Socio-cyberphysical System for Proactive Driver Support

Approach and Case Study

Alexander Smirnov

1,2

, Nikolay Shilov

1,2

and Oleg Gusikhin

SPIIRAS, 39, 14 Line, St. Petersburg, Russia

ITMO University, 49, Kronverkskiy pr., St. Petersburg, Russia

Ford Motor Company, 2101, Village Rd., Dearborn, Michigan, U.S.A.

Keywords: Infomobility, Information Driver Support System, Proactive Collaborative Recommendation System,

Service Fusion, Parking Assistance.

Abstract: Recent developments in the areas of decision support, data and decision mining, on-board infotainment

systems have produced valuable results that can be used to support people in different aspects of their lives.

Infomobile driver support is one of the possible applications of these, what can significant increase the

quality of the user experience. The paper presents a developed approach and enabling technologies for

implementation of an intelligent driver support system that takes advantages provided by such modern

developing technologies as context-based collaborative recommendation systems, proactive information

support, smart space, and V2V communication. The developed concept is illustrated via a parking assistance

scenario.

1 INTRODUCTION

Recent developments in the areas of decision

support, data and decision mining, recommendation

systems have produced valuable results that can be

used to support people in different aspects of their

lives.

Cyberphysical systems tightly integrate

heterogeneous resources of the physical world and

IT world. Socio-cyberphysical systems go

significantly beyond the ideas of the current progress

in cyber-physical systems, socio-technical systems

and cyber-social systems to support computing for

human experience (Sheth, et al., 2013). They tightly

integrate physical, cyber, and social worlds based on

interactions between these worlds in real time. Such

systems rely on communication, computation and

control infrastructures commonly consisting of

several levels for the three worlds with various

resources as sensors, actuators, computational

resources, services, humans, etc. (Teslya, et al.,

2014).One of promising tasks is integration of

different mobile applications with on-board

infotainment systems.

There are exist various techniques aimed at

driver support based on the analysis of information

from various devices and sensors. Those, which are

based on the information accumulated within one

car, are commercially available (e.g., parking

assistance systems). Cyberphysical networks provide

for extended possibilities in this area. Integration of

several nearby cars with their sensors into one

cyberphysical network makes it possible to increase

the quality of situation detection (e.g., sharing

information about free parking slots) and to provide

for certain situation development prediction (sharing

information about parking slots that are currently

being occupied or will be in the nearest future). The

concept of socio-cyberphysical systems adds one

more dimension – humans (drivers in this particular

example). Analyzing drivers’ needs, preferences and

intentions could significantly improve the situation

detection and situation development mechanisms.

Such systems can be classified as infomobile

driver support assuming distribution of dynamic and

selected multi-modal information to the users, both

pre-trip and, more importantly, on-trip (Ambrosino

et. al., 2010). It is a new way of service organization

appeared together with the development of personal

mobile and wearable devices capable to present user

multimodal information at any time. Recent

advances in car on-board infotainment systems make

it possible to organize infomobile driver support.

Configuration of socio-cyberphysical systems

belonging to the class of variable systems with

289

Smirnov A., Shilov N. and Gusikhin O..

Socio-cyberphysical System for Proactive Driver Support - Approach and Case Study.

DOI: 10.5220/0005540702890295

In Proceedings of the 12th International Conference on Informatics in Control, Automation and Robotics (ICINCO-2015), pages 289-295

ISBN: 978-989-758-123-6

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

dynamic structures is a very complex task. Their

resources are too numerous, mobile with a

changeable composition. However, taking into

account not only combination of information from

cars, including speed, location, free parking spots,

directions (from the navigation system, etc.) but also

application of behavior analysis techniques for

predicting future (both short term [few seconds] and

long term [minutes-hours]) actions of drivers (some

drivers might prefer to park “next to the door” even

if it is expensive, others prefer to have a walk and

save on parking fee; some might be seeking for a

parking spot, others are about to leave, etc.) could

significantly improve the efficiency of parking

situation prediction and consequently improve it via

regulation of its controllable components.

In dynamic environments correct decisions can

only be made in the right context related to the

current situation (Smirnov, et al., 2010; Smirnov, et

al., 2005). Context is any information that can be

used to characterize the situation of an entity where

an entity is a person, place, or object that is

considered relevant to the interaction between a user

and an application, including the user and

applications themselves (Dey, et al., 2001). Thus,

context-driven decision support is required in

situations happening in dynamic, rapidly changing,

and often unpredictable distributed environments

such as roads.

The paper proposes an approach that has a

service-oriented architecture. The services are

integrated through service fusion, originating from

the concept of knowledge fusion, which implies a

synergistic use of knowledge from different sources

in order to obtain new information (Smirnov, et al.,

2003; Smirnov, et al., 2015). Thus, service fusion in

this work can be defined as synergistic use of

different services to have new driver support

possibilities not achievable via usage of the services

separately. Context-based service fusion can provide

a new, previously unavailable level of personalised

on-board information support via finding

compromise decisions taking into account proposals

of various services and driver preferences.

2 CASE STUDY SCENARIO

The following scenario can be considered as the

basis for the case study. A driver has a meeting

scheduled for 3pm. The navigation system leads the

driver to the meeting place. The driver’s profile has

information that the driver prefers free parking and

does not mind to have a walk for 200-300 meters.

Figure 1: Parking spot information sharing: the car doesn’t

fit into the free parking spot but it shares this information

so that other cars can use the spot.

Analysing the information about available parking

places nearby, as well about which parking spots

are currently free / will be free in few minutes, the

system proposes a parking spot to the driver.

The required information can be acquired from a

number of sources. Usually, there can be several

independent parking structures nearby with their on-

line services that provide availability (each parking

spot might have a sensor indicating if it is busy or

free), price, and wait time. The municipal street

parking can also provide a service with

corresponding parameters such as parking time limit

(e.g., 30 minutes or two hours, and the parking

enforced from 7:00 am till 19:00). There can also be

free street parking in vicinity. The availability of this

free parking may be provided by 3rd party service

that can provide a probability of finding a spot in the

given block at the given time. Alternatively, the

availability can be shared by other cars searching for

parking spots, leaving, or just passing by (Figure 1).

Free parking spots can be estimated via such

systems as Active Park Assist, which are already

available in the market. The parking facilities might

also have customer ratings stored in a social

network.

To lead the driver to the chosen parking spot the

system should be integrated with a navigation

service, which would calculate the route (e.g., going

through several possible parking spots) takin into

account the driver’s schedule and the context of the

current situation (traffic, weather conditions, etc.).

Such distributed system is reasonable to be

implemented based on the service-oriented

architecture. The developed approach to

configuration of a service-based socio-cyberphysical

network is presented in the next section. Section 4

addresses the issue of proactive driver support,

which is essential because the driver has very

limited possibilities to input requests and checking

all possible solutions. Such systems require behavior

analysis techniques described in section 5.

Searching

for a free

parking

spot

Found a free

parking spot 4.5 m

long at

[

coordinates

]

Smart space or

V2V

communication

ICINCO2015-12thInternationalConferenceonInformaticsinControl,AutomationandRobotics

290

A developed research prototype is presented in

section 6. Major results are summarized in the

conclusion.

3 PROPOSED APPROACH

The proposed approach to configuration of socio-

cyberphysical network assuming negotiation of

socio-inspired services. The approach is based on

the following principles: self-management and

responsibility of the services, and decentralization

with network organization (without any social

hierarchy of command and control) and co-operation

between services. This is motivated by the fact that

in order to operate efficiently, the system has to

model driver behavior (taking into account

preferences, earlier made decisions; see section 5),

what is achieved due to the usage of socio-inspired

services.

Self-organization of services is considered as a

threefold process of (i) cognition (where subjective

context-dependent knowledge is produced),

(ii) communication (where system-specific objecti-

fycation or subjectification of knowledge takes

place), and (iii) synergetical co-operation (where

objectified, emergent knowledge is produced). The

Individually acquired context-dependent (subjective)

knowledge is put to use efficiently by entering a

social co-ordination and co-operation process. The

objective knowledge is stored in structures and

enables time-space distanciation of social

relationships.

In order to achieve the realism and dynamics of

the self-organizing system, its components (services)

have to be creative, knowledgeable, active, and

social. The services that are parts of a system

permanently change their joint environment what

results in a synergetic collaboration and leads to

achieving a certain level of collective intelligence.

This is also supported by the fact that individual

service behavior is partially determined by the social

environment the services are contributing to (called

“norms”). For this purpose a protocol has been

developed based on the BarterCast approach

(Seuken, et al., 2014) that originates from the idea of

building a network by a service representing all

interactions it knows about.

The overall scheme of the approach is shown in

Figure 2. The lower part of the figure represents the

socio-cyberphysical network and the corresponding

service network. The upper part represents the socio-

inspired service modelling behavior of a member of

the socio-cyberphysical network. The detailed

Figure 2: Socio-cyberphysical network configuration

approach overview.

description of the service is given in (Smirnov, et al.,

2013). The interoperability at the technological level

is provided via usage of common standards and

protocols, the interoperability at the level of

semantics is provided via usage of common

semantics and terminology described via ontologies

(please, see Smirnov, et al., 2007b for a detailed

description of the interoperability support).

4 PROACTIVE

RECOMMENDATIONS

Recommendation systems are widely used in the

Internet for suggesting products, activities (including

tourism), etc. for a single user considering his/her

interests and tastes (e.g., Garcia, et al., 2009). The

collaborative / group recommendation systems try to

find users who share similar interests with the given

user and recommend items they choose to that user

(Kumar and Thambidurai, 2010). Сontext-driven

collaborative algorithms of recommendation

generation increase the quality and speed of decision

making due to taking into account not only

previously made decisions but also on the contexts

of situations in which the decisions were made.

Modern recommendation systems mostly work

on the scheme “request-response”; the active party

Service

network

Service

Socio-

cyberphysical

network

Relationship

Correspondence

Information flow

Network membe

Profile

Environment

New knowledge,

norms, policies

Activity

Self-contextualization

Self-management

Subjective

(contextualized)

knowledge

Socio-cyberphysicalSystemforProactiveDriverSupport-ApproachandCaseStudy

291

in the interaction is the user, who makes a request,

which can contain some additional constraints for

the alternative solutions. As a response, the

recommendation system offers a list of

recommendations (recommended solutions). In on-

board recommendation systems, users cannot

browse through many search results and suffer from

other restrictions in the user experience, because of

user interface limitations such as small display sizes

or missing keyboards. In such environments, having

the user not to submit any request or query to get a

recommendation could possibly improve the user

experience (Bader, 2013; Woerdnl, et al., 2011).

At the moment, there are hardly any systems that

could offer recommendations in the proactive way,

without user request, on the basis of the current

situation and user profile analysis (Ricci, 2011). One

of the reasons of this situation is the high risk of

obtrusive offering non-relevant information to the

user. However, modern mobile devices equipped

with various sensors, make it possible to produce

proactive recommendations, which would be useful

and convenient due to creation of more precise user

behavior models.

Appearance of mobile devices caused a more

intensive development of proactive systems. The

next step in the development of proactive systems is

appearance of systems based on prediction of the

user behavior, his/her future locations or actions, as

well as situation recognition. Today, integration of

proactive systems with recommendation systems can

be considered as a perspective research field.

5 BEHAVIOR ANALYSIS

The developed approach assumes description of

functionality, preferences and strategies of the socio-

cyberphysical network members via updatable and

extendable profiles. Usage of the profiles makes it

possible to “individualize” the proactive

recommenddations. For this reason methods of

human preferences revealing have been developed.

The preferences are revealed via the analysis of

the situations the network member faces most often,

parameters of objects and actions most often

occurring or avoiding in the decisions (actions)

made by the network member, optimization criteria

the network member most often follows or not. One

of the main features of the developed profile model

is presence of the information related to antecedents

and consequences of the made decisions and under-

taken actions what makes it possible to perform the

functional analysis of the human behavior.

The functional behavior analysis is one of the

behavior analysis techniques considering frequency

of key behavior events related to certain human

activity (Kraus, 1995). It is also known as ABC

analysis (antecedent, behavior, consequence) and is

based on identification of both antecedents and

consequences of the behavior. As a result, it is

possible to build a conditional behavior model,

which would let one know (to predict) how a human

(e.g., a driver or a pedestrian) would act in a given

situation. For example, the research of application of

this technique to the driver behavior prediction has

resulted in some positive results (Taniguchi, et al.,

2012).

The result of such an analysis produces typical

decisions (actions) made by the considered person in

certain situations (behavior patterns). Example of

behavior pattern is presented below:

 Context: the traffic is heavy; the traffic in the

lane in the left moves a bit faster; the driver is

hurrying.

 Antecedent: there is a traffic congestion ahead

in the lane; the vehicle ahead slows down.

 Possible behavior: stay in the lane and slow

down; switch to the left lane.

 Preferred behavior: switch to the left lane.

 Consequence: the vehicle moves faster than

vehicles in the congested lane, but slower than it

moved before; vehicles in the left lane behind

slow down.

The behavior pattern revealing techniques used

in the proposed approach include:

1. Revealing human behavior patterns for problems

with the same structure but different parameters.

In this case, the structural knowledge constituent

will be the same, and the parametric knowledge

constituent will be different.

2. Revealing human behavior patterns for different

problems solved by the same person. This

technique assumes analysis of structures of

different problems trying to find similarities

associated with the same decisions / actions.

3. Revealing human behavior patterns based on the

optimization criteria (problem parameters with

highest or lowest values) the person tends to

follow or avoid (e.g., the driver prefers moves

faster or with less maneuvers). Aggregated (e.g.,

weighted average) criteria can also be analyzed.

4. The above techniques applied not to one person

but to different persons with similar profiles.

This technique utilizes collaborative filtering

mechanisms (Schafer, et al., 2007).

ICINCO2015-12thInternationalConferenceonInformaticsinControl,AutomationandRobotics

292

To implement the first three techniques the

following methods have been developed:

1. Decision / action clustering method. The

decisions made by the person and actions

undertaken are grouped into clusters. Based on

the clusters built the common properties

(parameters) of the problems and decisions /

actions grouped into one cluster are identified.

The results of this method can be refined if there

is enough historical data accumulated and

clustering can be done taking into account the

context of the situation when corresponding

decisions have been made (including and

preferences of the person at the moment of

decision making as well as information about

behavior antecedents and consequences).

2. The alternative analysis method. Unlike the

previous method searching for similar person’s

decisions, this method is aimed at the analysis of

differences between decisions made by the

person and actions undertaken. Based on the

analysis of the identified differences taking into

account the situation context (as well as

preferences of the person and information about

behavior antecedents and consequences) namely

definition of the main generic differences of the

made decisions, the behavior patterns are

revealed.

To implement the fourth technique of human

behavior pattern revealing, a method based on the

collaborative filtering mechanisms used for building

collaborative recommendation systems. This

technique would enable to predict human behavior

even in situations, in which this person has never

got. For this reason, the decisions made by persons

with similar properties are used.

Application of the above techniques would

enable to generate proactive recommendations based

on prediction of behavior of real people (e.g., via

usage of opportunistic planning (Hayes-Roth, 1980)

mechanisms).

6 PROTOTYPE DESCRIPTION

The developed research prototype has a service-

oriented architecture based on the usage of the smart

space concept implemented in the Smart-M3

platform. The smart spaces technology (Balandin

and Waris, 2009; Korzun, et al., 2013) aims at the

seamless integration of different devices by

developing ubiquitous computing environments,

where different services can share information with

each other, perform computations, and interact with

each other for joint task solving. A detailed

description of the developed smart space-based

architecture can be found in (Smirnov, et al., 2014b).

The interaction between services is presented in

Figure 3. It is based on usage of AppLink for

interaction with the vehicle. In addition to the

information already stored in the services

(associated databases, user settings, revealed

preferences, etc.), they acquire the following

information from other services, namely:

 Local road infrastructure provides information

about parking places, their restrictions and

prices.

 Nearby cars share information, which parking

spots are currently free and occupied, will be free

if the car is about to leave, are about to be

occupied if the corresponding driver is going to

park at the selected spot.

Figure 3: Service interaction example.

AppLink SmartPhone Driver profile

Road

infrastructure

GPS, route

Schedule

Parking locations

Recommended parking spot

Car Driver Cloud

Recommendat

ion Service

Other cars

On-board

information

system

Free parking spots, intentions

GPS, route

Free

arkin

ots, intentions

Preferences, intentions

Socio-cyberphysicalSystemforProactiveDriverSupport-ApproachandCaseStudy

293

Figure 4: Example of In-vehicle information support implemented in AppLink emulator.

 Weather acquisition service provides information

about the current and forecasted weather

conditions.

 Recommendation service obtains driver’s

schedule from his/her smartphone to estimate

current time restrictions, predefined driver

preferences and information obtained from the

above mentioned sources.

The generated solutions are transferred to the

AppLink screen so that the driver could choose the

most appropriate one, and to the in-car navigation

system (Figure 4). If parking can be paid online, the

payment can be done automatically, when the driver

is parking.

7 CONCLUSIONS

The paper presents a developed approach and

enabling technologies for implementation of an

intelligent driver support system that takes

advantages provided by such modern developing

technologies as context-based collaborative

recommendation systems, proactive information

support, smart space, and V2V communication. The

developed concept is illustrated via a parking

assistance scenario.

The work is at an early stage of development.

The paper proposes generic solutions for the key

problems that may arise during the implementation

of the proposed system. Particular methods and

models supporting these solutions are subjects of the

future research.

ACKNOWLEDGEMENTS

The research was partially supported by projects

funded by grants # 15-07-08092, # 15-07-08391,

# 13-07-00271, # 13-07-12095, # 14-07-00345 and

of the Russian Foundation for Basic Research,

project 213 (program 8) of the Presidium of the

Russian Academy of Sciences, and project # 2.2 of

the basic research program “Intelligent information

technologies, system analysis and automation” of the

Nanotechnology and Information technology

Department of the Russian Academy of Sciences.

This work was also partially financially supported

by Government of Russian Federation, Grant 074-

U01.

REFERENCES

Ambrosino, G., Nelson, J. D., Bastogi, B., Viti, A.,

Romazzotti, D., Ercoli, E., 2010. The Role and

Perspectives of the Large-Scale Flexible Transport

Agency in the Management of Public Transport in

Urban Areas. In: Ambrosino, G., Boero, M., Nelson, J.

D., Romanazzo, M. (Eds), Infomobility Systems and

Sustainable Transport Services, ENEA, 156-165.

(2010).

Bader, R., 2013. Proactive Recommender Systems in

Automotive Scenarios. PhD Theses. Technische

Universitat Munchen, Fakultat fur Informatik

Lehrstuhl fur angewandte Informatik / Kooperative

Systeme.

Balandin, S., Waris, H., 2009. Key Properties in the

Development of Smart Spaces. In Proc. of the 5th Int’l

ICINCO2015-12thInternationalConferenceonInformaticsinControl,AutomationandRobotics

294

Conf. Universal Access in Human-Computer

Interaction, Springer-Verlag, 3-12.

Dey, A. K., Salber, D., Abowd, G. D., 2001. A Conceptual

Framework and a Toolkit for Supporting the Rapid

Prototyping of Context-Aware Applications. Context-

Aware Computing, A Special Triple Issue of Human-

Computer Interaction, 16, 229-241.

Garcia, I., Sebastia, L., Onaindia, E., Guzman, C., 2009. A

Group Recommender System for Tourist Activities.

EC-Web 2009: Proceedings of E-Commerce and Web

Technologies, The 10th International Conference

(2009), LNCS, Springer, 5692, 26-37.

Hayes-Roth, B., 1980. Human Planning Processes.

Scientific Report (1980),

http://www.rand.org/content/dam/rand/pubs/reports/20

07/R2670.pdf.

Korzun, D., Balandin, S., Gurtov, A., 2013. Deployment

of Smart Spaces in Internet of Things: Overview of the

Design Challenges. Internet of Things, Smart Spaces,

and Next Generation Networking, Springer, LNCS,

8121, 48-59.

Kraus, S. J.: Attitudes and the Prediction of Behavior: A

Meta-Analysis of the Empirical Literature. Personality

and Social Psychology Bulletin, 21(1), 58-75.

Kumar, A., Thambidurai, P., 2010. Collaborative Web

Recommendation Systems Based on an Effective

Fuzzy Association Rule Mining Algorithm (FARM).

Indian Journal of Computer Science and Engineering,

Engg Journals Publications, 1(3), 184-191.

Ricci, F., 2011. Mobile Recommender Systems. J. of IT &

Tourism, 1, 205-231.

Schafer, J. B., Frankowski, D., Herlocker, J., Sen, S.,

2007. Collaborative Filtering Recommender Systems.

The Adaptive Web, Springer Berlin Heidelberg, 291-

324.

Seuken, S., Meulpolder, M., Parkes, D. C., Pouwelse, J.

A., Tang, J., Epema, D. H. J., 2014. Work Accounting

Mechanisms: Theory and Practice. Working Paper.

Department of Informatics, University of Zurich.

Sheth, A. P., Anantharam, P., Henson, C. A., 2013.

Physical-Cyber-Social Computing: An Early 21st

Century Approach. IEEE Intell. Syst., 28(1), 78-82.

Smirnov A., Pashkin M., Chilov N., Levashova T., 2003.

KSNet-Approach to Knowledge Fusion from

Distributed Sources. Computing and Informatics. 22,

105-142.

Smirnov, A., Pashkin, M., Levashova, T., Chilov, N.,

2005. Fusion-Based Knowledge Logistics for

Intelligent Decision Support in Network-Centric

Environment. In Klir, G. J. (Ed.) International Journal

of General Systems, Taylor & Francis, 34(6), 673-690.

Smirnov, A., Shilov, N., Levashova, T., Sheremetov, L.,

Contreras, M., 2007. Ontology-Driven Intelligent

Service for Configuration Support in Networked

Organizations. Knowledge and Information Systems.

Special Issue on Current Challenges and Trends in

Intelligent Computing and Knowledge Management in

Industry, Springer London, 12(2), 229-253.

Smirnov A., Levashova T., Shilov N., Kashevnik A.,

2010. Hybrid Technology for Self-Organization of

Resources of Pervasive Environment for Operational

Decision Support.

International Journal on Artificial

Intelligence Tools, World Scientific Publishing

Company, 19(2), 211-229.

Smirnov, A., Kashevnik, A., Shilov, N., Makklya, A.,

Gusikhin, O., 2013. Context-Aware Service

Composition in Cyber Physical Human System for

Transportation Safety. Proceedings of the 13

International Conference on ITS Telecommunications

(ITST2013), IEEE, 139-144.

Smirnov, A., Kashevnik, A., Shilov, N., Gusikhin, O.,

2014. Context-Driven On-Board Information Support:

Smart Space-Based Architecture. The 6

International

Congress on Ultra Modern Telecommunications and

Control Systems (ICUMT 2014), IEEE, 195-200.

Smirnov, A., Levashova, T., Shilov, N., 2015. Patterns for

Context-Based Knowledge Fusion in Decision

Support. Information Fusion, 21(1), 114-129.

Taniguchi, T., Nagasaka, S., Hitomi, K., Chandrasiri, N.

P., Bando, T., 2012. Semiotic Prediction of Driving

Behavior Using Unsupervised Double Articulation

Analyzer. Intelligent Vehicles Symposium (IV), IEEE,

849-854.

Teslya, N., Smirnov, A., Levashova, T., Shilov, N., 2014.

Ontology for Resource Self-Organisation in Cyber-

Physical-Social Systems. Klinov, P., Mouromtsev, D.

(eds.) 5

International Conference of Knowledge

Engineering and the Semantic Web (KESW 2014),

CCIS, Springer International Publishing Switzerland,

468, 184-195.

Woerdnl, W., Huebner, J., Bader, R., Gallego-Vico, D.,

2011. A Model for Proactivity in Mobile, Context-

Aware Recommender Systems. Proceedings of the

fifth ACM conference on Recommender systems, 273-

276.

Socio-cyberphysicalSystemforProactiveDriverSupport-ApproachandCaseStudy

295