=Paper=
{{Paper
|id=None
|storemode=property
|title=Towards Egocentric Fuel Efficiency Feedback
|pdfUrl=https://ceur-ws.org/Vol-722/paper10.pdf
|volume=Vol-722
}}
==Towards Egocentric Fuel Efficiency Feedback==
https://ceur-ws.org/Vol-722/paper10.pdf
Towards Egocentric Fuel Efficiency Feedback
Tiago Camacho Filipe Quintal Michelle Scott Vassilis Kostakos
Ian Oakley
Madeira Interactive Technologies Institute (M-ITI)
Funchal, 9000-390, Portugal
{tiago, filipe, mscott, vassilis, ian}@m-iti.org
ABSTRACT tem failed because, effectively, the on-board equipment
Motivated by anecdotal evidence, we hypothesize that measured pure fuel consumption which in turn was in-
an egocentric approach is more appropriate and relevant tricately related to the steep terrain of the environment.
to providing fuel efficiency feedback than a systemic ap- On the other hand, drivers perceived the feedback as a
proach. In this paper we describe a proposed study to reflection of their skills.
test this hypothesis, and present the design of a fuel ef-
ficiency feedback system for public transit bus drivers. Our anecdotal experience with the public transport au-
thority’s feedback system caused us to hypothesize that
Author Keywords providing feedback on specific driver behaviour, as op-
Feedback systems, fuel efficiency, public transit bus drivers posed to overall fuel efficiency, may be a more appropri-
ate way for motivating driver behaviour change. Adopt-
ing a systemic approach to this issue, we argue that
ACM Classification Keywords existing feedback mechanisms relating to efficiency pro-
H5.m. Information Interfaces and Presentation (e.g., vide a view of the complete system, parts of which the
HCI): Miscellaneous driver has simply no way of effecting (such as the steep
terrain). Hence we argue that efficiency feedback fo-
Introduction cusing on parts of the system that the driver can ef-
In 2010 the public transport authority in Madeira, Por- fect (such as acceleration) may result in more efficient
tugal, installed onboard electronic equipment that gau- driving behaviour. We term this approach to feedback
ged driving fuel efficiency by presenting the driver with egocentric.
very simple feedback: 3 green lights progressively sug-
gested that efficiency was increasingly optimum, while In this paper we describe a fuel efficiency reporting
3 red lights progressively suggested that driving effi- and advisory system that takes advantage of the multi-
ciency was increasingly sub-optimal. The system was sensor and interactive nature of modern smart-phones
intended to give drivers feedback on their driving and to present feedback to drivers. More specifically, we
to help them achieve optimal driving efficiency. The are interested in deploying the system in public transit
result was negative: drivers complained to human re- buses to measure its effectiveness on positively influenc-
sources that the system constantly showed 3 red lights, ing drivers’ behavior. By continuously capturing real-
suggesting that their driving was bad. Human resources time sensor data, we can calculate the Vehicle Specific
complained to operations that the system was bad for Power (VSP), a surrogate variable that strongly corre-
morale. lates with both fuel consumption and pollutant emission
levels, providing a systemic view of efficiency [11]. Cru-
In response, operations attempted to “calibrate” the cially, we are able to manipulate the calculation of VSP
system by tweaking its thresholds. The result was that to ignore environment variables and provide egocentric
the feedback became useless and largely inaccurate, ul- feedback. Taking advantage of this manipulation, we
timately resulting in the abolishment of the system. In propose a study where we intend to test our hypothe-
our discussions with the transport authority, it became sis about the benefits of egocentric over systemic feed-
clear that in addition to the misinterpretation of the back. We believe that through the use of our system we
feedback by the professional drivers as a rating of their can promote not only short-term but also medium/long-
driving, the mountainous terrain of Madeira caused gen- term positive changes in public transit bus drivers’ be-
uinely inefficient driving. There was simply no way to haviours.
avoid steep hills that took a significant toll on fuel con-
sumption, thereby skewing the feedback towards inef- Related Work
ficient driving. The attempts at calibrating the sys- Research suggests that it is possible to achieve up to
15% of fuel consumption decrease when appropriate driv-
ing behavior is used [2, 6–8, 12]. Independent of contex-
Copyright � c 2011 for the individual papers by the papers’ authors. Copy- tual settings, appropriate driving behavior is character-
ing permitted only f or private and academic purposes. This volume is pub- ized by a combination of two main factors: speed and
lished and copyrighted by the editors of PINC2011.
acceleration. Specifically, it is believed that smoothness emission levels for the model to be effective.
of driving (i.e. slow acceleration levels) has a consider-
able effect on fuel consumption. Therefore, fuel effi- Devices such as smart-phones possess a wide variety of
ciency systems should be dedicated to promoting ad- sensors, like GPS and accelerometers, that enable calcu-
equate driver feedback in relation to these two essen- lation of vehicle dynamics and consequently VSP values.
tial factors, i.e., reasonable speeds and low accelera- It is then possible to approximate fuel consumption us-
tion/deceleration levels. Accurately accounting for all ing solely internal smart-phone sensors. These devices
factors that influence fuel consumption and consequent can be easily incorporated into vehicles, and their abil-
pollutant emissions can be a complex exercise. Nev- ity to provide a rich and extensible interaction platform
ertheless, And & Fwa present a possible vehicular fuel make them a feasible alternative mechanism to provide
consumption explanatory framework [1]: Physical char- drivers with fuel efficiency feedback. Furthermore, and
acteristics of the vehicle; vehicle usage and route char- comparing with usual commercial systems such as Sca-
acteristics; road characteristics; and driver’s behavior. nia Fuel-Saving Driver Support System1 , smart-phones
are not restricted to specific vehicles, and can even be
Of these factors, engine efficiency (physical characteris- device independent, which is the case when using devel-
tics of the vehicle) is considered the most important [4]. opment platforms such as Google’s Android.
Still, the driver’s attitude and behavior towards the ma-
neuvering of the vehicle can considerably impact fuel Receiving timely feedback is key to motivating behaviour
consumption levels. Therefore, it is commonly argued change, people need to be aware of their behaviour in or-
that smoothness of driving leads to higher efficiency of der to change it. Fischer found the most successful feed-
fuel consumption. back was given frequently, clearly presented, used com-
puterised tools and allowed historic or normative com-
Raw fuel consumption levels and pollutant emissions parisons [9]. Our mobile interface reflects these types of
can be calculated through the use of Portable Emis- feedback. Utilising a mobile display allows frequent op-
sions Measurement Systems (PEMS). These are con- portunities for self-reflection and should increase driver
nected to vehicles through their On-Board Diagnostic awareness of their behaviour.
(OBD) interface, letting the PEMS system access the
vehicle’s on-board computer and calculate multiple pa- Consolvo, McDonald, and Landay [3] suggest a num-
rameters [13]. Still, PEMS systems work primarily as ber of design strategies for persuasive technologies that
a diagnostic/analysis tool, not as a feedback support wish to motivate behaviour change. These strategies
mechanism. Furthermore, PEMS systems fail to reflect are based on psychological theories and recent persua-
contextual characteristics such as road gradient values. sive technology research and we have chosen to follow
It is common to augment PEMS with GPS for analysis some of their guidelines.
purposes [13].
First, we make use of abstractions rather than counting
The Vehicle Specific Power (VSP) approach is used to solely on raw data to display to drivers. Secondly, the
approximate and predict actual emissions levels and fuel data shown should be unobtrusive. This is of paramount
consumptions [10, 11]. VSP is a model that tries to ex- importance for safety reasons, as we need the mobile dis-
plain consumption and emission levels from a physical play to support ignorability and not distract the driver
perspective; it corresponds to the Power Demand or Ve- unnecessarily. Thirdly, since the data is to be presented
hicle Engine Load values, therefore correlating strongly in public, we need to present it in a way that the driver
with fuel consumption and pollutant emission levels [13]. will not feel uncomfortable if others are aware of it.
The VSP model depends on three variable factors: speed, Fourthly, we decided to ensure that only positive feed-
acceleration, and road grade. Through the combination back is given, not punishing any “bad” behavior. Con-
of these factors, along with vehicle specific air and roll cretely, we aim at rewarding possible low consumption
resistance coefficients, VSP values are calculated as fol- levels, but not use punishment for poor performance.
lows [11]: This decision is supported by the notion that positive
feedback can indeed increase intrisic motivation by af-
V SP = v ∗ (a + g ∗ sin(ϕ) + rcoef ) + acoef ∗ v 3 (1) firming competence [5]. The anecdotal evidence from
the use of a commercial system by the public transit
company also supports this notion. Finally, we have
where v is speed in m/s, a is acceleration in m/s2 , g is chosen to provide historical feedback. Doing so allows
9.807 m/s2 , ϕ is the road gradient value, rcoef is the the driver to reflect on past behaviours in order to make
rolling resistance term coefficient, and acoef is the air more informed decisions on current behaviour.
drag term coefficient. Another characteristic of VSP
is its ability to support payload modeling, especially
Research Methodology
important in situations where this value has noticeable
impact, such as is the case with public transit buses We propose an experimental approach to study to what
[11]. Still, VSP does require that we calibrate the model extent we can, through the use of egocentric feedback,
for each type of vehicle, as it is necessary to obtain 1
http://www.scania.com/media/feature-
the ground truth for fuel consumption and pollutant stories/sustainability/every-drop-of-fuel-counts.aspx
Real-time Historical User
Interface
VSP Real-time& VSP Historical &VSP Real-time
feedback
egoVSP Real-time &egoVSP Historical &egoVSP Real-time Processing
Pipeline
VSP
Acc.
Transformed 10 l 50 0.5
Raw Input Output
Component Component
Table 1. 2x2 design of combination factors
...
1 N
Historical
feedback Driver
Sensor
Data
Storage
influence public transit bus drivers driving behavior. In
our study we are interested in the following research
questions:
Figure 1. Overall view of system functionality
• Can we accurately establish driving behavior profiles
for bus drivers through the use of VSP calculations?
is flexible and extensible enough to provide support for
• To which extent can we positively influence driving any kind of vehicle.
behavior through the use of egocentric feedback tech-
niques? An overview of the architectural design is seen in Fig.
1, where the mechanism that is used to produce the fi-
• Is the use of real-time more effective than the use nal output to the driver is visible. Raw sensor data is
of historic feedback, or is a combination of the two sampled at several times per second, before it is passed
approaches most effective? to a real-time processing pipeline. This allows us to
execute tasks in parallel that may require some compu-
Consequently, and based on the previous mentioned re- tational complexity, therefore increasing system over-
search questions, we raised the following hypotheses: all speed and responsiveness. The advantage of such a
scheme becomes more evident when, for example, the
• H1. The use of the VSP surrogate variable (and its system is required to perform continuous sensor data
derivatives) allows for accurate driving profile char- integration by means of a Kalman Filter.
acterisation
• H2. The use of egocentric driver feedback improves The calculation of the vehicle dynamics and the VSP
average fuel consumption levels modeling is also included in the processing pipeline. Af-
ter exiting the pipeline, the transformed output is then
• H3. The use of real-time feedback does not signifi- fed to the feedback mechanism, which transmits spe-
cantly influence driving behavior cific information to the driver, according to the type
of feedback used. All data is continuously stored in a
To test these hypotheses we propose to develop an An- local database, so that further off-line analysis may be
droid based software to continuously collect sensor in- performed. Repeated sampling from sensors will un-
formation so that trip instantaneous parameters, such doubtedly drain the battery in its full in a matter of
as speed and acceleration, can be calculated. We will hours, so there is the need of ensuring that the device
also consider the use of additional variable(s) to model is fed continuous power by connecting it to the vehicle’s
the influence of passenger payload on the overall vehicle internal electric circuit.
weight. Then, we intend to install equipment on-board
public transit buses and calibrate the VSP model. The Drivers initiate interaction through the system’s main
ground truth establishment of instantaneous fuel con- menu (see Fig. 2). In order to use the system, drivers
sumption levels is a necessary condition for the success must register themselves before receiving a 3 digit PIN
of the VSP model. This may be achieved through the code that uniquely identifies them. Vehicles registration
use of a PEMS system or a similar mechanism. Sub- and VSP model calibration is also required to be per-
sequently, we will develop a derivative of VSP called formed, but this may be done by the developers before
egoVSP, which ignores road gradient and is defined as the system is made available to the drivers. This will
follows be the case when doing the experimental study with the
egoV SP = v ∗ (a + rcoef ) + acoef ∗ v 3 (2) public transit bus drivers. Besides the VSP model cal-
ibration, it is also possible to calibrate both the device
accelerometer, as well set up the desired orientation of
Terms of the equation are defined equally as in eq. 1. the phone inside the vehicle. This last step has some
These two fuel efficiency models, VSP and egoVSP, are limitations, as currently we are working with a phone
one of the two variables we intend to manipulate in our with only one accelerometer and no gyroscope, which
study. The other variable is the type of feedback to limits the phone’s orientation recognition. Just before
provide: real-time versus historical. Table 1 shows the starting a trip, the driver introduces his PIN code and
possible combinations of these two variables. indicates the vehicle that he is currently using. After
this the trip is marked as initiated.
Ongoing Work
As it stands, the system is a working prototype. Tar- In order to test the effectiveness of the feedback system,
geted mainly at public transit bus drivers, the system we propose using two different types of feedback: real-
ported by the European Union project INTERVIR+
and the local public transit company, Horarios do Fun-
chal, S.A.
REFERENCES
1. B.W. Ang and T.F. Fwa. Study on
fuel-consumption characteristics of public buses.
Energy, (14):797–803, 1989.
2. E.A. Barth. Evaluation of Gastell: a device to
modify driving habits, 1981.
3. S. Consolvo, D.W. McDonald, and J.A Landay.
Theory-driven design strategies for technologies
that support behaviour change in everyday life. In
Proceedings of the 27th international conference on
Human factors in computing systems - CHI ’09,
pages 405–414, New York, NY, USA, 2009.
4. J. Deccico and M. Ross. Recent advances in
automative technology and the cost-effectiveness
of fuel economy improvement. Transportation
Figure 2. System main menu Research Part D, 1(2):79–96, 1996.
5. E.L. Deci and R.M. Ryan. The support of
time and historical. In the first, we will show a real-time autonomy and the control of behaviour. Journal of
VSP graph that represents an approximation to the ac- Personality and Social Psychology,
tual VSP value. The graph is an abstract representa- 53(6):1024–1037, 1987.
tion, where it goes from green (low VSP values) to red
(high values) with an approximate quadratic function 6. Y. Ding and H. Rakha. Trip-based explanatory
increase. Additionally, actual fuel consumption, speed, variables for estimating vehicle fuel-consumption
and acceleration values are to be represented. and emission rates. Water, Air and Soil Pollution:
Focus, 2(5-6):61–77, 2002.
In regard to the historical feedback, our system will
7. E. Ericsson. Independent driving pattern factors
make available two modes to the driver. The first will
and their influence on fuel use and exhaust
show the distribution of time in the pre-defined VSP
emission factors. Transportation Research Part D,
bins, and the second will show a heat map of the route,
6:325–345, 2001.
indicating VSP “hot zones”. The use of historical feed-
back gives the driver a more broad perspective of his 8. L. Evans. Driver behavior effects on fuel
driving behavior, as it recalls and identifies potential consumption in urban driving. Human Factors,
patterns that may be improved. Furthermore, histori- 21(4):389–398, 1979.
cal feedback will only take place when the driver is not
actively driving. 9. C. Fischer. Feedback on household electricity
consumption: a tool for saving energy? Energy
Conclusion
Efficiency, 1:79–104, 2008.
In this paper we have argued that egocentric feedback 10. HC Frey and NM Rouphail. Speed-and
on fuel efficiency can be more effective than systemic Facility-Specific Emission Estimates for On-Road
feedback on motivating driving behaviour change. Mo- Light-Duty Vehicles on the Basis of Real-World
tivated by anecdotal evidence, we hypothesise that an Speed Profiles. Transportation Research Record,
egocentric approach is more appropriate and relevant. pages 0–21, 2006.
By re-defining the VSP surrogate metric, we are able to
switch between systemic and egocentric feedback while 11. HC Frey, NM Rouphail, Haibo Zhai, TL Farias,
maintaining minimal changes between our experimen- and GA. Comparing real-world fuel consumption
tal conditions. Orthogonal to the manipulation of the for diesel-and hydrogen-fueled transit buses and
efficiency model, we describe our interest in testing the implication for emissions. Research Part D:,
effect of instantaneous versus historic data in the feed- 12:281–291, 2007.
back system. 12. M.H.L. Waters and I.B. Laker. Research on fuel
conservation for cars, 1980.
Acknowledgments
This work is supported by the Portuguese Foundation 13. L. Yu, Z. Wang, and Q. Shi. PEMS-Based
for Science and Technology (FCT) grant CMU-PT/ HU- Approach to Developing and Evaluating Driving
MACH/ 004/ 2008 (SINAIS). This work is also sup- Cycles For Air Quality Assessment, 2010.