Patras, Department of Electrical Engineering, Greece

Sparkworks ITC Ltd, U.K.

Sapienza University of Rome, Italy

Abstract. Driving style is seen not only to become a signiﬁcant cause of green-

house gas (GHG) and other air pollutant emissions but also a critical parameter

regarding road safety, with huge social & ﬁnancial adverse effects. In this work

we provide a conceptual architecture of a highly innovative and interactive Seri-

ous Games platform that will empower and guide users to adopt an eco-friendly

driving style. This will be achieved, without distracting users from safe driv-

ing, through a multidisciplinary approach aiming at the development of a user

friendly, unobtrusive multi-player gaming environment, where the users will not

only play collaboratively/competitively using their mobile device but also using

the car itself and their own bodies, thus turning eco-driving into an immersive and

highly motivating experience. The sensing infrastructure of the proposed system

will not only acquire data related to driving from an OBD sensor that will capture

a complex set of parameters related to eco-driving, but will also sense environ-

mental and physiological parameters of the driver, so as to better position the state

of the system (car) in context (environment, user). The use of virtual user mod-

els and cognitive modeling of the users, will further boost personalization and

adaptation of the game itself with respect to the needs of the individual driver.

The impact of such a holistic and innovative approach is huge and the founda-

tions laid here are expected to result in a widespread adoption of sensor-based

gamiﬁcation platforms in areas going far beyond eco-driving

Keywords: Gamiﬁcation · Ecodriving · Augmented reality · Virtual reality

1 Introduction

Road transport is one of the major causes of the environmental pollution. According to

a recent study [4] it is responsible for about 30% on the total emissions of CO2 into

the atmosphere. Among the actions individuals can take to reduce their green-house

gases associated with personal transportation, there is to operate their current vehicles

more efﬁciently [7]. Recent studies [5] have shown that in certain situations the driver’s

driving style can result in differences in terms of fuel consumption (and therefore CO2

emissions) from 2 up to 35% between a calm driver and an aggressive one [58]. One

100

Stavros, N., Lalos, A., Tselios, C., Bitzas, D., Amaxilatis, D., Chatzigiannakis, I., Gerasimos, A. and Moustakas, K.

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information.

DOI: 10.5220/0008862301000123

In OPPORTUNITIES AND CHALLENGES for European Projects (EPS Portugal 2017/2018 2017), pages 100-123

ISBN: 978-989-758-361-2

important action to reduce the environmental impact caused by road transport is there-

fore to educate drivers to adopt a driving style that is as eco-friendly as possible. At this

point, it should be also mentioned that numerous studies have underlined the substan-

tial ecological [34], economic but, also, road safety adverse beneﬁts that can be derived

from adopting eco-driving behaviors [7, 38]. Road trafﬁc injuries are a major public

health problem in the WHO European Region and cause the premature deaths of some

120 000 people every year, they are the leading cause of death in children and young

adults aged 5 to 29 years. In addition to these deaths, about 2.4 million people are es-

timated to be so seriously injured as to require hospital admission each year. Although

there are several potential causes of trafﬁc crashes, and the injury severity sustained in

the crashes, a leading cause is aggressive driving. Driving style is inﬂuenced by a com-

plex mixture of technical, social, psychological [31] and cultural factors, with the latter

tending to have a greater inﬂuence than the former: How a driver reacts to a given situ-

ation is more likely to be inﬂuenced by factors such as his age, values, social position,

concentration, stress and attitude to risk, than by trafﬁc regulations or the driver’s ability

to use the vehicle controls [46]. It is more a question of attitude than ability. To change

their driving behavior, car drivers have to change their automated behavior into delib-

erately controlled behaviour – a process that requires considerable effort. More impor-

tantly, car drivers need to be motivated to change their behavior [32] (i.e., they need to

form an intention [31] to practice eco-driving), and second, they have to put this inten-

tion into practice. Many studies have report cultural, technical, and educational barriers

inhibiting its adoption. Recent studies [18] also found that eco-driving behaviours are

related to time saving and safety goals, therefore, can be easily violated in demanding

trafﬁc environments or in time pressure situations. The most important educational ele-

ment in changing driver behavior is the positive feedback from taking the desired action.

Behavioral theory strongly conﬁrms that unless the individual can see or feel the results

of his/her actions—preferably on an immediate and continuous basis—that individual

is unlikely to maintain the behavior over time. Therefore, as eco-driving results are so

small, unless personalized feedback on the collective effort is provided, the driver is

unlikely to perceive important changes in his driving style affecting the fuel economy

and the society in general. To this end, innovative strategies that will increase driver

motivation to make changes in their driving behaviour are required. Serious gaming

and Gamiﬁcation has been positioned as a powerful approach, tool, or set of techniques

that guides targeted behaviour change to improve the way that various activities are un-

dertaken so that those involved begin to take the desired actions while they experience

more fun, enjoyment, and pleasure in their tasks [47]. The gamiﬁcation term refers to

a designed behaviour shift through playful experiences and became popular in 2010

and dominates especially areas with human interaction and focus on the quality of the

experience since then [42]. However, in order to effectively motivate users to adopt de-

sired behaviors, the games should provide information so that the users themselves will

be able to evaluate their behaviour and increase their awareness on the negative conse-

quences that it may have. In this context, we aim to develop highly innovative and Inter-

active Serious Game Platform that will motivate users to adopt an eco-friendly driving

style without distracting them from safe driving. The developed playful interventions

will be adapted to the user characteristics through an intelligent unobtrusive sensing

101

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

101

platform that will utilize late-breaking personalized user models so as to optimize the

individuals driving behaviour shift through playful experiences that will take place in

the real world and real-time rather than in a virtual/ simulation environment, which is

the current state-of-the art in eco-driving education. Our major objective is to provide a

highly innovative and interactive Serious Games platform, named GamECAR, empow-

ering drivers to manage their driving behavior by: i) conveying educational messages,

ii) encouraging certain activities through game-based elements such as competition or

rewards and iii) providing them with user friendly tools that will increase their aware-

ness of their ecodriving state and effectiveness of the intervention strategies. Critically,

the proposed system is expected to guide users to adopt an eco-friendly driving style,

without distracting them from safe driving, through a multidisciplinary approach aim-

ing at the development of a user friendly, unobtrusive multi-player gaming environment,

where the users will not only play collaboratively/competitively using their mobile de-

vice but also using the car itself and their own bodies, thus turning eco-driving into an

immersive and highly motivating experience. A pipeline of advanced algorithms and

data fusion techniques starting from raw measurement processing, feature extraction,

EcoDriving indicators and personal proﬁle data will ensure EcoDriving state awareness

and a timely optimal intervention strategy will be provided using the digital gaming

technology. In addition, a ”personal driving guidance system” will empower drivers

to optimize their action plans towards personalized preset goals and guidelines (e.g.,

maintain speed, switch off engine when stationary, shift up to low revs, etc). Finally the

system will provide to the user i) Serious Games (including Augmented Reality (AR)

games) responsible for monitoring driving behavior & delivering action personalized

plans that will help user to maintain a green driving style without distracting them from

safe driving and ii) a multiplayer gaming environment where users can monitor their

ecodriving score evolution, set missions and invite other to participate collaboratively

or competitively. The rest of the paper is organized as follows, in section 2 we give

an overview of the existing eco-driving assisting systems, in section 3 we analyze the

details of our approach and ﬁnally in section 4 we end with discussion, conclusion and

future direction related to this topic.

2 Related Work

The current ecodriving systems can be divided into two major categories: i) the eco-

driving real life applications and ii) the simulators. The most representative approaches

available in the market are shortly described in the rest part of this section.

2.1 Ecodriving Mobile Applications and Driving Assistance Systems

Smartphone Applications: Studies have shown that eco-driving smartphone applications

have impact on fuel efﬁciency (Tulusan et al., 2012). Greenmeter[43], FuelGood [54],

TEXACARe [53] and Geco [40] are good examples of smartphone applications that

have been developed in order to track fuel consumption and increase efﬁciency. Some

of them works in real time since others provide a summary at the end of each jour-

ney. TEXACARe and Geco provide a score based on actions that have impact in fuel

102

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

102

consumption and driver’s behavior Remote Control Solution: WeNow [59] is a solution

mainly aimed at vehicle ﬂeets and aids them in increasing their overall fuel efﬁciency.

The device of the platform is connected to the OBD II interface of the vehicle in or-

der to be able to collect the relevant vehicle data such as mileage, fuel consumption,

etc. Advanced Driver-Assistance Systems: FIAT

eco:DRIVE APP [24], Ford Smart

Eco Driving [25], BMW Eco Pro[10], Honda

Insight Eco Assist system [27], Nis-

san Eco-Drive Support Technology, Subaru Ecology [49] and Mitsubishi ECO Drive

Support [36] are fully implemented systems. Some of them are based on apps that com-

municate directly with the car which purpose is to improve the driving style, lowering

fuel consumption and tracking the carbon dioxide emissions. Others are build-in sys-

tems providing feedback in real-time for maximizing the fuel efﬁciency. Finally, one

approach acts as a medium between driver’s actions and car’s engine inputs to avoid

non-eco behaviors, although the system turns this functionality off in cases of emer-

gency. traction (e.i. driving at dusk, on slippery surfaces, snowy environments etc.). In

this simulator, ecodriving is just an extra feature.

2.2 Ecodriving Simulators & Games

This section presents three simulators (ST Software Simulator Systems with EcoDriv-

ing Package, DriveSim and Simescar Silver – Simumak) and four games (EcoDriver,

Stagecoach, Truck Fuel Eco Driving and SafetyDrivingSimulator: Car) that focus on

teaching principles of EcoDriving to drivers. ST Software Simulator Systems Eco Driv-

ing Package [51]: The simulator teaches how to save up to 20% or 30% of fuel by

applying another style of driving. It offers a way of experiencing a new driving style

arising from being mindful about fuel consumption. The simulator focuses on both the

ﬁnancial and environmental gain obtained from this driving style.

DriveSim simulator [21]: The simulation program DriveSim allows the driver to

practice as if he/she were commanding a real vehicle, thanks to its realistic situations

and environment. DriveSim scenarios include real trafﬁc and pedestrians. This program

offers the possibility of doing different tours with any climatic settings, timing and

traction: driving at dusk, on slippery surfaces, snowy environments, with rain or even

practice emergency braking with and without ABS. DriveSim provides the opportunity

for initial training on track, overtaking practice, driving on urban roads, service roads,

roundabouts and efﬁcient driving, among many other options. EcoDriver [22] tests the

players driving in an endless, randomly generated environment, in an attempt to demon-

strate the advantages of driving safely, economically and in an environmentally friendly

way to the player. Players must master a dangerous terrain and treacherous trafﬁc sit-

uations, all the while saving fuel and being aware of the situation on the road. Truck

Fuel Eco Driving: This game places the player behind the driving wheel of a Renault

Premium Optifuel, in order to let them discover the principles of economic driving.

Different challenges in urban environments and secondary road networks are posed, in

which the player needs to obtain the lowest fuel consumption possible and the highest

commercial speed possible.

Table 1 depicts existing worldwide competing technology or products in this market

place. Most of the available solutions simply provide educational messages without

taking into account speciﬁc characteristics of the individual driver. In addition none

103

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

103

Table 1. Related projects and competitive products.

GamECAR

Greenmeter

Garmin Mechanic with

ecoRoute

FIAT eco:DRIVE APP

CarbonDiem

FuelGood

Honda Insight Eco

Assist system

EcoDrive One

Speedometer

Simulators (e.g. ST

Software Simulator

systems Eco driving

package , etc. . . )

Games (e.g.

EcoDriver,stagecoach

,Truck Fuel eco driving)

Driver behavior

sensing

3 3 3 3 3 3 3 3 7 7

Proper vehicle

maintenance

3 7 3 7 7 7 7 7 7 7

Mixture of

social status,

ﬁnancial status

parameters

3 7 7 7 7 7 7 7 7 7

Environmental

sensing

3 7 7 7 7 7 7 7 7 7

Physiological

data

3 7 7 7 7 7 7 7 7 7

Direct

feedback while

driving

3 3 3 7 3 3 3 3 7 7

Indirect

feedback

3 7 7 3 3 3 3 3 7 7

Education and

feedback

support

3 7 7 3 7 3 3 7 7 7

User models 3 7 7 7 7 7 7 7 7 7

Customized

personalized

Interventions

3 7 7 7 7 3 7 7 7 7

Interactive

Serious Games

3 7 7 7 7 7 7 7 7 7

Information

visualization

3 3 3 3 3 3 3 3 7 7

Customized

information

visualization

3 3 3 7 3 3 7 3 7 7

Driver

Community

Platform

3 7 7 3 7 7 7 7 7 7

Multiplayer

game support

3 7 7 3 7 7 7 7 7 3

Vendor

GamECAR

HUNTER

GARMIN

FIAT

CarbonDiem

Energy Saving Trust

HONDA

JoyNow!

Various vendors

104

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

104

of them guides targeted behaviour change to improve the way that various ecodriving

activities are undertaken so that those involved begin to take the desired actions in order

to adopt a more efﬁcient driving style.

3 GamECAR Approach and Methodology

An ecodriving coaching solution needs to realize the difﬁculty of trying to change be-

havioral patterns that have been internalized over years and should try to take account

of a driver’s social and cultural preconceptions. For example, it will be hard to encour-

age a driver to let the car coast in neutral if he or she believes that this is forbidden or

wrong. And it will be hard to encourage a driver to shift up early whilst applying mod-

erate gas pedal if he thinks this combination might strain an engine or waste fuel. In

addition, unlike private car drivers who respond well to ﬁnancial incentives for adopt-

ing sustainable driving behavior, corporate drivers are not ﬁnancially motivated, as their

fuel costs are reimbursed by their employer. Therefore, a coaching solution needs to be

aware of the driver’s reservations in order to address them head on, for example, by

explaining that regardless of the accelerator pedal position, a modern engine will not

inject excess fuel. A powerful approach, that have been proved very effective in guiding

targeted behaviour change to improve the way that various activities are undertaken is

gamiﬁcation. GamECAR aims to better understand and change the driving behavior and

its relation to a complex mixture of technical, social, psychological and cultural factors

by using the Gamiﬁcation approach. The proposed system aims to provide both direct

and indirect feedback in a playful manner to the driver, through user speciﬁc Interactive

Serious Games that will not distract him/her from safe driving. The distraction moni-

toring is based on physiological measurements (e.g., heart rate, respiration rate, etc.).

It will strengthen the driver’s motivation of adopting an ecodriving style through the

delivery of personalized feedback, monitoring driving inattention, guidance and educa-

tion. Through user speciﬁc interventions, ecodriving game aims to deﬁne an ecodriving

score. The score is initially constructed from prior knowledge on the ﬁeld and then

globally updated based on analysis of long-term observations of all drivers’ states. This

update will be then applied to the individual user models, modifying them accordingly,

to ﬁt different needs per driver.

The developed interactive Serious Games platform will empower and guide users to

adopt an eco-friendly driving style. This will be achieved, without distracting users from

safe driving, through a multidisciplinary approachaiming at the development of a user

friendly, unobtrusive multi-player gaming environment, where the users will not only

play collaboratively/competitively using their mobile device but also using the car itself

and their own bodies, thus turning eco-driving into an immersive and highly motivating

experience. The sensing infrastructure of GamECAR, shown in Figure 2 and 1 will not

only acquire data related to driving from an OBD sensor that will capture a complex set

of parameters related to ecodriving, but will also sense environmental and physiological

parameters of the driver, so as to better position the state of the system (car) in context

(environment, user). The use of virtual user models and cognitive modeling of the users,

will further boost personalization and adaptation of the game itself with respect to the

needs of the individual driver.

105

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

105

3.1 GamECAR Architecture

As far as the architectural view is concerned, GamECAR platform consists of various

components, depicted in Figure 1, which are described as follows: Car Sensor Network:

It consists of the wearable wireless body area sensors and an OBD II sensor, designed to

continuously monitor vehicle characteristics, driving style and a cluster of physiological

parameters while driving. This car sensor network (CSN), records several parameters

such as heart rate and respiration rate, which are used to monitor driver inattention

and in order to estimate correlations with the driving style. Mobile Device/Gateway:

The CSN system will be connected to the Mobile device (e.g., Mobile phone, Tablet),

which will send the data to the Streaming Data Sub-System. This sub-system is initially

responsible for fusing the streaming data to the Mobile Device/Gateway wirelessly via

Bluetooth, to enable the Real Time Analysis component, where the handling of risk

situations is conducted with the use of real-time data mining techniques. The Streaming

Data Sub-System also sends data for storage to CSN. Real time analysis conducted in

Streaming Data Sub-System, will inter-connect with the AR GAME, so as to produce

warnings to drivers in case of danger. It will also interact with the VUM, to update it

in case of risk. Ofﬂine Analysis Server: It consists of a central server infrastructure that

is responsible for Ofﬂine Analysis of data and communication with a remote Database,

that will be mainly used for scientiﬁc analysis and VUM handling. Data from Real

Time Analysis conducted in Streaming Data Sub-System will be transmitted once a day

through the Internet to the Ofﬂine Analysis Server, to aid ofﬂine analysis. In the Ofﬂine

Analysis Server, tests for ecodriving assessment will be carried out, in the direction

of a formal ecodriving indicator, reﬂecting changes in driving styles through the long-

term adjustment of the VUM. The adjustment parameters as well as the results will be

transmitted to the Remote DB, for storage and feedback to the ofﬂine analysis, as well

as to the VUM, through data fusion. Data fusion will be carried out under a Big Data

analysis scenario, taking into account the VUM adjustments in case of risk in CSN,

hence the VUM and data fusion interaction. Additionally, all data obtained from the

Streaming Data Sub-System will be stored in the remote DB, to capture driver current

status and correlate it to his/her driving behavior history, and combine it with the already

fused VUM data to construct a long-term frailty monitoring system. Intervention Sub-

System: Implements the Interactive Serious Games supporting AR rendering, interacts

with the Ofﬂine Analysis Server module, but through the aforementioned ofﬂine-online

data broadcasting, will also be able to comprise short-term events. This sub-system

will receive input from the remote EcoDriving assessment implemented in the Ofﬂine

Analysis Server, to produce driving style recommendations and vehicle maintenance

plans to older vehicles. These changes will then be communicated back to the ofﬂine

system, in order to reach the VUM, and also be stored for EcoDriving score evaluation.

3.2 Data Analysis

Towards assessing driver’s behavior and driving performance, a large amount of data

need to be sampled, transmitted, stored and processed. Sensor devices often failure

to provide continuous data streamings, due to transmission errors, threshold errors, or

106

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

106

Fig. 1. GamECAR system architecture.

generic system failures. To provide reliable inferences, further data processing is re-

quired, which provides more solid results when is based on continuous values without

missing entries and other type of abnormalities. Possible data inconsistencies and other

anomalies need to be identiﬁed while applying data mining techniques is an essential

step towards the extraction of ecodriving performance indicators and driving style clas-

siﬁcation.

The three main input sources of the project are focused on (i) the vehicle, (ii) the en-

vironment and (iii) the driver, acquiring and analysing all relevant data available to con-

struct a complete and personalized proﬁle for each instance. The data is then passed to

the EcoDriving assistance module, which comprises of three sub-modules, responsible

for feedback generation, game elements and information visualization. Functionalities

of the online data analysis and ofﬂine data analysis are also presented, along with their

general design directives and features. These include real time information about 1) Ac-

tive transmission gear 2) Intake air temperature 3) Engine coolant and oil temperature

4) Fuel type, consumption and tank level 5) Throttle position 6) Engine revolutions per

minute (RPM) 7) Calculated engine load 8) Vehicle speed 9) Emission requirements to

which the vehicle was designed

all particularly relevant towards constructing and maintaining a multi-fold and ac-

curate model for the driver, which is appropriately parameterized according to his/her

vehicle. It is important that these data should be further normalized to account for sev-

eral varying parameters, such as road type and trafﬁc, weather conditions, etc. Most of

this information is readily available from relevant cloud services. The main purpose of

this post processing is to prohibit factors beyond driver’s control affect the evaluation of

his/her Eco friendliness. This is important for conveying a sense of fairness to the partic-

ipating drivers/players, which in turn preserves engagement and motivation in the long

run. Apart from vehicular and environmental data, the monitoring of the driver per se,

via indicative biosignals can be proven crucial in forming an accurate proﬁle of not only

107

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

107

driving behaviour under certain circumstances, but also of broader personality traits in

long-term frames. GamECAR is going to utilize common wearables which provide real

time measurements for heart rate, respiration rate and muscle activity. This type of data

is highly correlated to the driver’s psychological state, and are parsed together with the

vehicle data to the EcoDriving assistance module. An overall presentation of the online

analysis dataﬂow is presented in Figure 2

Data Preprocessing. As global telecommunication market shifts towards the 5G era

and the fascinating technological advances it proclaims [9, 55], it becomes obvious that

new applications and verticals with a tremendous affect in our everyday lives are ap-

proaching. The deployment of a highly heterogeneous, always available, brisk and agile

network, offering inherent support for billions of interconnected devices with less than

1 millisecond end-to-end latency [6], will transform almost every application from sim-

ple daily entertainment to autonomous private transportation. The automotive domain is

somehow divided into two separate yet highly consolidating tracks, autonomous driving

vehicles (ADV) [56] and vehicle-to-everything (V2X) communication [57], with V2X

act as a key technology enabler for ADV by allowing moving vehicles to approach each

other more safely, thus enabling trafﬁc ﬂow optimization techniques and increased sit-

uational awareness. Vehicle-oriented notiﬁcations and alerts are propagated to nearby

infrastructure, properly equipped pedestrians or other vehicles, rendering every involv-

ing entity capable of reacting fast and efﬁciently regardless of the situation [26] for

instance, to the now hazardous cases of blind intersections, closed curves, lane switch-

ing and overtaking which often cause fatal car accidents worldwide. Smart Cities all

around the globe are deploying dedicated infrastructure, partially based on resource

savvy architectures such as Multi-Access Edge Computing (MEC) or Fog Computing

[6, 1] paradigms, for facilitating V2X communication in a seamless manner, since cit-

izen security and effective public/private transportation are considered top priorities.

However, one of the keys to efﬁcient and secure V2X communication applications is

the ability to deliver informative messages, even based on partially fragmented datasets.

Recently, novel signal processing tools such as matrix completion (MC) have been

widely adopted for mitigating similar effects in many research ﬁelds such as computer

vision [60] or signal reconstruction from partial observations. Their strong beneﬁt is

that they ensure exact reconstruction of a low rank matrix given a sufﬁcient number of

observations [13, 11]. Several modiﬁcations of the classical MC problem allowed the

application in undirected graphs [28] and subsequently in 3D models[57]. Furthermore,

[12] presented the recovery of signals from a sampling of its Fourier spectrum via ma-

trix completion. In this section, motivated by the necessity for fault tolerant systems

that will extract information from incomplete datasets, as well as the omnipresent de-

mand of limited data transfer through the congested Smart City networks, we present

an efﬁcient and low cost method of managing non-uniformities and uncertainties in

vehicle-related data. Capitalizing on the low-rank property of the generated matrices

compiled by partitioning and stacking timeseries data, we formulate an approach that

offers higher reconstruction quality as compared to the traditional MC approach, at fast

execution times. More speciﬁcally, the contributions of the proposed work [39] can

be summarized on the following points: 1) Formulation of a Laplacian based matrix

108

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

108

completion to cope with missing entries in timeseries 2) Establishment of an efﬁcient

approach to deal with computational complexity of the issue under investigation This

solution will allows us to retrieve adequately detailed information from smaller or com-

promised datasets through an algebraic interpolation process that intends to effectively

ﬁll the missing information gaps without changing the overall essence of the actual

dataset. Depending on the use case, the proposed solution could be used for analyt-

ics retrieved from artiﬁcially reconstructed data or simply as a method of lowering the

threshold of the necessary amount of data needed for a speciﬁc decision.

Fig. 2. Online analysis datastream.

Extraction of Informative Indicators. For obtaining all available data from the driver,

the vehicle and the surrounding environment of each driving session, a dedicated net-

work of sensors was utilized. More speciﬁcally, the driver may wear (i) the Xiaomi

MiBand 1S tracker which provides heart rate information, (ii) the Polar H7 band also

for recording heart rate signals and (iii) the Spire Stone respirator which measures the

expansion and contraction of the individuals’ torso and categorizes its breath as calm,

tense or focused. An OBD II Bluetooth adapter is attached to the car’s parallel port,

transmitting vehicle-related data such as speed, fuel consumption, engine RPM and

throttle position to a paired mobile device, wirelessly, using Bluetooth communication

protocol. On the aforementioned mobile device runs a custom Android application in

charge of (i) collecting all wirelessly transmitted data traces from both the wearables

and the OBD, (ii) obtain enviromental data related to weather, trafﬁc and road con-

dition from third-party online repositories, (iii) recording additional data from all its

embedded sensors, such as 3D acceleration, geolocation data and timestamps, (iv) ag-

gregate all data into one dataset before transmitting them to an online repository for

further analysis and long-term storage, the GamECAR Database. The application also

includes a real-time data preprocessing module which applies several algorithmic tech-

niques to the accumulated data, as well as a feature extraction module responsible for

identifying high level features and detect events while driving. Such events may involve

shift up and braking or changes in the vehicle’s acceleration and the driver’s heart rate.

These extracted features are the direct input of the physiological score calculation mod-

ule and are also cross-correlated with metrics retrieved from the GamECAR Database.

109

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

109

Provided that the GamECAR database contains analyzed data for each user, the phys-

iological score integrates historical, preprocessed and real-time data, thus providing a

holistic overview of an individual’s behaviour on the road.

Parameters for the Ecoscore Estimation. Based on state of the art gamiﬁcation ap-

proaches [16], the ecoscore aims to penalize high RPM values during gear shift-up

and cruising, abrupt braking, high acceleration. In the same direction, the aggressive-

ness score aims to penalize high lateral acceleration, abrupt braking and high variances

in throttle position and RPM. Ecoscore is a driving style indicator, namely a tool that

shows drivers’ ECO driving performance and, consequently, their efﬁciency during their

trips. By analyzing the driving behavior, individual drivers can dramatically improve

their driving style. In this way, we take into account the main parameters which can

negatively affect the ecoscore. The used formula for the estimation of the ecoscore aims

to penalize high RPM values during gear shift-up and cruising, abrupt braking and high

acceleration. Speciﬁcally, for the calculation of the ecoscore the following parameters

are used:

1) Shift ups RPM ecoscore parameter 2) Brake ecoscore parameter 3) Acceleration

ecoscore parameter 4) RPM ecoscore parameter 5) Cruising ecoscore parameter

The shift up RPM ecoscore parameter describes the evaluation of shift-up events as

a factor to calculate ecoscore. The general idea is to punish shift up events with RPM

values higher than 2500 and reward shift up events with RPM values between 2000-

2500. In Figure 3, an example is presented showing how the RPM is affected when a

”gear shift up” event happens. The reward is less if the driver shifted up after several

seconds above 2000 rpm. Fig. 4 shows that the duration of a shift up event has an impact

27/06/2018 Milan, 14-15 Nov. 2017 30

0.5

1.5

2.5

3.5

4.5

0-2000 2000-2100 2100-2200 2200-2300 2300-2400 2400-2500 2500-2600 2600-2700 2700-2800 2800-2900 2900-3000 3000-6000

Shift ups

RPM when shift up categories

RPM Histogram

Fig. 3. RPM and gear shift up events.

and speciﬁcally different impact for positive and negative factor values. The punishment

is more when the duration is several seconds after 2000 rpm. For the estimation of shift

up RPM ecoscore we use: (i) the peak RPM during the event, (ii) the time required to

complete a shift-up event, (iii) the vehicle speed, (iv) the mean vehicle speed, (v) the

mean throttle position, (vi) the speed variance and (vii) the throttle variance.

The brake ecoscore parameter describes the evaluation of break events as a factor

to calculate ecoscore. For the calculation of brake ecoscore parameter, we are looking

110

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

110

27/06/2018 Milan, 14-15 Nov. 2017 31

Fig. 4. Shift up events and different impact for positive and negative factor values.

for brake events which are divided into four categories: (i) soft engine brake, (ii) hard

engine brake, (iii) pedal brake and (iv) emergency brake.

The acceleration ecoscore parameter describes the evaluation of acceleration events

as a factor to calculate ecoscore. The acceleration ecoscore parameter estimation mainly

depends on (i) the brake event and (ii) the velocity reached during acceleration event.

The RPM ecoscore parameter describes the evaluation of acceleration events as a

factor to calculate ecoscore. For the estimation of RPM ecoscore, we take into account:

(i) the RPM event, (ii) an RPM event occurs when RPM > 2500 (these events are

also validated from Throttle position values if available), (iii) the maximum RPM value

during RPM event, (iv) the duration of an RPM event measured in seconds and (v) the

duration of temporal window.

The cruising ecoscore parameter evaluates the ecoscore during cruising. The cruis-

ing ecoscore is estimated using: (i) the vehicle speed, (ii) the mean vehicle speed, (iii)

the mean throttle position, (iv) the speed variance, (v) the throttle variance and (vi) the

temporal window.

Aggressiveness Score Estimation. The aggressiveness ecoscore calculation is based on

the following parameters: a) throttle position variance and engine RPM variance, b)

braking intensity, c) acceleration magnitude. High variance of engine RPM indicates

driver nervousness and improper vehicle handling. For example, signiﬁcant ﬂuctua-

tions in engine RPM are the result of a driver who is not focused on smooth cruising

and overall safe driving. Instead, they tend to accelerate many times for small dura-

tions, thus minimizing their fuel efﬁciency, eco-friendliness and safety for themselves

and others. The RPM variance parameter calculation is based on a 30 seconds tempo-

ral window for the calculate of RPM position variance. The standard deviation is then

111

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

111

normalized and scale it appropriately with experimentally deﬁned parameter µ to con-

tribute to the ﬁnal aggressiveness score. Furthermore, braking intensity is computed

during online analysis, based on the correlation between vehicle deceleration (as mea-

sured by vehicle speed decrease) and braking duration. This is intended to contribute

to the computation of aggressiveness score in a cumulative manner. Abrupt braking

events indicate poor planning from the driver’s part and a low level of situation aware-

ness regarding road conditions and surrounding trafﬁc. Finally, lateral and directional

acceleration obtained from the mobile device’s accelerometer can provide a meaning-

ful measurement of steering and overall driving aggressiveness. While abrupt braking

and engine RPM variance are a good metric of aggressiveness in slow sections of the

driving route, they fail to provide similarly robust information in situations that require

greater velocity. To alleviate this, the lateral and directional acceleration magnitude is

integrated across time, so as to provide higher sensitivity for assessing aggressiveness

in such cases.

Physiological Score Estimation. Driving is an activity that requires considerable alert-

ness. Insufﬁcient attention, imperfect perception, inadequate information processing,

and sub-optimal arousal are possible causes of poor human performance. Understand-

ing of these causes and the implementation of effective remedies is of key importance

to increase trafﬁc safety and improve driver’s wellbeing. Several non-invasive measure-

ments have been considered in the literature. These include Electroencephalography

(EEG) based methods [20, 3, 2], Electrocardiography (ECG) and Heart Rate Variabil-

ity(HRV) related methods [35, 30, 14], image processing based approaches [33] as well

as tissue imaging and blood oxygenation measurement based studies [15]. In several

cases the HRV timeseries are used in conjunction with other validated methods such

as galvanic skin response[17, 45, 19, 50], thoracic electrical bioimpedance (TEB) [37],

EMG [50], respiration [29]. The extracted features from the aforementioned metrics are

used in most cases as input to classic classiﬁcation approaches such as Naive Bayes,

SVM , KNN [14], neural networks [20], fuzzy clustering [30]. Recently, a study used

ECG data to derive mental stress in open road driving [41]. Electro cardiogram and

heart rate variability generate a series of time-domain, frequency domain-features or

nonlinear features computed over 24 h, ﬁve-minute-long short-term period, or ultra-

short-term period for under ﬁve-minute duration. Commonly referenced feature is the

LF/HF ratio associated with the sympatho-vagal balance [8, 44]. In a healthy organism,

there is a dynamic relative balance between the sympathetic nervous system(SNS) and

parasympathetic nervous system (PNS). SNS activation increases heart rate, whereas

PNS activation decreases it. Spectral analysis of HRV shows several frequency bands,

two of which are important here: a low-frequency component (LF; 0.04-0.15Hz) medi-

ated by both PNS and SNS, and a highfrequency component (HF; 0.15-0.4Hz) mediated

by PNS activation. As a result, the ratio of LF to HF power is sometimes used as an in-

dex of autonomic balance [23]. However later studies prove that this simple rule does

not apply in all cases [35] and that it is not a conclusive feature [48]. In most studies a

series of features are used all-together with state of the art classiﬁcation approaches to

differentiate stressful situations in a case speciﬁc manner.

The physiological score calculation is part of the Online Analysis module, in the

same level with the ecoscore and aggressiveness calculation, which together form the

112

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

112

computational core for the evaluation of eco-friendliness during each driving session.

The main input of the physiological score module comes from the real-time data pre-

processing module, which in turn receives its input from the Sensor Layer. The cal-

culation of physiological score is based solely on heart rate measurements due to the

seamless connection of available HR sensor to mobile devices without the mediation of

propriety services. As features for the physiological score calculation, the value heart

rate variability and the window based LF/HF frequency ratio describing the autonomic

nervous system balance are considered. Personalized thresholds are extracted based on

LF/HF frequency ratio, mean heart rate and standard deviation for a temporal window

using simple decision rules. The desired outcome of the physiological score would be

a binary decision where TRUE or 1 declares that the driver is in aggravated state and

FALSE or 0 if the driver is calm. Thus, two classiﬁcation approaches are employed:

– Decision Rule 1: The driver is assumed as stressed if his/her heart rate is higher

than the mean value plus the standard deviation of his/her heart rate based on the

personalized normal driving.

– Decision Rule 2: Given heart rate mean value and heart rate standard deviation

and LF/HF ratio for each annotated temporal we employ Random Forest and AD-

ABoost classiﬁers

For the evaluation a public dataset was employed [52] that includes data for n =

68 volunteers that drove the same highway under four different conditions: (a) No dis-

traction, (b) Cognitive distraction, (c) Emotional distraction and (d) Sensorimotor dis-

traction. Driving with no distraction means driving without secondary activities. This

type of driving is called Normal Drive (ND). A cognitive driving means driving under

a cognitive stressor. The cognitive stressor was mathematical questions and analytical

questions posed orally by the experimenter. An emotional driving means driving under

an emotional stressor. The emotional stressor was emotionally stirring questions posed

orally by the experimenter in two phases. Sensorimotor driving means driving under a

sensorimotor stressor. The sensorimotor stressor was texting back words, sent one by

one to the subject’s smartphone. The phase layout within each stressful driving con-

tentions; namely cognitive, emotional and sensorimotor stress, was as follows: 1) Phase

1: Driving without distractions for ∼ 80 s. 2) Phase 2: Driving while engaging in a

secondary activity for ∼ 160 s. 3) Phase 3: Driving without distractions for ∼ 240 s

(coincided with the detour). 4) Phase 4: Driving while engaging in a secondary activity

for ∼ 160 s. 5) Phase 5: Driving without distractions for ∼ 120 s.

Across all subjects, heart rates were calculated for each selected temporal window.

Then, the mean and standard deviation were calculated for each driving phase and for

each driver individually. Additionally, we separated each stress-loaded driving condi-

tion (e.g., Cognitive, Emotional, Sensorimotor) into two phases; namely the ”quite”

phase [Cognitive Drive (CD), Emotional Drive (ED), Sensorimotor Drive (SD)] and the

”original stress” phase [Stress Cognitive Drive (SCD), Stress Emotional Drive (SED),

Stress Sensorimotor Drive (SSD)]. In the ”original stress” phases the subjects are di-

rectly under a stress event (Phases 2, 4) while as ”quite” phases, it is represented the

period of time before, between and after a stress event (Phases 1, 3, 5).

113

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

113

In Figure 5, the heart rate distribution of different drivers for different driving con-

ditions is presented in boxplots. As it can be observed the drivers’ heart rate is higher

when they drive under stress events in comparison with the normal driving phases

●●

●●●●●●●

●●●●●

●●●●

1 ND

2 CD

3 ED

4 SD

5 SCD

6 SED

7 SSD

Driving Section

bpm

●

●●

●

●●

●

●●

●

●●

●

●●

●

●●

●

●●

●

●●

●

●●

●

●●

●

●●

●

●●

●

●●●

●

●●

●

1 ND

2 CD

3 ED

4 SD

5 SCD

6 SED

7 SSD

100

Driving Section

bpm

●

●●

●

●●

●

●●

●

●●

●

●●●

●

●●

●

●●

●

●●

●

1 ND

2 CD

3 ED

4 SD

5 SCD

6 SED

7 SSD

100

Driving Section

bpm

●

●●

●

●●

●

●●

●●●●

●

●●●

●●

●

1 ND

2 CD

3 ED

4 SD

5 SCD

6 SED

7 SSD

100

Driving Section

bpm

Fig. 5. Boxplot with the heart rate values in different driving sections for the ﬁrst 4 drivers.

3.3 Gamiﬁcation

Based on the state of the art approaches[16] and the ideas and concepts of gamiﬁcation

that have been proven to be most effective within an educational context – we can start

to map out how the gamiﬁcation of GamECAR will look. Furthermore, we can marry

the different driving activities involved in GamECAR along with a set of desirable be-

haviours we want the gamiﬁed environment to stimulate and generate, from that, the

game elements that we can utilise as the basis for the gamiﬁcation design. In general,

our gamiﬁed elements should be geared towards encouraging and supporting end users

to adopt the following desirable behaviours 1) Improved eco-driving skills 2) Improved

safe-driving skills 3) Better knowledge about the impact of different driving practices

4) Increased use of GamECAR while driving 5) Increased sharing of the GamECAR

app with others

The selected game elements for GamECAR derive from the identiﬁed gamiﬁcation

trends and best practices. As we mapped out there, these trends relate to the identiﬁed

core drivers typical in gamiﬁed environments and to the four game phases that are

commonly seen. The diagram in Figure 6 below explains the links between all these

ideas and elements:

Based on this knowledge, a set of game elements and design proposals have been

developed that reﬂect the presented gamiﬁcation strategies. Table 2 shows these game

114

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

114

Fig. 6. Gamiﬁcation approach.

elements, the concrete design proposals, and how they relate to different gamiﬁcation

strategies and trends. The introduced design proposals follow a basic rule. Each of them

pursues one of the desirable behaviors in GamECAR, and complies with the presented

gamiﬁcation trends and good practices. The diagram in Figure 7 presents the overall

gamiﬁcation design where all the proposals are connected to work together. The main

activity that the user will undertake as part of their interaction with GamECAR is driv-

ing.

Table 2. Corresponce of gamiﬁcation trends with core drives and game phases.

Gamiﬁcation Trends Core Drives Phases

Storytelling Unpredictability & Curiosity Discovery

Goals Development and

Accomplishments , Loss &

Avoidance

Scaffolding, Endgame

Challenges and Quests Social Inﬂuence & Relatedness

& Development and

Accomplishments, Loss &

Avoidance

Scaffolding, Endgame

Customization Ownership & Possession Scaffolding, Discovery,

Onboarding

Progress Scarcity & Impatience Discovery, Onboarding,

Scaffolding, Endgame

Feedback Loss & Avoidance, Social

Inﬂuence

Scaffolding, Endgame

Access / Unlocking of content Unpredictability & Curiosity Onboarding, Scaffolding

Building Motivation. After each trip, an overall EcoScore for the trip is be calculated

and presented to the user, but as a part of the overall experience, other more concrete be-

haviours and parameters related to speciﬁc driving skills are also calculated and stored

as speciﬁc data (and accessible as scores or as the trigger for advice/feedback to the

user). This includes the following: 1) Fuel consumption 2) Gear change 3) Acceleration

4) Braking 5) Aggressiveness All of these parameters and scores are in fact related to

the speciﬁc desired behaviours pursued in GamECAR. And there is also an implicit

aspect to this: just using the GamECAR application while driving is one of the desired

behaviors and therefore just by following this behaviour, the user should improve his

or her status within the gamiﬁed system. As a part of each speciﬁc trip EcoScore, other

gamiﬁed elements will be earned or achieved during each active driving session. This

will include: 1) Driving skills awards 2) Badges 3) Avatar outﬁts 4) Knowledge cards

Game elements will be customized behind the scenes in order to appeal to different

115

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

115

types of players during the different phases of the game. GamECAR drivers will gain

levels in the game, which require progressively more experience points (XP) as pre-

sented in Figure 8. These levels also determine the game phase, with each phase having

distinctive characteristics, features and objectives

Fig. 7. Overall presentation of gamiﬁcation design.

Game Phases. The game evolves in four discrete , coherent phases, namely, discovery,

onboarding, scaffolding and endgame.

– Discovery phase : Levels 1-4 – estimated completion after 90-120 minutes of driv-

ing depending on Ecoscore. This phase’s main purpose is to familiarize the par-

ticipant with the game’s elements, scoring features, feedback, levelling system

and driving statistics extraction. To this end, the user gets introduced to Levels,

Ecoscore, HUD alerts and basic driving statistics.

– Onboarding: Levels 5-14 – estimated completion after 6-8 hours of driving, approx-

imately 9-16 days with 30 minutes of daily driving and a decent number of com-

pleted missions. This phase aims to expose the player to more advanced features

of the game. Apart from the discovery phase elements, this phase includes leader-

boards and leaderboard points (LP), daily missions with lootbox rewards, large-

scope missions with XP/LP, knowledge cards, lootboxes (drop likelihood depends

on eco-driving performance, up to 5 drops per month) and proﬁle customization

(e.g. avatar).

– Scaffolding: Levels 15-39 – estimated time of completion after 45-57 hours of driv-

ing, approximately 2.5-4 months with 30 minutes of driving daily and a decent

number of completed missions. Scaffolding is the phase of the game that aims at

creating desirable habits to the player through his/her core drives. Along with all

the aforementioned features, the user now is granted the ability to create his/her

own crews in order to recruit other players as he gains a widely respected “veteran”

status.

– Endgame: Levels 40-50 (max) – estimated time to reach max level is approximately

1.5-2 months with 30 minutes of driving daily and a decent number of completed

missions. In the ﬁnal stage of the game, the user starts getting personalized mes-

sages according to his/her weak driving aspects. He also gains access to advance

information visualizations regarding his/her performance compared to global statis-

116

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

116

tics. Finally, most of his visible status is subject to inactivity decay, i.e. it is gradu-

ally diminished while the player remains inactive.

Fig. 8. Experience points distribution required for the player to level up as a function of current

level.

The features are then further customized based on the detected type of each player.

Players are classiﬁed into 3 predeﬁned classes, namely, Constant Improver, Safety en-

thusiast and Smart saver. The directions according to which the gamiﬁcation features

are customized as shown in Figure 10. The ﬁgure shows the core drives that gamiﬁca-

tion features should take advantage of in order to be most effective depending on each

player’s type and game phase. The ﬁgure is separated in two rows (representing early

and late phases of the game) and three columns, corresponding to the three aforemen-

tioned player types. During early phases, the focus is on core drives like Epic Meaning,

while later the focus shifts to Loss and Avoidance, Accomplishment and Social Inﬂu-

ence [16]. Similarly, Accomplishment and Ownership are key drives for the Constant

Improver, as are Loss and Avoidance and Ownership for the Safety Enthusiast and So-

cial Inﬂuence and Scarcity for the Smart Saver.

Table 3. Game elements correspondence to gamiﬁcation strategies

Game

elements

Story-

telling

Goals Chal-lenges Customi-

zation

Progress Feed-back Unlocking

Levels EcoScore Missions Rank Badges

Ranking Rank Trophies Rank Rank, VCP

Search Videos Cards Cards Avatar outﬁt Cards

Points EcoScore Rank EcoScore

Virtual

Currency

Avatar outﬁt Cards Cards

Special Events Videos Badges Badges Cards

Social Area VCP Coop.

Missions

Avatar outﬁt Skills Proﬁle

Epic

Challenge

Missions Missions Missions

Customi-

zation

Avatar outﬁt Avatar outﬁt Avatar outﬁt

Access Item Videos Cards

Chat VCP

Skills EcoScore Skills Skills

Rewards Album Trophies Trophies

Album,

Trophies

Trophies Cards

117

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

117

Fig. 9. GamECAR reward system.

Fig. 10. Distribution of core drives for each phase of the game and each player type.

The main tool for customizing the game experience and stimulating the appropriate

core drives are mission objectives, which are customized according to these diagrams.

For example, a mission for the Constant Improver should challenge him to beat his/her

previous performance or invite him to exceed the scores of his/her peers. Similarly, a

player in the Safety Enthusiast should face missions that are somewhat easier than the

corresponding missions for the other two categories, but place a heavy focus on consis-

tent performance and calm/eco-friendly road behaviour. Finally, missions posed to the

Smart Saver should feature as either objective or highlight the beneﬁts of ecodriving

regarding reduced fuel consumption and positive environmental impact.

Use Cases. In this section two use cases are presented, the case of a new player and the

case of an experienced player.

1. A new user. John is a sales representative for a large pharmaceutical company. He

lives on the outskirts of Athens but travels a great deal across mainland Greece to

meet with clients and maintain his relationships. He has become increasingly in-

terested in environmental issues, every since the birth of his two children and he is

worried about the impact that pollution is having on air quality and on the environ-

ment in general. His job requires him to drive a lot but he wants to do something.

John read about GamECAR and immediately cherished the idea: he wanted to im-

prove his driving, be more fuel efﬁcient and reduce his carbon footprint. Since John

was paying for his fuel consumption, reducing his fuel bills would be great. But he

118

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

118

needed help with his driving style. How would he know if he was actually having

any impact? Also, he leads a busy life and often would be rushing around without

really thinking about the way he was driving. Having something he could switch on

at the start of each journey, something that would give him feedback and something

where he had targets he could aim for would be great. He downloaded the GamE-

CAR application and got started. First, he set himself up as a user. It was a quick

process and allowed him to create his own proﬁle and avatar.

2. The experienced user. Jaime was introduced to GamECAR by his work friend John.

They had been friends for years and really enjoyed competition. They would battle

over football and over tennis. And now they had GamECAR. Jaime spent some time

becoming familiar with the game and, like John, started to get his scores up and his

driver ranking up. He liked the way the app gave you individual targets around dif-

ferent driving skills and allowed you to track your progress. But he wanted more

engagement. After a few weeks of using the application, he opened it up one day

and was given a Loot Box. At ﬁrst, he didn’t know what it was. But once he opened

it, he started to understand the world of Knowledge Cards and Missions. Having

been driving with GamECAR for a few weeks, he already had some good knowl-

edge of what helps you to be a better, more eco-friendly driver so the Knowledge

Cards should be simple! Once he answered the cards he had been given, a Mission

opened up. It was tough.

4 Discussion and Conclusion

Current self-management systems related to ecodriving, despite their complexity and

sophisticated nature, are practically proven inefﬁcient, since they do not take into ac-

count a mixture of complex characteristics including age, sex, social status, ﬁnancial

status, physiological measurements, driving behavior measurements and vehicle char-

acteristics. The proposed system is designed to ﬁll this commitment gap by introducing

an innovative and interactive Serious game that guide targeted Driving behavior changes

to improve the way that various EcoDriving activities are undertaken so that the drivers

involved will be educated to take the desired actions and more importantly will be fa-

miliar with AR displays and applications while driving. Though GamECAR is expected

to be beneﬁcial solution in the ﬁeld of ecodriving education, the use of such systems

is not without challenges and limitations, most notably that of realism gap due to the

vast number of parameters that needs to be implemented. But they are still obstacles to

overcome, even if the system was ﬂawless there are still lots of steps to be taken before

the use of AR driver-assistance systems in real life, such as ethical issues like the ban

of using technological means while driving. Additionally, Augmented reality head up

displays (AR HUD) are not yet fully capable of coping tasks like that, the limited ﬁeld

of view and the low refresh rate are two examples of device limitations that need to be

improved. Ultimately, the AR technology has a limited target group, people that are in

early adulthood since it is challenging for someone older to get used to cutting-edge

technologies. On the other hand, the use of gamiﬁcation in our system is a favorable

way to teach teenagers how to drive using the AR drive-assistance system from the

very beginning.

119

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

119

Acknowledgments. This work has been supported by the H2020-ICT-24- 2016 IA

project GamECAR (Grant No. 732068).

References

1. Akrivopoulos, O., Chatzigiannakis, I., Tselios, C., Antoniou, A.: On the deployment of

healthcare applications over fog computing infrastructure. In: 2017 IEEE 41st Annual Com-

puter Software and Applications Conference (COMPSAC). vol. 2, pp. 288–293 (July 2017).

https://doi.org/10.1109/COMPSAC.2017.178

2. Al-shargie, F., Tang, T.B., Badruddin, N., Dass, S.C., Kiguchi, M.: Mental stress assessment

based on feature level fusion of fnirs and eeg signals. In: Intelligent and Advanced Systems

(ICIAS), 2016 6th International Conference on. pp. 1–5. IEEE (2016)

3. Al-shargie, F., Tang, T.B., Badruddin, N., Kiguchi, M.: Simultaneous measurement of eeg-

fnirs in classifying and localizing brain activation to mental stress. In: Signal and Image

Processing Applications (ICSIPA), 2015 IEEE International Conference on. pp. 282–286.

IEEE (2015)

4. Alessandrini, A., Cattivera, A., Filippi, F., Ortenzi, F.: Driving style inﬂuence on car co2

emissions. In: 2012 international emission inventory conference (2012)

5. Alessandrini, A., Orecchini, F., Ortenzi, F., Campbell, F.V.: Drive-style emissions testing on

the latest two honda hybrid technologies. European Transport Research Review 1(2), 57–66

(2009)

6. Amaxilatis, D., Akrivopoulos, O., Chatzigiannakis, I., Tselios, C.: Enabling stream process-

ing for people-centric IoT based on the fog computing paradigm. In: 2017 22nd IEEE In-

ternational Conference on Emerging Technologies and Factory Automation (ETFA). pp. 1–8

(Sept 2017). https://doi.org/10.1109/ETFA.2017.8247674

7. Barkenbus, J.N.: Eco-driving: An overlooked climate change initiative. Energy Policy 38(2),

762–769 (2010)

8. Bianchi, A., Mainardi, L., Meloni, C., Chierchiu, S., Cerutti, S.: Continuous monitoring of

the sympatho-vagal balance through spectral analysis. IEEE Engineering in Medicine and

Biology Magazine 16(5), 64–73 (1997). https://doi.org/10.1109/51.620497

9. Bianchi, G., Biton, E., Blefari-Melazzi, N., Borges, I., Chiaraviglio, L., Cruz Ramos, P.,

Eardley, P., Fontes, F., McGrath, M.J., Natarianni, L., et al.: Superﬂuidity: a ﬂexible func-

tional architecture for 5G networks. Transactions on Emerging Telecommunications Tech-

nologies 27(9), 1178–1186 (2016)

10. BMW: Eco pro (2012), http://www.bmw.com.af/com/en/insights/technology

/efﬁcientdynamics/phase 1/measures ecopro.html

11. Cai, J.F., Cand

es, E.J., Shen, Z.: A singular value thresholding algorithm for matrix comple-

tion. SIAM Journal on Optimization 20(4), 1956–1982 (2010)

12. Candes, E.J., Plan, Y.: Matrix completion with noise. Proceedings of the IEEE 98(6), 925–

936 (2010)

13. Cand

es, E.J., Recht, B.: Exact matrix completion via convex optimization. Foundations of

Computational mathematics 9(6), 717 (2009)

14. Castaldo, R., Xu, W., Melillo, P., Pecchia, L., Santamaria, L., James, C.: Detection of mental

stress due to oral academic examination via ultra-short-term hrv analysis. In: Engineering in

Medicine and Biology Society (EMBC), 2016 IEEE 38th Annual International Conference

of the. pp. 3805–3808. IEEE (2016)

15. Chen, T., Yuen, P., Richardson, M., Liu, G., She, Z.: Detection of psychological stress using a

hyperspectral imaging technique. IEEE Transactions on Affective Computing 5(4), 391–405

(oct 2014). https://doi.org/10.1109/TAFFC.2014.2362513

120

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

120

16. Chou, Y.k.: Actionable gamiﬁcation. Beyond Points, Badges, and Leaderboards (2015)

17. Ciabattoni, L., Ferracuti, F., Longhi, S., Pepa, L., Romeo, L., Verdini, F.: Real-time mental

stress detection based on smartwatch. In: Consumer Electronics (ICCE), 2017 IEEE Interna-

tional Conference on. pp. 110–111. IEEE (2017)

18. Cristea, M., Paran, F., Delhomme, P.: The role of motivations for eco-driving and social

norms on behavioural intentions regarding speed limits and time headway. World Academy

of Science, Engineering and Technology 6, 1307–6884 (2012)

19. De Santos Sierra, A., Sanchez Avila, C., Guerra Casanova, J., Bailador Del Pozo, G.: A

stress-detection system based on physiological signals and fuzzy logic. IEEE Transactions on

Industrial Electronics 58(10), 4857–4865 (2011). https://doi.org/10.1109/TIE.2010.2103538

20. del R Millan, J., Mourino, J., Franze, M., Cincotti, F., Varsta, M., Heikkonen, J., Ba-

biloni, F.: A local neural classiﬁer for the recognition of EEG patterns associated

to mental tasks. IEEE Transactions on Neural Networks (3), 678–686 (may 2002).

https://doi.org/10.1109/TNN.2002.1000132

21. DriveSim: Drivesim simulator (2014), http://drivesimsimulator.com/en/

22. Eco-Drive, Q.A.: Ecodriver (2015), http://www.play-ecodriver.ch/en/

23. Electrophysiology, T.F.o.t.E.S.o.C.t.N.A.S.o.P.: Heart Rate Variability : Standards of Mea-

surement, Physiological Interpretation, and Clinical Use. Circulation 93(5), 1043–1065 (mar

1996). https://doi.org/10.1161/01.CIR.93.5.1043

24. Fiat: eco:drive app (2014), http://ecodrive.driveuconnect.eu/

25. Ford: Smart eco driving (2013), https://ford.devpost.com/submissions/16640-smart-eco-

driving

26. Hobert, L., Festag, A., Llatser, I., Altomare, L., Visintainer, F., Kovacs, A.: Enhancements

of v2x communication in support of cooperative autonomous driving. IEEE communications

magazine 53(12), 64–70 (2015)

27. Honda: Insight eco assist system (2017), http://world.honda.com/INSIGHT/eco/

28. Kalofolias, V., Bresson, X., Bronstein, M., Vandergheynst, P.: Matrix completion on graphs.

arXiv preprint arXiv:1408.1717 (2014)

29. Kontaxis, S., L

azaro, J., Hernando, A., Arza, A., Garz

on, J.M., Gil, E., Laguna, P., Aguil

J., Bail

on, R.: Mental stress detection using cardiorespiratory wavelet cross-bispectrum. In:

Computing in Cardiology Conference (CinC), 2016. pp. 725–728. IEEE (2016)

30. Kumar, M., Weippert, M., Vilbrandt, R., Kreuzfeld, S., Stoll, R.: Fuzzy Evaluation of Heart

Rate Signals for Mental Stress Assessment. IEEE Transactions on Fuzzy Systems 15(5),

791–808 (oct 2007). https://doi.org/10.1109/TFUZZ.2006.889825

31. Lauper, E., Moser, S., Fischer, M., Matthies, E., Kaufmann-Hayoz, R.: Psychological predic-

tors of eco-driving: A longitudinal study. Transportation research part F: trafﬁc psychology

and behaviour 33, 27–37 (2015)

32. Magana, V.C., Munoz-Organero, M.: Gafu: using a gamiﬁcation tool to save fuel. IEEE

Intelligent Transportation Systems Magazine 7(2), 58–70 (2015)

33. McDuff, D., Gontarek, S., Picard, R.: Remote measurement of cognitive stress via

heart rate variability. In: 2014 36th Annual International Conference of the IEEE

Engineering in Medicine and Biology Society. pp. 2957–2960. IEEE (aug 2014).

https://doi.org/10.1109/EMBC.2014.6944243

34. Mensing, F., Bideaux, E., Trigui, R., Ribet, J., Jeanneret, B.: Eco-driving: An economic or

ecologic driving style? Transportation Research Part C: Emerging Technologies 38, 110–121

(2014)

35. Mili, G.: Low to High Frequency Ratio of Heart Rate Variability Spectra Fails to Describe

Sympatho-Vagal Balance in Cardiac Patients. Coll. Antropol 29172(1), 295–300 (2005)

36. Mitsubishi: Eco drive support (2013), http://www.mitsubishi-

motors.com/en/innovation/technology/library/eco drive.html

121

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

121

37. Mohino-Herranz, I., Gil-Pita, R., Ferreira, J., Rosa-Zurera, M., Seoane, F.: Assessment of

mental, emotional and physical stress through analysis of physiological signals using smart-

phones. Sensors 15(10), 25607–25627 (2015)

38. Mundke, R., Malewar, S., Arya, K.: Use of data recorder for driver rating. Tech. rep., SAE

Technical Paper (2006)

39. Nousias, S., Tselios, C., Bintzas, D., Orﬁla, O., Jamson, S., Mejuto, P., Amaxilatis, D.,

Akrivopoulos, O., Chatzigiannakis, I., Lalos, A.S., Moustakas, K.: Managing nonuniformi-

ties and uncertainties in vehicle-oriented sensor data over next generation networks. In: IEEE

Internaltional Conference on Pervasive computing and communications (2018)

40. Nouvelles, E.: Geco (2014), https://www.geco-drive.fr/

41. Parvaneh, S., Toosizadeh, N., Moharreri, S.: Impact of mental stress on heart rate asymmetry.

In: Computing in Cardiology Conference (CinC), 2015. pp. 793–796. IEEE (2015)

42. Reiners, T., Wood, L.C.: Gami cation in Education and Business. Springer (2015)

43. research, H., technology: Greenmeter (2018), http://hunter.pairsite.com/greenmeter/

44. Rigas, G., Goletsis, Y., Bougia, P., Fotiadis, D.I.: Towards driver’s state recognition on real

driving conditions. International Journal of Vehicular Technology 2011 (2011)

45. Salai, M., Vass

anyi, I., K

osa, I.: Stress detection using low cost heart rate sensors. Journal of

healthcare engineering 2016 (2016)

46. Saynor, B.: Ecowill deliverable 6.1: Guidelines for delivering short-duration ecodriving train-

ing courses (2011)

47. Schuller, B.W., Dunwell, I., Weninger, F., Paletta, L.: Serious gaming for behavior change:

The state of play. IEEE pervasive computing 12(3), 48–55 (2013)

48. Shaffer, F., McCraty, R., Zerr, C.L.: A healthy heart is not a metronome: an integrative review

of the heart’s anatomy and heart rate variability. Frontiers in Psychology 5(September), 1–19

(2014). https://doi.org/10.3389/fpsyg.2014.01040

49. Subaru: Ecology (2016), http://www.subaru.com.mt/eco driving.html

50. Sun, F., Kuo, C., Cheng, H., Buthpitiya, S., Collins, P., Griss, M.: Activity aware mental

stress detection using physiological sensors. Mobile Computing, Applications, and Services

76, 1–20 (2012)

51. Systems, S.S.S.: Eco driving package (2007), https://www.stsoftware.nl/

52. Taamneh, S., Tsiamyrtzis, P., Dcosta, M., Buddharaju, P., Khatri, A., Manser, M., Ferris, T.,

Wunderlich, R., Pavlidis, I.: A multimodal dataset for various forms of distracted driving.

Scientiﬁc data 4, 170110 (2017)

53. Texa: Texacare (2016), http://www.energysavingtrust.org.uk/scotland/

businesses–organisations/ transport/fuelgood-driver-training

54. Trust, E.S.: Fuelgood (2016), http://www.energysavingtrust.org.uk/scotland/

businesses-organisations/transport/fuelgood-driver-training

55. Tselios, C., Tsolis, G.: On QoE-awareness through virtualized probes in 5G networks.

In: 2016 IEEE 21st International Workshop on Computer Aided Modelling and De-

sign of Communication Links and Networks (CAMAD). pp. 159–164 (Oct 2016).

https://doi.org/10.1109/CAMAD.2016.7790351

56. Vlachos, E., Lalos, A.S., Berberidis, K., Tselios, C.: Autonomous driving in 5g: Mitigating

interference in ofdm-based vehicular communications. In: 2017 IEEE 22nd International

Workshop on Computer Aided Modeling and Design of Communication Links and Networks

(CAMAD). pp. 1–6 (June 2017). https://doi.org/10.1109/CAMAD.2017.8031619

57. Vlachos, E., Lalos, A.S., Moustakas, K., Berberidis, K.: Efﬁcient graph-based matrix com-

pletion on incomplete animated models. In: Multimedia and Expo (ICME), 2017 IEEE In-

ternational Conference on. pp. 1548–1553. IEEE (2017)

58. Wengraf, I.: Easy on the gas: The effectiveness of eco-driving. Royal Automobile Club Foun-

dation, Tech. Rep. (2012)

122

EPS Portugal 2017/2018 2017 - OPPORTUNITIES AND CHALLENGES for European Projects

122

59. WeNow: Wenow (2016), https://www.wenow.com/en/eco-driving-your-way-towards-

energy-efﬁciency/

60. Zhou, X., Yang, C., Zhao, H., Yu, W.: Low-rank modeling and its applications in image

analysis. ACM Computing Surveys (CSUR) 47(2), 36 (2015)

123

Gamiﬁcation of EcoDriving Behaviours through Intelligent Management of Dynamic Car and Driver Information

123