A Wizard-of-Oz Experiment To Demonstrate Water . - ASME

3y ago
1.36 MB
12 Pages
Last View : 7d ago
Last Download : 5m ago
Upload by : Matteo Vollmer

Proceedings of the ASME 2019International Design Engineering Technical Conferencesand Computers and Information in Engineering ConferenceIDETC/CIE2019August 18-21, 2019, Anaheim, CA, USADETC2019-98468A WIZARD-OF-OZ EXPERIMENT TO DEMONSTRATE WATER REDUCTIONAND USER TRAINING WITH AN "AUTONOMOUS" FAUCETWilliam JouStanford UniversityStanford, CA, USASamantha M. BeaulieuStanford UniversityStanford, CA, USAAdrienne K. LimStanford UniversityStanford, CA, USAErin F. MacDonaldStanford UniversityStanford, CA, USAABSTRACTResource-conserving products and commercial smartproducts abound in the market, but the intersection of the two islargely unexplored from the human-centered-designcommunity. Research has shown that people (users) havedifferent cognitive styles that influence their methods ofapproaching challenges and how they interpret the world.Utilizing this knowledge of cognitive styles, energy conservingproducts could (1) reduce resource consumption of its users and(2) increase user satisfaction with interacting with thoseproducts. Passive products—such as a flow-limitingshowerhead—do not seek to change the user behavior andsolely change the behavior of the product to conserve water. Inthis work, we design and test an "active smart" product to see ifit can change users through product interaction. A customfaucet was designed and built to conduct an experiment withthe Wizard of Oz (WoZ) technique of remotely operating adevice to create the impression of autonomy/smartness.Participants were asked to wash multiple sets of dishes to testif: (1) participants use less water when washing dishes with asmart faucet and (2) participants remember this behaviorchange and use less water in a alter interaction with a normalfaucet. Results confirmed the hypotheses and showed that thoseinteracting with the faucet reduced their consumption by 26.5%during WoZ treatment and, importantly, 10.9% while washingafter interacting with the WoZ treatment. Limitations includethe implementation of the smart algorithm and the willingnessto-pay for a smart faucet in the home. This study demonstratesthat smart products can conserve resources and train for furtherconservation even when the user is not using the smart product.1. INTRODUCTION1.1 MotivationIn the first six months of 2018, households consumed over20% of the United States’ total energy usage, making theresidential sector the third highest contributor [1]. However,despite knowing that reducing energy consumption could leadto significant monetary savings, users are still reluctant tochange habits [2, 3]. Studies have shown that attitudes towardconservation do not influence consumer behavior. Therefore,energy saving displays equally affected everyone, regardless ofwhether they identified as active energy savers [4, 5, 6].Faucets alone compose over 15% of household waterusage. Given the potential savings that could come fromreducing water usage at the sink, smart faucets are a potentialproduct that could dramatically benefit users and theenvironment. Typical automatic faucets also consume morewater than manually operated faucets [7, 8, 9]. The studypresented here suggests that the proposed faucet can train usersto reduce water consumption by adapting and modifying userbehavior.1.2 BackgroundWhile commercial smart products today, such as the Nestthermostat, may learn about the preferences of a user, they donot learn about how a user actually thinks [10]. Current smartproducts assume that all people are fundamentally the same, butresearch has shown that people have cognitive styles thatgovern how they approach different tasks and interact with theworld [11], for example, someone might be more analytical,1Copyright 2019 ASME

while someone else may be more intuitive. We are interested incognitive styles because we believe that interacting with aproduct that can understand the user improves trust, whichwould in turn open an avenue to training users to adoptenvironmentally friendly behaviors. For example, a smartfaucet that better understands its users could potentiallyleverage that knowledge to train users to reduce waterconsumption or lower the average water temperature. Ourmotivation here is to take the framework of leveragingcognitive states to develop physical “telepathic” products thatbetter understands its users. Through various studies, we founda smart faucet as a potential “telepathic” product that coulddramatically benefit users and the environment. Studies byMayer et al. that showed that faucets compose of 15% ofhousehold water usage [12] and Gauley and Koeller whichrevealed that typical automatic faucets use more water thanregular faucets [13].Exploratory studies by Ramaswamy and MacDonald foundthat parameters of a user’s cognitive style such as patience,temperature sensitivity, and resource consciousness affectedhow they used faucets [14]. Other studies show that cognitivestyles can be utilized to encourage pro-environmental behavior.For example, by considering different behavioral groupings ofindividuals, one study shows that policy makers will have moresuccess promoting their energy-conservation initiatives [15]. Astudy by MacDonald and She identifies seven cognitiveconcepts to incorporate into eco-products to influenceconsumer behavior: responsibility, complex decision-makingskills, decision heuristics, the altruism-sacrifice link, trust,cognitive dissonance, and motivation [16]. MacDonald thenproposes specific recommendations, which we incorporatedinto our study. One of these recommendations is that trust canbe instilled into a product that is similar to the user. Thissimilarity can be in the form of physical or personality traitsthat resemble the user’s [17]. By understanding and adapting tothe user’s cognitive style, our faucet will be able to form thistrust between user and product.This paper explores what the physical implementation ofsuch a system could achieve in terms of user behavior changeand real-life water conservation. For even if such a systemcould theoretically work, we wanted to first answer thequestions: (1) Would such a system save water? (2) Would therebe any residual characteristics shown by participants afterinteracting a smart faucet? (3) Would participants enjoy theexperience and consider bringing such a product home?To answer the questions we had, we built a custom faucet –depicted in Figure 1 – to conduct a human experiment. Thefaucet was initially used for pilot testing conducted byRamaswamy and MacDonald [14] and has since been modifiedto allow for remote control of the temperature and flow of thewater as well as water consumption tracking. While otherstudies solely rely on feedback from participants in the forms ofenergy surveys or energy-flow limiting devices, we sought tocreate a real-time interactive product to elicit true participantresponses of a perceived interaction with an autonomous devicethat “understands” them.FIGURE 1. MODIFIED PHYSICAL FAUCET FOR COMPUTERCONTROL2. LITERATURE REVIEW2.1 Water ConservationThere have been numerous studies conducted oneducational or behavioral methods for encouraging waterconservation. Flack et. al. tried to implement a policy-relatedapproach to enforce installation of water-efficient fixtures;however, they found it difficult to enforce long-term. Theyconducted a survey of 19 communities, in which they enforceda variation of water-conservation methods: some communitieswere metered for the amount of water they used, while othershad their water use restricted to certain days of the week andhours of the day. Other communities were given plumbingfixtures with water-saving devices. Flack et. al. found that,while water-saving devices were feasible, they were difficult toenforce. Participants were supportive of the idea, but they werereluctant to introduce new devices to their homes [18].Other studies attempted to use feedback techniques toconvey how much water was being consumed in real time.These studies found that immediate feedback was moreimportant for changing consumers’ behaviors than long-termfeedback, such as a bill at the end of the month, as shown in astudy by Chetty et. al. [19, 20]. One study by Kuznetsov et. al.utilized an in-shower LED display to indicate water usage inreal time. The display is green when the water is first turned on,yellow if the water remains on for more than the previouslymeasured average duration, and red once the water has beenrunning for longer than one standard deviation above average.If the water is left running for a significantly long period oftime, the red light starts to flash. This feedback proved usefulwhen the device was present, but it was ineffective in changinglong-term behavior as participants did not maintain waterefficient habits after the device was removed [21].Meanwhile, numerous studies have attempted to comparedifferent methods aimed at encouraging water saving behavior.2Copyright 2019 ASME

A study by Aitken and McMahon found that cognitivedissonance, when combined with feedback about water usage,helped reduce water consumption in the short term whencompared to feedback alone [22]. Other studies found thatwater efficient devices, such as flow-limiting faucets, weremore effective than other techniques such as educationmaterials and real time feedback displays. Hopp et. al. foundthat low-flow shower heads, combined with flow-limitingfaucets and dual-flush toilets could save at most about 114gallons of water per day for a family of four [23]. Studies alsofound that using devices marked as water-efficient could alsolead to significant savings [24]. However, Geller et. al. findsthat these devices do not promote any change in behavior anddidn’t save as much water as expected based on manufacturingratings. Therefore, they speculate that the installation of watersaving devices, such as flow limiting faucets, might have ledresidents to be less conscious of their water usage tocompensate for presumed savings [25].difficult to understand, reporting that many of its features donot operate as expected and the system could not understandthe intent behind an individual’s behavior [10].2.2.2 Active Products Meanwhile, active products, suchas the smart home energy efficiency devices presented by Jahnet. al, aim to train users to better use certain resources. Thesystem implemented by Jahn et. al. connects various devicesand appliances within a home to coordinate energy savings.Their system incorporates feedback from multiple devices,which provides users with a visual of their energy usage, andhow the system is adapting. This added awareness andknowledge empowers users to identify where and how theywaste the most energy and change their habits to save moreenergy [32].Eco-feedback devices represent a large subset of activeproducts in the energy realm. These devices operate on the ideathat educating people on their energy consumption throughactive displays will make them more aware of how their actionsimpact the environment and, therefore, encouraging behaviorchange. However, there are few that explore or measure thebehavior change aspect [33]. One study that aims to understandthe saliency affect of eco-feedback devices, conducted byLynham et. al., conducted a three phase experiment, in whichthree groups have their electricity consumption measured for 30days. While group 1 acts as a control, having their electricityuse monitored for all three periods, the two experimentalgroups receive an in-home display (IHD) that gives real timefeedback on electricity consumption for 30 days. Finally, one ofthe two experimental groups, the continued treatment group,keeps the IHD while the other, the discontinued treatmentgroup, has theirs removed for a final 30 days. While there wassome learning effect briefly after the IHD is removed, itdeclined over time. The study found that while knowledge andunderstanding of electricity consumption increased over longperiods of time, the IHD did not change the participant’s habits[34].A review of thirty-eight studies aimed at household energyconservation by Abrahamse et. al., shows that certaintechniques have proved more successful at promoting userbehavior change than others. While some methods, likeproviding the user information about energy waste, have notbeen shown to motivate behavior change or energy savings,other methods, like rewards or active feedback, have had higherlevels of effectiveness (with some degree of variability) [35]. Astudy by McClelland and Cook used household monitorsdisplaying electricity use in cents per hour and found thathouseholds with the installed monitor used 12% less electricity[36]. However, despite this success, these studies either onlytake place in the short term or even show that the effectivenessof the method diminishes with time [35].2.2 Smart Products in the Energy RealmWhen we discuss smart devices, we place them into one oftwo categories: passive or active conservation products. Passiveproducts, such as Nest, may provide personalized settings forcomfort, convenience, and energy savings; however, they donot attempt to change user behavior [10]. Although our studyfocuses specifically on water, studies on smart products in thewater conservation realm are sparse. For a more comprehensivebackground, we investigated energy conservation smart devicesfor insight as well.2.2.1 Passive Products Han and Lim designed asystem that can provide a sustainable experience forhomeowners by linking new and easy to use digitaltechnologies aimed at conserving energy automatically. Thisdesign relies on sensors that track the current temperature,lighting, etc. and automatically adjust to a more energy efficientsetting on its own [26]. Automated devices that use sensors,like the prior example, have been proposed as a furtheradvancement to the simpler mechanical based approach toreducing energy consumption, like water flow-restrictingdevices [27, 28]. Another system created by Capone et. al.adopts a generalized method for household appliancemanagement and then gathers information from a sensor toadapt to user behavior. The user profiling process includes amechanism for recording events that can help characterize auser’s interactions with their home devices and then utilizes alearning algorithm to meet the user requirements [29]. Ourdesign, like Han and Lim and Capone et. al., adapts toindividual users by tailoring the faucet to behave similarly tousers’ current behaviors, while still promoting waterconservation through small changes.Given the difficulty of changing user habits, studies haveproposed employing technology-centered approaches toencourage energy savings without harsh changes in behavior[30, 31]. However, despite these advantages, many users feeldisconnected from the technology behind the passive systems.For example, Nest users found the system frustrating and2.3 Cognitive StylesStudies have found links between peoples’ cognitive stylesand learning behavior. These styles can be broken down intomany dimensions including field perception, impulsivity when3Copyright 2019 ASME

turning the handles by the user, and “smart” faucet operation iswhen both the wizard and the user can control the faucetconcurrently. The following section discusses the designparameters and construction of the WoZ faucet.making decisions, and convergent versus divergent and holisticversus focused problem-solving strategies [37].Many different models of cognitive styles exist. Of note,Witkin et. al. discusses the application of learning styles to aneducation environment. This study suggests that the fieldperception dimension of cognitive style can be applied tofurther understand how students learn and teachers teach, howstudents and teachers interact, and how students ultimatelymake choices about their education and work [38]. Hauser et.al. discusses morphing the content, look, and feel of websites tomatch the user’s cognitive style, which can be inferred fromclickstream data [39]. Similarly, research by Urban et al. provedthat morphing advertisements to the cognitive style of thepotential customer increased click-through rates by 245% [40].Research into cognitive styles is fairly limited with fewapplications [41]. However, what exists can lead to importantinsights when designing a smart faucet to understand and adaptto the way its users think.3.1 Faucet DesignSince the faucet must be able to simultaneously share control ofthe temperature and water flow with both the participant and awizard, a custom faucet was required. Shared control wasaccomplished through retrofitting the current water lines withelectronic servomotors. Secondary to the simultaneous control,the water used by each participant must be tracked for each setof dishes that they washed. Since, the freestanding faucet couldnot be hooked up to a main drainage line, waste water wasstored in a 50-qt bucket below the faucet basin. A platform witha load cell was then designed to support the waste bucket andprovided real-time tracking of water consumption.Aesthetically, the faucet was designed with the intent toblend into a typical household environment and evoke minimalinitial emotions of novelty or curiousness from users.Therefore, electronics are concealed in plastic housingsalongside weight monitoring devices and water collectionbuckets that are hidden in the cabinet below the sink. However,the servomotors that control the faucet do emit an audible noisewhen they turn.As seen in Figure 2, the Raspberry Pi can control the coldand hot lines via gear trains that open each respective valve.Servos were connected to each gear train and could beindividually controlled so that all usable faucet settings couldbe reproduced. For ease of control during experiments, a GUI(Graphical User Interface) was created for an operator to easilyselect the desired flow and temperature settings. Weighttracking is controlled by a separate script, which tares,measures weight at a fixed time step, and then saves the data toa text file for each phase of the experiment.2.4 Research HypothesesWe hypothesize that an active or “smart” faucetintervention that is able to control the output flow andtemperature based on differences in a user’s cognitive style andtask could be a more effective intervention for waterconservation and prolonged user-behavior change than theexperiments presented in the previous section. Therefore, ourhypotheses are as follows:Hypothesis 1: The use of a smart faucet intervention decreasesthe water consumption of a user for a given activity.Hypothesis 2: The interaction with a smart faucet decreases thewater consumption of a user immediately after the interventionis discontinued.3. METHODThe user experiment described in this paper was a betweensubject experiment conducted at Stanford University. Userswashed 3 sets of dishes with a custom faucet to identify thewater saving potential of a “smart” faucet, possible userreactions to a perceived autonomous product, and residualeffects users may carry with them after (drawing of bottomhere) interacting with the faucet. Users placed in theexperimental group interacted with a “smart” faucet that wouldadjust its flow and temperature according to the task andbehavior of the user. While a future implementation of thissystem could automate the “smart” behavior, we utilize aWizard-of-Oz (WoZ) [42] method to remotely control thefaucet. WoZ control refers to when a “wizard”, a member of theexperimental design team, remotely controls a robot to performan action. This can be employed so that a human is able tocontrol a potentially dangerous interaction, or in this case, toprototype and learn about a potential design before fullyimplementing the robot's behavior algorithmically. Althoughthe wizard can control the faucet, all users can also control thefaucet normally, so control over the faucet settings are shared.For the purposes of this paper, “normal” faucet operation iswhen the faucet only acts upon commands given to it throughFIGURE 2 FAUCET SETUP FOR WOZ CONTR


Related Documents:

(b) A wizard follows the user and updates his or her location in the wizard UI on a tablet PC. Figure 2. Topiary’s wizard UI. The wizard map represents entities’ current location and orientation; to simulate updates, a wizard drags them on the map. The end-user screen lets a wizard mon

Recover files detected by their signatures wizard on page 69 Recover files from a formatted partition wizard on page 73 Recover files from a deleted partitions wizard on page 75 Recover files from a physical disk wizard on page 77 Restore a deleted partition wizard on page 82 Create a new partition wizard on page 83

The User Profile Wizard User Guide. This document! The Deployment Files folder. This folder contains the files needed to migrate a workstation to a new domain. User Profile Wizard The User Profile Wizard Command Line Console The User Profile Wizard Deployment Kit We will reference these icons throughout this user guide.

Experiment 15Diodes Lab – Part 1 Experiment 16Diodes Lab – Part 2 Experiment 17Transistor Lab 1 Experiment 18Transistor Lab 2 Experiment 19Transistor Lab 3 Experiment 20Induction, Magnet and Coil Additional Equipment needed: Please refer to the Equipment Needed section in the beginni

Experiment 2 Franck-Hertz Experiment Physics 2150 Experiment 2 University of Colorado1 Introduction The Franck-Hertz experiment demonstrates the existence of Bohr atomic energy levels. In this experiment you will determine the first excitation potential of Argon contained in a Franck-He

Experiment 9 Determination of Enthalpy and Entropy Changes 83 Experiment 10 Electrochemistry 88 Experiment 11 Sulfide Separations 99 Experiment 12 Separation and Identification Using Paper Chrom. 104 Experiment 13 Cis and Trans K[Cr(H2O)2(C2O4)2] 110 Experiment 14 Coordination Complexes of Co(III) 116

Wing Type Surface Setup is the last step in the Wizard. Press the in the upper right corner to save settings and exit the Wizard. The application will display the AS3X Settings menu after exiting the Wizard. Exit the Wizard

The Royal Shakespeare Company’s stage adaptation of The Wizard of Oz is a musical based on the novel The Wonderful Wizard of Oz by L. Frank Baum, and the 1939 Metro-Goldwyn-Mayer film, The Wizard of Oz, with a book by John Kane, music by Harold Arlen and lyrics by E.Y. Harburg. It debuted at the Barbican Theatre in London’s West