Brotate and Tribike: Designing Smartphone Control for Cycling
Paweł W. Woźniak, Lex Dekker, Francisco Kiss, Ella Velner, Andrea Kuijt, Stella Donker
BBrotate and Tribike: Designing Smartphone Control for Cycling
Paweł W. Woźniak
Utrecht UniversityUtrecht, the Netherlands [email protected]
Lex Dekker
Utrecht UniversityUtrecht, the Netherlands
Francisco Kiss
University of StuttgartStuttgart, Germany [email protected]
Ella Velner
University of TwenteEnschede, the Netherlands [email protected]
Andrea Kuijt
Utrecht UniversityUtrecht, the Netherlands
Stella F. Donker
Utrecht UniversityUtrecht, the Netherlands [email protected]
ABSTRACT
The more people commute by bicycle, the higher is the numberof cyclists using their smartphones while cycling and compromis-ing traffic safety. We have designed, implemented and evaluatedtwo prototypes for smartphone control devices that do not requirethe cyclists to remove their hands from the handlebars—the three-button device Tribike and the rotation-controlled Brotate. The de-vices were the result of a user-centred design process where weidentified the key features needed for a on-bike smartphone controldevice. We evaluated the devices in a biking exercise with 19 par-ticipants, where users completed a series of common smartphonetasks. The study showed that Brotate allowed for significantly morelateral control of the bicycle and both devices reduced the cognitiveload required to use the smartphone. Our work contributes insightsinto designing interfaces for cycling.
CCS CONCEPTS • Human-centered computing → Interactive systems and tools . KEYWORDS cycling; smartphone; physical activity; traffic
ACM Reference Format:
Paweł W. Woźniak, Lex Dekker, Francisco Kiss, Ella Velner, Andrea Kuijt,and Stella F. Donker. 2020. Brotate and Tribike: Designing SmartphoneControl for Cycling. In
ACM, New York, NY, USA, 12 pages. https://doi.org/10.1145/3379503.3405660
All around the world, the number of bike commuters is on therise [10, 15]. Research has shown that cycling provides significanthealth benefits and contributes to more liveable and sustainablecities [39], and coincides with the rise of smartphones as a primarymeans of communication [11, 52]. Studies have shown that cyclistsaround the globe often listen to music or conduct phone calls whilecycling [13]. Multiple studies [18, 43] have shown that cyclists are
MobileHCI ’20, October 5–8, 2020, Oldenburg, Germany © 2020 Association for Computing Machinery.This is the author’s version of the work. It is posted here for your personal use. Not forredistribution. The definitive Version of Record was published in , https://doi.org/10.1145/3379503.3405660 . likely to use their smartphones while in motion, removing theirhand from the handlebars and thus increasing the probability ofan accident as single-handed cycling significantly reduces lateralcontrol [45].As an ever increasing number of smartphone users cycle andtraffic infringement penalties for using a smartphone while cyclingonly have limited effects [18], an alternative approach is needed.Enabling users to use their smartphone while cycling while min-imising the decrease in safety for the usage may not only increasetraffic safety, but also contribute to further development of cyclingas a sustainable means of commuting. Consequently, designingeffective smartphone controls for in-ride interactions that minimisedistraction and allow for directing more attention to traffic emergesas a challenge for Human-Computer Interaction (HCI).A number of past works in HCI have addressed interacting withtechnology on the bicycle, both during and after the ride. Multipleways of providing input were proposed including gestures [8], on-body tapping [44] and voice interaction [40], however this workmostly focused on performance in using interaction techniques anddid not investigate the impact of the input methods on the controlof the bicycle (which has safety implications) or performance insecondary tasks such as smartphone control. In contrast, our workexplores the means of performing smartphone tasks while cyclingusing input methods that do not require removing one’s hand fromthe handlebars, and studies how providing input in-ride affects thestability of the bicycle.To that end, we conducted a user-centred design process to de-sign alternative handlebar devices for controlling the smartphonewhile cycling. We first conducted a survey to determine whichsmartphone functions were most commonly used during cycling.The next step was a pre-study with low-fidelity prototypes to de-termine the two most promising designs. We then built functionalprototypes of Brotate—a smartphone controller where actions areperformed by rotating the handlebar grip around an axis parallelto the handlebars, and Tribike—a device with three buttons placeddirectly next to the handlebar grip. We then compared the twoprototypes against using the smartphone held in one hand in an ex-periment with n =
19 participants. We contribute: (1) the first, to ourknowledge, systematic study of input techniques for smartphonecontrols during cycling; (2) a generative design contribution of twoexamples of eyes-free smartphone controls for bicycles developedin a user-centred design process and (3) experimentally determiningthe desirable properties for future improved smartphone controlsfor cycling. a r X i v : . [ c s . H C ] S e p obileHCI ’20, October 5–8, 2020, Oldenburg, Germany Paweł W. Woźniak, Lex Dekker, Francisco Kiss, Ella Velner, Andrea Kuijt, and Stella F. Donker In this section, we first review past work on interaction for cyclists.We then explore similar designs from the automotive domain anddescribe our inspirations from work on providing input duringphysical activity.
Designing interactive artefacts and services for cyclists is a recur-ring topic in HCI research. A significant body of work is dedicatedto providing effective output for cyclists, especially for the pur-poses of navigation. Poppinga et al. [33, 34] built a tactile displayfor communicating directional cues. They found that users wereable to process signals communicated through vibration motors onthe handlebars with medium accuracy. Huxtable et al [17] proposedmoving the vibration motors to a user’s wrists, and Steltenpohl etal.âĂŹs Vibrobelt [42] was a device that provided output throughvibration on the waist. Using Vibrobelt led to fewer navigationerrors than a smartphone mounted on the handlebars. Anotherapproach to navigation used street surface for display as in SmartFlashlight [7]. Matviienko et al. [28] found that auditory feedbackwas preferred by child cyclists. The variety of systems developedfor cyclists in past research reflects a demand for the ability toeffectively interact with information while cycling, which our workexplores. Further, the research also includes a range of design possi-bilities for modifying existing bike equipment, which inspired ourdesign.Another strain of HCI work investigates the use of interactivetechnologies for improving cycling safety. Carton [4] proposed asmart glove for additional safety in indicating directions, result-ing in additional visibility. Gesture Bike [8] implemented gesturerecognition to support the automatic activation of turn signals forbicycles. The authors found that users preferred input methodswhere interaction did not require them to remove their hands fromtheir handlebars, as this was perceived as a safer behavior. Trafficresearch confirms that single-handed control should be avoided,especially at low speed [38]. Matviienko et al. [27] proposed addinga multimodal warning signal to bicycles to increase awareness incritical cycling situations, with warning lights mounted at multi-ple points. Finally, BikeSafe [14] showed that smartphones couldbe effectively used to detect dangerous cycling behavior and thuscontribute to preventing dangerous traffic incidents. Past work il-lustrates the potential of interactive technologies to augment thesafety of cycling, on which this paper attempts to build. We alsoobserve that all the works cited here found that the way the cyclistmoved with reference to the bicycle was of particular concern forinteraction design. Our work investigates this aspect further.Research has also explored design alternatives for input methodsfor in-ride interactions. Vechev et al. [44] proposed tapping differ-ent areas on the user’s body to activate features while cycling. Thisstrategy, however, requires removing one’s hands from the handle-bars, as does gestural input [4, 8]. Sörös et al. [40] suggested thatvoice navigation could be used to control an augmented reality dis-play for cycling. Speech interaction, however, is sensitive to noiseand a set of design limitations [12]. Hochleitner et al. [16] designedsmartphone controls for cycling and compared touch, button andwristband interaction. The focus of the study was designing game controls for outback cycling. In contrast, our work focuses on usingsmartphones in traffic and the impact of smartphone operation onbicycle control. Overall, past research offers little consensus onwhat the effective and safe ways of providing input during cyclingare. As past work in traffic studies provide empirical evidence ofwhat behaviours are unsafe [45], the challenge for HCI is to designinput methods that discourage those unsafe behaviours. Our workexplores and compares in-ride input methods and examines theirpotential for safety and performance. We contribute the first study,to our knowledge, that empirically examines the impact of using asmartphone on performance in everyday tasks and bicycle control.Augmenting the user experience of cycling has also been ex-plored in past research. Rowland et al. [37] postulated includingmore context-aware interactions in cycling technology and embrac-ing the qualitative experience of a bike ride. Further, they stressedhow eyes-free audio control was important to a smooth cyclingexperience. Our work investigates the ways users can effectivelycontrol in-ride audio. A number of systems stressed the social ex-perience of cycling and highlighted the role of the cyclist as part ofa larger community of those in traffic. Biketastic [35] explored howroutes could be annotated to estimate how friendly they are forcyclists. Ari [1] was an e-bike designed to help users cross only ongreen lights. GameLight [51] attempted to transform cycling into asocial exergame. Walmink et al. [46] built a system that displayedthe rider’s heart rate on the back of their helmet for social shar-ing. Our work explores a different dimension of the social aspectof cycling. We investigate the means for cyclists to display safebehaviours.
A bicycle is part of a modern traffic environment that also includesother vehicles. Consequently, work that explores input while driv-ing cars or motorbikes is related to our inquiry. Despite the dif-ferences that characterise the experience of driving each of thesemeans of transportation, some research focuses on common aspectsand useful insights can be derived for the particular case of bicycles.The automobile interface field proposed several designs that keepthe driver’s hand on the steering wheel while providing second-task input. Modern cars with button-equipped steering wheels area prime example. Research proposed alternative solutions such asmulti-touch [32] or pressure-based input [31]. Similarly, commercialmotorcycles use buttons located next to the handgrips to controlsecondary functions [26]. These works not only provide inspirationfor the design space we explored, but also showed that a numberof complex tasks can be effectively and safely performed whilecontrolling a vehicle. Our work explores this paradigm in a bicyclecontext.
As cycling is an outdoor physical activity, our work is also inspiredby past efforts on how to use these methods during physical ac-tivity [19]. Past work suggests that the number and complexityof available interactions under physical exertions (which cyclingis likely to cause) should be limited. Wozniak et al. [49] showedthis requirement to be valid for running and Mencarini et al. [29]for climbing. Their findings echo the constraints for interaction in rotate and Tribike: Designing Smartphone Control for Cycling MobileHCI ’20, October 5–8, 2020, Oldenburg, Germany motion as postulated by Marshall and Tennent [25]. In line withthis work, our approach first identifies the most necessary inputs,considers motion constraints and investigates how users interactwith their environment while providing input. We adopt a motion-centric design process, where we aim to minimise the complexityof the interactions while biking and design controls that affect theuser’s and the bicycle’s movement to the least degree.
Brotate and Tribike were developed in an iterative user-centreddesign process. Here, we provide an account of how we reached thefinal designs of our prototypes. Research in traffic safety shows thatcyclists use mobile phones regularly [18, 43] and that this usageconstitutes a risk to traffic safety [45], but does not establish whatfunctions of the phone are most commonly used. Consequently,we started designing a new control device for the smartphone byinvestigating what should be controlled. To that end, we conductedan online survey.
We designed an online survey to investigate user requirements forinteracting with a smartphone while cycling. Further, we investi-gated the user experience and social acceptability of in-ride mobilephone use. In the first part of the survey, we asked participants torank the frequency of performing smartphone actions on a six-pointLikert scale from ‘never’ to ‘very often’. Then we asked how oftenand how they operated their phone while cycling: using their hand,headphone remote or dedicated Bluetooth device. In the second partof the survey, we presented the users with four actions performedwhile biking, in a randomised order: calling with the phone nextto one ear, writing a text message with one hand, writing a textmessage with two hands and operating the headphone remote withone hand. As suggested by Williamson and Brewster [36], we pre-sented alternative contexts in which the actions could be performed:a busy crossing where cyclists should give way, a crossing withtraffic lights and a quiet street. We then applied the method usedby Montero et al. [30], asking an open question: ‘What would youthink if you saw someone else performing this gesture?’ and askedthe participants to rate their answer to: ‘How would you feel aboutperforming this gesture in the following situations?’ on a six-pointLikert scale from ‘embarrassed’ to ‘comfortable’. The survey designand the complete answer set are available as auxiliary material.
We recruited 154 participants (119 male and35 female, aged from 17 to 71, M = . SD = .
48) via socialmedia posts and snowball sampling. Forty participants identifiedas moderate cyclists, cycling 4 to 6 hours a week, and 55 reportedcycling more than 6 hours a week. Participation was voluntary.
Most participants (95 participants, 61%) reportedthat they used their mobile phone on their bike. The majority ofthem operated the phone with one hand. Eleven users always con-trolled their phone using a Bluetooth remote control, while 61 outof 95 never used such a device. We used a two-way ANOVA onaligned-rank-transformed [47] data to investigate the effect of typeof action and context on the acceptability of the action. Therewas a significant main effect of type of action ( F , = .
89, busy crossing traffic light quiet streettwo-hands 1.4 1.7 2.9one-hand 1.7 2.1 3.4calling 2.1 2.6 4.1remote 3.2 3.8 5.0
Table 1: Quantitative results of the survey. Acceptance was rated ona six-point Likert scale from ’embarrassed’ to ’comfortable’ for fourTYPES OF ACTION and three CONTEXTS. All pairs showed signifi-cant differences. p < . F , = . p < . F , = . p > . p < .
001 level. Holding thephone in both hands was perceived as most unacceptable and theparticipants were most likely to accept in-ride smartphone use onquiet roads. Table 1 presents the results in detail. Two researchersused affinity diagrams to analyse the survey data from the openquestions. Participants’ comments focused on an inherent need forsafety, changes from typical cycling posture, and responses to thetraffic situation. We observed that users found many of the behav-iors in the survey embarrassing and potentially dangerous. Lowusage rates for Bluetooth controls showed that consumer-gradedevices were not used widely. Finally, controlling music, answer-ing calls and activating the voice assistant were identified as thetop functions to be controlled. Interestingly, social media use wasranked as the least often used.
The next step in our design process was creating speculative proto-types of possible solutions. Having previously analysed relevantresearch work, we also analysed commercial solutions for inspi-ration. We reviewed commercially available solutions to find thatdevices featured five or more buttons arranged in a circular or linearpattern . They were also all intended to be mounted in the middleof the handlebars. Consequently, we designed alternatives with lesscomplexity that also minimised the hand movement required tooperate the device.We built four low-fidelity prototypes of possible smartphonecontrol devices, presented in Figure 1. First, the button device, in-spired by headphone remotes and motorbike controls, featuredthree buttons placed in the direct vicinity of the handlebar grip.Second, the touchpad device, inspired by touch controls in mod-ern cars, featured a touch-sensitive surface next to the grip. Third,the rotation device, which mimicked rotational bells, offered inputthrough rotating it around the handlebar axis and an additionalbutton. Finally, the lever device used derailleur control levers withan additional button. All prototypes were built using existing bicy-cle parts and moldable clay. We then conducted a study to gatheruser feedback about the prototypes. See https://buy.garmin.com/en-GB/GB/p/621230 or https://cobi.bike/product obileHCI ’20, October 5–8, 2020, Oldenburg, Germany Paweł W. Woźniak, Lex Dekker, Francisco Kiss, Ella Velner, Andrea Kuijt, and Stella F. Donker Figure 1: The four low-fidelity prototypes used in our design pro-cess. Bar charts below the prototype pictures show SUS scores foreach of the devices.Figure 2: The pre-study apparatus. The four low-fidelity prototypeswere mounted on the handlebars to allow users to imagine usingthe devices while cycling.
We recruited 14 participants (9 male and 5 fe-male, aged 22–61, M = . SD = .
05) through snowball sam-pling. All participants reported cycling regularly, i.e. more than 4hours a week. They all stated that they used their mobile phoneswhile cycling.
We mounted handlebars with a stem on a woodenplate to conduct our study. The four prototypes could be easilymounted and removed from the handlebars. The setup of this canbe found in Figure 2. The participants were presented with each ofthe prototypes in a counter-balanced order. We used a think-aloudprotocol where we asked the participants to imagine how theywould use each of the devices to control their smartphone. Afterthe participants interacted with the prototype, we administeredthe SUS scale to measure the anticipated usability of each of thedevices. After all four devices were presented, we asked partici-pants to rank them in order of preference. Finally, for the devicesthat incorporated motion, i.e. the rotation and touchpad devices, weconducted a gesture elicitation for the three most-requested smart-phone functions—controlling calls, music and the voice assistant.The elicitation was conducted according to the procedure designedby Wobbrock et al. [48] with 9 referents, see Table 2.
We used a one-way ANOVA on aligned-rank-trans-formed [47] data to compare SUS scores between the devices. Therewas a significant main effect of the type of device, F , = . p < .
05. Post-hoc analysis revealed that the touchpad device scoredsignificantly lower than the other devices, all at p < .
05. In termsof rankings, participants opinions were divided, with the buttonranked best and the touch device being less preferred. The partici-pants’ remarks were transcribed verbatim. As in the online survey,two researchers analysed the data with affinity diagramming, focus-ing on the differences between the devices. Participants commentedextensively on how the devices could be integrated into everydaycommuting and how they affected using other bicycle controls. The lever device was identified by most participants as potentially inter-fering with other actions and using a form reserved for derailleurcontrol. Many participants also suggested using a longer grip andplacing the devices further towards the centre of the handlebars,to assure a firmer grip for bicycle control. Elicitation results wereclassified and gesture sets were compiled using agreement scoresas suggested by Wobbrock et al. [48].
The survey showed that many users needed to operate their phoneswhile cycling with a general negativity to use additional devicesto do so. We gathered the insights gained in our design processso far using affinity diagramming and reviewed requirements inother work that investigated handlebar controls, e.g. [21]. Basedon that, we defined design requirements for future prototypes. Thesmartphone control system should: • provide easy access to music control and answering calls.These were the top functions in our survey; • enable the user to keep both of their hands on the handlebar.This requirement provides increased safety. It also reflectslegal regulations in some countries. • limit the hand movement on the handlebars. A steady handposition on the handlebars ensures minimal deviation fromthe straight line [45]; • not interfere with bicycle controls, e.g. brakes, derailleurlevers, or the bell. This was a concern often mentioned byparticipants when reflecting on the low-fidelity prototypes; • enable eyes-free operation as a secondary task, which wasboth a voiced user need and a safety requirement; • convey an impression of safe use. The majority of the partic-ipants in the survey and low-fidelity prototype study com-mented extensively on the need for a bicycle device to evokesafety.In light of these design requirements, we decided to further refineand evaluate the button and rotation devices. We eliminated the touchpad device that users perceived as significantly less usable.We did not develop the lever further, because of the participants’remarks on it interfering with bicycle controls. Building on thatinsight, we decided to place our devices on the left side of thehandlebars as designed and artefact for commuters. City bikesusually feature less controls on the left side of the handlebar as theyoften have only one brake lever and zero or one derailleur controls. rotate and Tribike: Designing Smartphone Control for Cycling MobileHCI ’20, October 5–8, 2020, Oldenburg, Germany Figure 3: The final prototype of Brotate and its supported actions.
Two devices—Brotate and Tribike were the final products of ourdesign process. These are refined versions of the earlier button and rotation prototypes. Both devices were built with our designrequirements in mind and in a form that considered the shape andfunction of the bicycle handlebars.
Brotate enables rotating part of the grip of the han-dlebars to control the cyclist’s smartphone. The device supportstwo basic movements—forward and backward. To extend the rangeof functions supported, we added a single button located directlyat the grip. Brotate is designed in a way that enables the cyclist’sthumb to constantly rest on the button, thus requiring minimalmovement for context-based actions. Figure 3 shows the device andthe range of supported motion. The control techniques used repre-sent the gesture set from the elicitation process conducted using thelow-fidelity prototype. Brotate provides tactile feedback throughresistance when rotating. The rotating part returns to its startingposition once an action is completed. Studies of the low-fidelityprototype revealed that a shorted grip was needed for Brotate sothat the flesh of the palm could rest comfortably on the rotatingpart also allowing to easily move away from it.
Tribike is heavily inspired by headphone remotes, asthese were positively perceived in the survey and the button devicewas highly ranked in the low-fidelity prototype study. Further,Tribike’s form factor borrows from devices seen on motorbikeswhich benefit from the proximity of the buttons to where handsare usually placed. As suggested by users, Tribike uses a buttonwrap-around alignment where buttons can be reached with thethumb with little rotation of the palm holding the handlebars. Thethree buttons are highly tactile and feature additional spacing tocounteract accidental presses. As the three horizontal buttons usethe same layout as a standard in-line headphone remote, with whichusers are familiar, we used standard mobile phone button patterns tocontrol the smartphone. Figure 4 shows the final Tribike prototype.
We built 3D models of both of the devices and 3D-printed therequired components, as illustrated in Figure 5. Brotate uses a de-sign where the outer part of the grip moves a Panasonic EWV-YG9U04B14 rotary potentiometer as it rotates. The base of the
Figure 4: The final version of Tribike with three action buttons. Thedevice uses the same button patterns as standard mobile phone in-line remotes.
Task Tribike Brotate
Answer call MB BDecline call MB twice B twiceVolume up TB Rotate upVolume down BB Rotate downNext song TB twice B + rotate upPrevious song BB twice B + rotate downPause music MB B
Table 2: The smartphone actions used as tasks in our experimentwith descriptions of how these tasks were performed with buttonpresses and/or rotations using Brotate (MB—middle button, TB—topbutton, BB—bottom button) and Tribike (B—button). Call-related ac-tions are only available when a call is active.Figure 5: Final 3D models of the Brotate (left) and Tribike (right).The spring inside Brotate, visible in the cross-section, provides tac-tile feedback when rotating the device. The recoil mechanism inBrotate includes a steel spring. potentiometer is fixed to the handlebars. To provide tactile feed-back and assure that the grip would return to its original position,we mounted a spring inside the device. The tactile buttons of bothdevices use flic programmable buttons, which feature embeddedBluetooth Low Energy (BLE) connectivity. Rotary input in Bro-tate is processed by a DFRobot Beetle BLE Arduino-compatiblemicrocontroller. We also built a custom Android application whichlogged all the device events for analysis. The devices connected tothe user’s smartphone using BLE and the inputs were mapped tosmartphone actions using the Tasker tool. https://flic.io/ https://play.google.com/store/apps/details?id=net.dinglisch.android.taskerm obileHCI ’20, October 5–8, 2020, Oldenburg, Germany Paweł W. Woźniak, Lex Dekker, Francisco Kiss, Ella Velner, Andrea Kuijt, and Stella F. Donker To evaluate Brotate and Tribike, we conducted a within-subjectsexperiment where we asked participants to complete smartphone-related tasks while cycling, using the two devices and controllingthe phone with one hand. We endeavored to investigate how effec-tive Brotate and Tribike were in controlling a smartphone whilecycling. Further, we studied the usability of the devices and theirimpact on the bicycle’s lateral control. Our work closely followsthe study conducted by De Waard et al. [45]. As our work looksspecifically at HCI for physical activity and not traffic safety, wechose to adapt a study from the traffic safety domain rather thandesigning another protocol which would require further validation.
We used social media and flyers to recruit 19 participants (9 maleand 10 female, aged 22–66, M = . SD = . Two equivalent unisex city bicycles were used in the experiment.One was fit with Brotate and the other with Tribike. Both prototypeswere fit next to the left handlebar grip of the respective bicycle. Wechose the left side of the handlebar as consumer-market bicyclesusually feature derailleur controls placed on the right side. Partici-pants were given a Huawei P30 Lite smartphone running Android9.0 with the study software installed and wired headphones. Theexperiment was recorded with a stationary camera and an actioncamera mounted on the bicycle that recorded all device interactions.An additional smartphone (Samsung Galaxy S8) was mounted onthe seat post to measure acceleration throughout the experiment.We also mounted one-way radios centrally on the frames of thebicycles so that the experimenters could communicate with theparticipant at all times.
Our study focused on evaluating how the users could leverageTribike and Brotate to operate the smartphones while cycling. Tothat end, we asked the participants to complete seven basic smart-phone actions based on our survey: turn the volume up, turn thevolume down, play the next song, play the previous song, pausethe music, decline a call and answer a call. We used Latin squaresto avoid order bias in administering the tasks. The task was com-municated to the user with pre-recorded voice instructions, e.g.‘Please switch to the next song now.’ with the exception of the callsto be received or declined where the calling person’s name indi-cated whether the participant was to decline or receive the call (thesmartphone was set to read caller names for incoming calls). When a call was received, an experimenter thanked the participant forcompleting the task and disconnected. Throughout the experiment,participants were asked to cycle straight at their preferred, moder-ate speed on 400-metre long straight track. The track was locatedin a location with limited visual distraction, featuring only grassand trees on its entire length. We assured that the individual taskswere distributed in a way that there was a minimum cycling timeof 15 seconds between completing one task and starting the next.This assured that participants could focus solely on cycling beforeattempting each task. Additionally, we ensured that no task wasperformed while the participant was turning around or within 10seconds before or after the turn. Figure 6 shows an example of a tasksequence in the experiment. The order of the tasks was differentwithin each trial so that the participants could not anticipate thenext action to be performed.After the participants arrived at the cycling track, we greetedthem, explained the purpose of the experiment, and asked themto complete a demographics, data processing and consent form.Afterwards, we asked the participants to sit on their bicycles andadjusted the seat height as desired. We then assigned the initialcondition to the participant, based on Latin-square order balancing.Next, for each condition, the participants completed a practicetask in which they could review all the tasks and actions whilestationary. We presented all the tasks in a fixed order and presentedthe required input. The participants were then free to use the deviceand complete tasks until they reported that they could comfortablyuse it while cycling. We then asked them to begin cycling, remindingthem to cycle at a moderate and comfortable speed, in a line asstraight as possible. The participants then completed the sequenceof smartphone tasks as shown in Figure 6. After the tasks werecompleted in each condition, the participants completed NASA TLXand SUS questionnaires. The tasks were then completed for theremaining conditions with rest time as required in between. Aftercompleting all tasks in the study, we conducted a semi-structuredinterview with each participant. The interview focused on theperceived differences between the conditions, how using the devicesaffected the ride experience and the usability of the devices. Afterthe conclusion of the experiment, we provided the participants withthe remuneration and offered refreshments.
Our experiment evaluated the following hypotheses, based assump-tion that Brotate and Tribike would alleviate the issues discoveredby De Waard et al. [9]:(1) Using Brotate and Tribike will reduce sideways movementwhile cycling compared to one-handed smartphone opera-tion. We investigated if our designs could limit the negativeimpact on lateral control caused by one-handed smartphoneuse while cycling [45].(2) Brotate and Tribike will enable performing smartphone tasksmore efficiently than one-handed control.(3) Brotate and Tribike will be perceived as more usable and re-quiring less cognitive effort to use than one-handed control. rotate and Tribike: Designing Smartphone Control for Cycling MobileHCI ’20, October 5–8, 2020, Oldenburg, Germany
Figure 6: An example sequence of tasks which the participants performed in the experiment. Note that the tasks were only constrainedby minimum time between them. Distance was determined by the participant’s preferred moderate cycling speed. Refer to Table 2 for therespective controller actions. Conditions were administered in a Latin square order while the order of the seven tasks was randomised.
Device used was the only condition in our study. The levels were:One Hand, Brotate and Tribike. We measured the total
TaskCompletion Time (TCT) for the seven tasks as the sum of the timesfor the individual tasks as a measure of effectiveness in control.The time was measured from the end of the voice command tothe smartphone registering the user action. In the case of declin-ing/receiving calls, the time was measured from the moment theparticipant’s phone started ringing. The participant’s phone alsorecorded the
Error Rate . Performing an incorrect action in a taskwas counted as an error. Raw
NASA TLX for and
SUS scores werecollected using printed questionnaires at the conclusion of eachcondition. We used
NASA TLX as a cognitive load measure as itwas linked to smartphone use while cycling by De Waard et al. [45].
SUS was used to measure the perceived usability of the system.Finally, the smartphone mounted under the bicycle seat recorded
Bicycle Tilt . De Waard et al. [45] empirically related lateral stabilitywhen cycling in a straight line to traffic safety. We used the iner-tial measurement unit (IMU) of the phone to capture the sidewaysmovement of the bicycle through the experiment thus measuringlateral control. This approach was inspired by past work [20] thatsuccessfully used IMUs to measure bicycle motion. This is an alter-native approach to De Waard et al. [45], which we used becausewe aimed to study the effect of the different devices on bicyclemovement, and not the nature of the movement itself. Our analysisinvestigates the amount of sway and/or tilt caused by using thesmartphone control devices while cycling thus providing a mea-sure of lateral control. We first applied a filter to the IMU signal toremove data from instances when the bicycle was stationary andwhen the participants were turning around. Possibly dangerouschanges in the position of the bicycle would result in changes inthe acceleration vectors acting on the bicycle. However, gravityand the acceleration caused by the participant pedalling also con-tribute to the overall acceleration. To factor out these vectors fromour analysis and focus solely in sideways acceleration change, wedefined a 100 ms sliding window and integrated the signal for thethree instantaneous acceleration values a x and a y as follows ( a z represents gravitational pull and is constant): ∫ t + mst (cid:113) a x + a y dt Finally, in order to account for each participant’s individualcomfortable speed, body movements and cycling style, the scoreswere standardised. Data was standardised using RâĂŹs normalizefunction. The initial acceleration period was removed from the datathrough matching with video timestamps.
We used a one-way ANOVA to investigate the effect of the deviceused on TCT, error rate, NASA TLX scores and Bicycle Tilt. Signifi-cant main effects were observed for all measures but the Error Rate.Table 3 presents the details of the analysis and Figure 7 illustratesthe results.To further confirm the bicycle motion analysis, we ran an ad-ditional analysis Root Mean Square (RMS) a y values, i.e. the rawsideways tilt of the bicycle. Hand control resulted in the largestRMS a y values recorded, while TriBike produced the lowest values.An one-way ANOVA showed that there was a significant effectof device used on RMS a y : F(2,58723)=78.9, p<.001. Tukey HSD re-vealed that all condition pairs were significantly different at p<.01.These results are in line with our derived measure of tilt.We applied the align rank transformation [47] on SUS data toanalyse it with a one-way ANOVA. Mean cell frequencies did notexceed 10 [24]. There was a significant effect of device used on theSUS score, F , = . , p < .
01. Post-hoc test with Tukey HSDshowed that there was a significant difference between One Handand Tribike, p < .
01. Figure 7 shows detailed results.The volume of recorded qualitative data was 200 minutes. Giventhe size of the data set, we adopted a pragmatic theme-based ap-proach to data analysis as suggested by Blandford et al. [2]. Record-ings of debriefing interviews were transcribed. Two researchersfirst identified relevant passages from the data, which were thenopen-coded independently by the two researchers. Based on thecodes and iterative discussion, we identified three themes in thedata: integration (with other bicycle components), learning(how to use the device), and movement. obileHCI ’20, October 5–8, 2020, Oldenburg, Germany Paweł W. Woźniak, Lex Dekker, Francisco Kiss, Ella Velner, Andrea Kuijt, and Stella F. Donker
Figure 7: Mean Task Completion Time (left) and Error Rates (centre left) in the experiment for the three experimental conditions. Errorbars visualise standard error. Normalised Mean Bike tilt values collected in our experiment. Error bars show standard error. The values werenormalised to account for between-participant differences in riding style. Mean SUS (centre right) and Raw NASA TLX (right) scores collectedin our experiment for the experimental conditions. Error bars show standard error.
TCT [ s ] Error Rate NASA TLX Bicycle tilt [ m / s ]Condition M SD M SD M SD M SD
One Hand 61 . ∗ .
67 0 .
03 0 .
00 45 . ∗† .
96 0 . ∗ . . ∗ .
06 0 .
07 0 .
01 29 . † .
73 0 . † . .
21 5 .
05 0 .
03 0 .
00 31 . ∗ . − . ∗† . ANOVA F , = . p < . F , = . p = . F , = . p < . F , = . p < . Table 3: Mean values and standard deviations for the TCT, Error Rate, NASA Task-Load Index and Bicycle tilt with from respective one-wayANOVAs. * and † show significantly different pairs from post-hoc testing using Tukey HSD, at the p < . level. Note that Bicycle tilt valueswere integrated and standardised, with normality requirements fulfilled (Anderson-Darling test, A = . , p < . ).Figure 8: Normalised Mean Bike tilt values collected in our experi-ment. Error bars show standard error. The values were normalisedto account for between-participant differences in riding style. The participants commented extensively on howTribike and Brotate could be placed in ways that do not conflict withexisting bicycle controls. Brotate was perceived as using existingbicycle control metaphors, such as the rotating bell:
The device [Brotate] is quite similar to a rotating bell, which is wellintegrated into your handlebars.
Further, the participants reported that they appreciated the factthat Brotate integrated into the handlebars and did not add anothervisible device to the bike controls. One participant contrasted thetwo devices:
While it’s [Brotate] a bit more complex than the button device, I likethe fact that it’s part of the handlebars.
In contrast, three of the users in the study were skeptical of inte-grating new devices into bicycle controls. One participant notedthat their core motivation behind not using a device for smartphonecontrol was the fact that he was already proficient with operatingthe smartphone using his hand:
I am used to operating my phone in one hand on a bike and lookingat it, so that’s what I prefer.
Anticipating how users may gain proficiency inusing Tribike and Brotate and considering the use of the devicesover a longer time period was a strong theme in the interviews.Given the novelty of the devices, the participants reported that thedevices appeared complex and training was needed:
I’m not used to operating such a device [Tribike], so, in the beginning,it felt a bit strange.
However, no users mentioned having trouble controlling smart-phone actions with Tribike or Brotate after completing the practice rotate and Tribike: Designing Smartphone Control for Cycling MobileHCI ’20, October 5–8, 2020, Oldenburg, Germany task and completing the experiment. Participants reported havingfamiliarised themselves with the operation of the devices and beingable to issue smartphone commands without looking:
I could operate the device [Brotate] without looking, especially now,when I’m more familiar with the functions.
Relating Tribike and Brotate played a role in how quickly userslearned how to use the devices. Tribike’s use of the headphoneremote layout was helpful to users. Consequently, Brotate wasperceived as more novel and required more time for acquainting:
Pressing the buttons feels more familiar than rotating the device.
This theme describes the users’ perception ofthe hand movements required to operate one’s smartphone whencycling and how these movements impacted their performance inthe tasks. Users felt that a key advantage of Brotate was the lack ofa need to re-position one’s hand to operate it:
I didn’t have to change my grip on the handlebar when using it.[Brotate]
In contrast, participants also remarked that wrist rotation wasnot a movement they associated with riding a bicycle. Some usersreflected that Brotate required movements that were suboptimal:
To make the movement [Rotate up], you need to slightly rotate yourwrist, which felt unnatural.
Finally, some users found it difficult to operate the devices due tothe size of their hands and palms. One user reported that he neededextra movement to hit the buttons on Tribike because of his largehands:
I needed to adjust my grip to operate the device [Tribike], maybebecause my hands are relatively big.
Throughout our design process and evaluation of Tribike and Bro-tate, we observed that the devices benefited interaction while cy-cling. Here, we summarise our findings and provide suggestionsfor future systems which aim to support cyclists.
One-handed smartphone use offers limited performancebenefits at the cost of cognitive load.
Our experiment revealedthat the participants were able to complete the tasks with the threedevices with equal accuracy. One-handed smartphone use was,however, faster than the other methods, which implies that H2 wasnot confirmed. As the majority of the study participants did use theirphones while cycling regularly using the one-handed method, webelieve that this result can be partially attributed to their acquiredproficiency. This was also confirmed by qualitative data in thelearning theme. Further, the results show a significant increase incognitive load when operating the phone with one hand. This resultis congruent with the findings by De Waard et al. [45]. This can beexplained by the fact that Tribike and Brotate do not require visualattention. While the phone can provide richer visual feedback, itappears that it does not aid the users in basic actions. Further, pastwork has shown that strong haptic feedback is particularly usefulwhen designing for dual tasks [3] such as cycling and using a phone.These findings also echo results from past studies in interactionduring physical activity which also showed a need for highly tactilecontrols, e.g. in the context of climbing [41] or running [6, 49]. SUS scale scores from our experiment provide additional motiva-tion for developing alternative means of controlling the smartphonewhile cycling—users ranked issuing commands by hand signifi-cantly lower despite the majority of the being proficient in one-handed operation. This is in contrast with the NASA TLX scores.Thus, H3 is partially confirmed. In contrast with past work [16], ourstudy provides empirical evidence that handlebar-mounted smart-phone controls can effectively reduce the cognitive load requiredto perform smartphone actions while cycling. This fact impliesthat using bicycle-specific controllers limits the attention requiredfor the interaction and leaves more cognitive capacity for traffic.Consequently, future bicycles should offer the means for eyes-freeinteraction with the cyclist’s smartphone. Using Brotate limits undesirable bicycle movements.
TheIMU measurements from our experiment showed that Brotate of-fered superior performance in terms of lateral control of the bicyclethrough the ride compared to the other conditions. This impliesthat H1 is confirmed for Brotate. Such a result suggests that thebicycle was more stable when using Brotate and, consequently, thecyclist had more control over the riding path. While De Waardet al. [45] also observed excessive swaying when the phone washeld in one hand, primarily caused by an altered body position onthe seat (operating a smartphone with one hand caused cyclists tosit more upright), the significant difference between Brotate andTribike should be attributed to other causes. Our design processassured that both devices were located as close as possible to thehandlebar grips and did not require excessive movement to activate.The difference between the two devices can be explained not by therange of movement required to perform the actions using Brotateand Tribike, but the direction of the motion. In the case of Brotate,the rotation move is performed parallel to the axis of motion ofthe bicycle. In contrast, placing the thumb on the top or bottombuttons on Tribike requires movement in a place perpendicularto the direction in which one is pedalling. As a consequence, ourwork implies that future devices for providing input while cyclingshould minimise the amount of movement along the handlebars.The study results for Brotate suggest that techniques that use han-dlebar grips offer most lateral control. Further, understanding themotion required to operate the device in the context of the motionof the bicycle is a key design consideration. New devices for cycling should leverage existing form fac-tors and user-specific configurations.
In our design process, webuilt devices that were explicitly inspired by existing bicycle con-trols. Results in the integration theme show that users appre-ciated the familiar form factors of a bell and headphone remote.Furthermore, we ensured that other controls of the bicycle were notaffected by Brotate and Tribike. The movement theme showed thatthe smartphone controller not only needed to integrate well with ex-isting controls, but it should also be subject to the same user-specificrequirements as other bicycle components. Palm size is a knownlimitation when designing for bicycle controls, which can be ob-served in commercial products such as integrated derailleur-brakecontrols [50]. This finding fits within a larger trend of augmentingexisting equipment in building technology for exertion [5, 22].While this observation sets design constraints for new devicesfor cycling, it also offers opportunities. Firstly, as bicycle technol-ogy evolves, cycling-related controls require less space and force obileHCI ’20, October 5–8, 2020, Oldenburg, Germany Paweł W. Woźniak, Lex Dekker, Francisco Kiss, Ella Velner, Andrea Kuijt, and Stella F. Donker to operate. For example, electronically controlled gear shifting nolonger requires the activation force of older, cable-based systems,and eBikes [1] often feature automatic gear shifting. Consequently,the number of controls required for cycling and their form factor isreduced. Our study illustrates that interactive devices for cyclingcan effectively re-purpose existing control metaphors for purposeslike smartphone control. Future systems for cycling should pri-oritise existing handlebar-based controls over novel methods toeffectively use the users familiarity with those controls and preventundesired operation.
As our work constitutes an exploratory inquiry into smartphonecontrol while cycling, we are aware of certain compromises and limi-tations to which our research is prone. We decided to not investigatecontrolling smartphones mounted in a holder on the handlebars.This decision was motivated by the fact that we wanted to focuson devices that limited the required visual attention. Additionally,past work in the automotive domain showed detrimental effects ofholders [23]. However, we do recognise that smartphones storedin holders may soon appear on the streets more frequently andmay need to be studied. Future research should compare eyes-freedevices like Brotate and Tribike with holder-based solutions interms of performance and safety. We also note that none of theparticipants reported ever using the smartphone buttons while notlooking at the phone, e.g. in oneâĂŹs pocket, which would be a wayto use tactile controls. While we hypothesise that such a solutionwould be safer than looking at the phone, it still requires moving ahand away from the handlebars thus negatively affecting stability.Further, we recognise that our results may be affected by the factthat most participants of the final study were proficient cyclists whoreported regularly using a smartphone while biking. We opted todesign for this user group, because we endeavored to understand thedesign possibilities for more efficient and potentially safer controlsfor them. However, the rising population of cyclists implies thatnew users are constantly introduced to smartphone usage whilecycling. Future research should investigate designing devices fornovice bicycle commuters who may have different needs. We alsonote that the actions we designed for were determined in our survey.However, we need to consider that a declared preference for callsand music control might be due to social desirability bias. Further,our user-centred design process was conducted with current usersof city bikes and thus the design of our devices is limited to thoseusers. More studies of behaviour of different types of cyclists areneeded in traffic research to better inform design.Finally, we believe that a broader discussion is needed about thesocial implications of our design. While increasing safety throughreducing the number of users holding their phones while cyclingis one of our primary motivations, there is a possibility that light-weight eyes free smartphone controls may increase the overallnumber of cyclists using smartphones. Certain social trade-offs areinvolved with cycling technology, such as the acceptability of suchbehaviors and using protection devices [9], and the balance of suchshould be monitored. We hope that social science studies in trafficcan benefit from our understanding of interacting with technology while cycling and effectively affect social acceptance and policy de-cisions. Further, smartphone use per se does have a negative impacton safety [9], and it remains a challenge to manage the cyclists’attention to assure comfort and safety. An emerging question ishow to design technology that would enable effective and safersmartphone use while cycling and, simultaneously, not encouragesmartphone interactions when they are not strictly necessary.
In this paper, we presented the design, implementation and evalua-tion of two smartphone controllers for cyclists—Brotate and Tribike.We first conducted a user survey to identify the most commonsmartphone actions used while cycling. We designed Brotate—adevice that uses handlebars grip rotation and Tribike—a handlebar-mounted device with three buttons in an iterative process. In anexperiment, we compared the two devices with one-handed smart-phone control. The study showed that both devices enabled efficientcontrol while significantly reducing cognitive load. Further, Brotatesignificantly improved lateral control during cycling. Our workconstitutes a structured exploration of smartphone input whilecycling. Our results show that future designers should focus onunderstanding and limiting movement across the handlebars, lever-aging existing interaction metaphors for bicycles and providinguser-specific solutions. We hope that our work inspires further in-quiries into input while cycling, which can contribute to increasingtraffic safety.
REFERENCES [1] Josh Andres, Tuomas Kari, Juerg von Kaenel, and Florian ’Floyd’ Mueller. 2019. Co-riding With My eBike to Get Green Lights. In
Proceedings of the 2019 on DesigningInteractive Systems Conference (San Diego, CA, USA) (DIS ’19) . ACM, New York,NY, USA, 1251–1263. https://doi.org/10.1145/3322276.3322307 [2] Ann Blandford, Dominic Furniss, and Stephann Makri. 2016.
Qualitative HCIresearch: Going behind the scenes . Vol. 9. Morgan & Claypool Publishers. 1–115pages.[3] Jennifer L. Burke, Matthew S. Prewett, Ashley A. Gray, Liuquin Yang, FrederickR. B. Stilson, Michael D. Coovert, Linda R. Elliot, and Elizabeth Redden. 2006.Comparing the Effects of Visual-auditory and Visual-tactile Feedback on UserPerformance: A Meta-analysis. In
Proceedings of the 8th International Conferenceon Multimodal Interfaces (Banff, Alberta, Canada) (ICMI ’06) . ACM, New York,NY, USA, 108–117. https://doi.org/10.1145/1180995.1181017 [4] Anthony Carton. 2012. Design of a Context Aware Signal Glove for Bicycleand Motorcycle Riders. In
Proceedings of the 2012 ACM Conference on UbiquitousComputing (Pittsburgh, Pennsylvania) (UbiComp ’12) . ACM, New York, NY, USA,635–636. https://doi.org/10.1145/2370216.2370341 [5] Ashley Colley, Paweł W. Woźniak, Francisco Kiss, and Jonna Häkkilä. 2018. ShoeIntegrated Displays: A Prototype Sports Shoe Display and Design Space. In
Proceedings of the 10th Nordic Conference on Human-Computer Interaction (Oslo,Norway) (NordiCHI âĂŹ18) . Association for Computing Machinery, New York,NY, USA, 39âĂŞ46. https://doi.org/10.1145/3240167.3240216 [6] Franco Curmi, Maria Angela Ferrario, and Jon Whittle. 2017. Embedding a CrowdInside a Relay Baton: A Case Study in a Non-Competitive Sporting Activity. In
Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems (Denver, Colorado, USA) (CHI ’17) . ACM, New York, NY, USA, 2359–2370. https://doi.org/10.1145/3025453.3025551 [7] Alexandru Dancu, Zlatko Franjcic, and Morten Fjeld. 2014. Smart Flashlight:Map Navigation Using a Bike-mounted Projector. In
Proceedings of the SIGCHIConference on Human Factors in Computing Systems (Toronto, Ontario, Canada) (CHI ’14) . ACM, New York, NY, USA, 3627–3630. https://doi.org/10.1145/2556288.2557289 [8] Alexandru Dancu, Velko Vechev, Adviye Ayça Ünlüer, Simon Nilson, OscarNygren, Simon Eliasson, Jean-Elie Barjonet, Joe Marshall, and Morten Fjeld.2015. Gesture Bike: Examining Projection Surfaces and Turn Signal Systems forUrban Cycling. In
Proceedings of the 2015 International Conference on InteractiveTabletops & Surfaces (Madeira, Portugal) (ITS ’15) . ACM, New York, NY, USA,151–159. https://doi.org/10.1145/2817721.2817748 rotate and Tribike: Designing Smartphone Control for Cycling MobileHCI ’20, October 5–8, 2020, Oldenburg, Germany [9] Dick de Waard, Paul Schepers, Wieke Ormel, and Karel Brookhuis. 2010.Mobile phone use while cycling: Incidence and effects on behaviour andsafety.
Ergonomics
53, 1 (2010), 30–42. https://doi.org/10.1080/00140130903381180 [10] Jennifer Dill and Theresa Carr. 2003. Bicycle commuting and facilities in majorUS cities: if you build them, commuters will use them.
Transportation ResearchRecord
Transport Reviews
36, 1 (2016), 1–8. https://doi.org/10.1080/01441647.2015.1114271 [12] Debjyoti Ghosh, Pin Sym Foong, Shengdong Zhao, Di Chen, and Morten Fjeld.2018. EDITalk: Towards Designing Eyes-free Interactions for Mobile Word Pro-cessing. In
Proceedings of the 2018 CHI Conference on Human Factors in ComputingSystems (Montreal QC, Canada) (CHI ’18) . ACM, New York, NY, USA, Article 403,10 pages. https://doi.org/10.1145/3173574.3173977 [13] C. Goldenbeld, M. Houtenbos, E. Ehlers, and D. De Waard. 2012. The use and riskof portable electronic devices while cycling among different age groups.
Journalof Safety Research
43, 1 (2012), 1 – 8. https://doi.org/10.1016/j.jsr.2011.08.007 [14] Weixi Gu, Yunxin Liu, Yuxun Zhou, Zimu Zhou, Costas J. Spanos, and Lin Zhang.2017. BikeSafe: Bicycle Behavior Monitoring via Smartphones. In
Proceedings ofthe 2017 ACM International Joint Conference on Pervasive and Ubiquitous Com-puting and Proceedings of the 2017 ACM International Symposium on WearableComputers (Maui, Hawaii) (UbiComp ’17) . ACM, New York, NY, USA, 45–48. https://doi.org/10.1145/3123024.3123158 [15] Lucas Harms, Luca Bertolini, and Marco te Brömmelstroet. 2014. Spatial and socialvariations in cycling patterns in a mature cycling country exploring differencesand trends.
Journal of Transport and Health
1, 4 (2014), 232–242. Walking andCycling: The contributions of health and transport geography.[16] Wolfgang Hochleitner, David Sellitsch, Daniel Rammer, Andrea Aschauer, ElkeMattheiss, Georg Regal, and Manfred Tscheligi. 2017. No Need to Stop: ExploringSmartphone Interaction Paradigms While Cycling. In
Proceedings of the 16thInternational Conference on Mobile and Ubiquitous Multimedia (Stuttgart, Ger-many) (MUM ’17) . ACM, New York, NY, USA, 177–187. https://doi.org/10.1145/3152832.3152871 [17] Brianna Jean Huxtable, Carlo Ka-Ho Lai, Johnson Wen Jun Zhu, Paulina Mun-YeeLam, Yeseul Tracy Choi, Carman Neustaedter, and Greg J. Corness. 2014. Ziklo:Bicycle Navigation Through Tactile Feedback. In
CHI ’14 Extended Abstractson Human Factors in Computing Systems (Toronto, Ontario, Canada) (CHI EA’14) . ACM, New York, NY, USA, 177–178. https://doi.org/10.1145/2559206.2579481 [18] Masao Ichikawa and Shinji Nakahara. 2008. Japanese High School Students’Usage of Mobile Phones While Cycling.
Traffic Injury Prevention
9, 1 (2008),42–47. https://doi.org/10.1080/15389580701718389 [19] Michael D. Jones, Zann Anderson, Jonna Häkkilä, Keith Cheverst, and FlorianDaiber. 2018. HCI Outdoors: Understanding Human-Computer Interaction inOutdoor Recreation. In
Extended Abstracts of the 2018 CHI Conference on Hu-man Factors in Computing Systems (Montreal QC, Canada) (CHI EA ’18) . ACM,New York, NY, USA, Article W12, 8 pages. https://doi.org/10.1145/3170427.3170624 [20] Shinhye Joo, Gunwoo Lee, Cheol Oh, and Keechoo Choi. 2018. A New Approachto Analyzing Cycling Stability Using an Inertial Measurement Unit Sensor.
KSCEJournal of Civil Engineering
22, 12 (2018), 5195–5202.[21] Francisco Kiss, Robin Boldt, Bastian Pfleging, and Stefan Schneegass. 2018. Navi-gation Systems for Motorcyclists: Exploring Wearable Tactile Feedback for RouteGuidance in the Real World. In
Proceedings of the 2018 CHI Conference on Hu-man Factors in Computing Systems (Montreal QC, Canada) (CHI ’18) . Associa-tion for Computing Machinery, New York, NY, USA, Article Paper 617, 7 pages. https://doi.org/10.1145/3173574.3174191 [22] Francisco Kiss, Paweł W. Woźniak, Felix Scheerer, Julia Dominiak, Andrzej Ro-manowski, and Albrecht Schmidt. 2019. Clairbuoyance: Improving DirectionalPerception for Swimmers. In
Proceedings of the 2019 CHI Conference on Hu-man Factors in Computing Systems (Glasgow, Scotland Uk) (CHI âĂŹ19) . As-sociation for Computing Machinery, New York, NY, USA, 1âĂŞ12. https://doi.org/10.1145/3290605.3300467 [23] Tuomo Kujala, Hilkka Grahn, and Niina Holmstedt. 2015. Hand-Held Textingis Less Distracting Than Texting with the Phone in a Holder: Anyway, Don’TDo It. In
Proceedings of the 19th International Academic Mindtrek Conference (Tampere, Finland) (AcademicMindTrek ’15) . ACM, New York, NY, USA, 98–105. https://doi.org/10.1145/2818187.2818270 [24] Haiko Luepsen. 2017. The aligned rank transform and discrete variables: A warn-ing.
Communications in Statistics - Simulation and Computation
46, 9 (Oct. 2017),6923–6936. https://doi.org/10.1080/03610918.2016.1217014 [25] Joe Marshall and Paul Tennent. 2013. Mobile Interaction Does Not Exist. In
CHI’13 Extended Abstracts on Human Factors in Computing Systems (Paris, France) (CHI EA ’13) . ACM, New York, NY, USA, 2069–2078. https://doi.org/10.1145/2468356.2468725 [26] Daniel A Matre. 1982. Motorcycle handgrip control. US Patent App. 06/154,698.[27] Andrii Matviienko, Swamy Ananthanarayan, Shadan Sadeghian Borojeni, Yan-nick Feld, Wilko Heuten, and Susanne Boll. 2018. Augmenting Bicycles and Helmets with Multimodal Warnings for Children. In
Proceedings of the 20th In-ternational Conference on Human-Computer Interaction with Mobile Devices andServices (Barcelona, Spain) (MobileHCI ’18) . ACM, New York, NY, USA, Article15, 13 pages. https://doi.org/10.1145/3229434.3229479 [28] Andrii Matviienko, Swamy Ananthanarayan, Abdallah El Ali, Wilko Heuten,and Susanne Boll. 2019. NaviBike: Comparing Unimodal Navigation Cues forChild Cyclists. In
Proceedings of the 2019 CHI Conference on Human Factors inComputing Systems (Glasgow, Scotland Uk) (CHI ’19) . ACM, New York, NY, USA,Article 620, 12 pages. https://doi.org/10.1145/3290605.3300850 [29] Eleonora Mencarini, Chiara Leonardi, Antonella De Angeli, and Massimo Zanca-naro. 2016. Design Opportunities for Wearable Devices in Learning to Climb. In
Proceedings of the 9th Nordic Conference on Human-Computer Interaction (Gothen-burg, Sweden) (NordiCHI ’16) . ACM, New York, NY, USA, Article 48, 10 pages. https://doi.org/10.1145/2971485.2971509 [30] Calkin S. Montero, Jason Alexander, Mark T. Marshall, and Sriram Subrama-nian. 2010. Would You Do That?: Understanding Social Acceptance of GesturalInterfaces. In
Proceedings of the 12th International Conference on Human Com-puter Interaction with Mobile Devices and Services (Lisbon, Portugal) (MobileHCI’10) . ACM, New York, NY, USA, 275–278. https://doi.org/10.1145/1851600.1851647 [31] Alexander Ng, Stephen A. Brewster, Frank Beruscha, and Wolfgang Krautter.2017. An Evaluation of Input Controls for In-Car Interactions. In
Proceedingsof the 2017 CHI Conference on Human Factors in Computing Systems (Denver,Colorado, USA) (CHI ’17) . ACM, New York, NY, USA, 2845–2852. https://doi.org/10.1145/3025453.3025736 [32] Max Pfeiffer, Dagmar Kern, Johannes Schöning, Tanja Döring, Antonio Krüger,and Albrecht Schmidt. 2010. A Multi-touch Enabled Steering Wheel: Exploringthe Design Space. In
CHI ’10 Extended Abstracts on Human Factors in ComputingSystems (Atlanta, Georgia, USA) (CHI EA ’10) . ACM, New York, NY, USA, 3355–3360. https://doi.org/10.1145/1753846.1753984 [33] Martin Pielot, Benjamin Poppinga, Wilko Heuten, and Susanne Boll. 2012. Tacticy-cle: Supporting Exploratory Bicycle Trips. In
Proceedings of the 14th InternationalConference on Human-computer Interaction with Mobile Devices and Services (SanFrancisco, California, USA) (MobileHCI ’12) . ACM, New York, NY, USA, 369–378. https://doi.org/10.1145/2371574.2371631 [34] Benjamin Poppinga, Martin Pielot, and Susanne Boll. 2009. Tacticycle: A TactileDisplay for Supporting Tourists on a Bicycle Trip. In
Proceedings of the 11thInternational Conference on Human-Computer Interaction with Mobile Devices andServices (Bonn, Germany) (MobileHCI ’09) . ACM, New York, NY, USA, Article 41,4 pages. https://doi.org/10.1145/1613858.1613911 [35] Sasank Reddy, Katie Shilton, Gleb Denisov, Christian Cenizal, Deborah Estrin,and Mani Srivastava. 2010. Biketastic: Sensing and Mapping for Better Biking. In
Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (Atlanta, Georgia, USA) (CHI ’10) . ACM, New York, NY, USA, 1817–1820. https://doi.org/10.1145/1753326.1753598 [36] Julie Rico and Stephen Brewster. 2010. Usable Gestures for Mobile Interfaces:Evaluating Social Acceptability. In
Proceedings of the SIGCHI Conference on HumanFactors in Computing Systems (Atlanta, Georgia, USA) (CHI ’10) . ACM, New York,NY, USA, 887–896. https://doi.org/10.1145/1753326.1753458 [37] Duncan Rowland, Martin Flintham, Leif Oppermann, Joe Marshall, Alan Cham-berlain, Boriana Koleva, Steve Benford, and Citlali Perez. 2009. UbikequitousComputing: Designing Interactive Experiences for Cyclists. In
Proceedings of the11th International Conference on Human-Computer Interaction with Mobile Devicesand Services (Bonn, Germany) (MobileHCI ’09) . ACM, New York, NY, USA, Article21, 11 pages. https://doi.org/10.1145/1613858.1613886 [38] Paul Schepers and K Klein Wolt. 2012. Single-bicycle crash types and character-istics.
Cycling Research International
2, 1 (2012), 119–135.[39] Patrick A. Singleton. 2019. Walking (and cycling) to well-being: Modal and otherdeterminants of subjective well-being during the commute.
Travel Behaviour andSociety
16 (2019), 249 – 261. https://doi.org/10.1016/j.tbs.2018.02.005 [40] Gábor Sörös, Florian Daiber, and Tomer Weller. 2013. Cyclo: A Personal BikeCoach Through the Glass. In
SIGGRAPH Asia 2013 Symposium on Mobile Graph-ics and Interactive Applications (Hong Kong, Hong Kong) (SA ’13) . ACM, NewYork, NY, USA, Article 99, 4 pages. https://doi.org/10.1145/2543651.2543660 [41] Daniel Spelmezan, Mareike Jacobs, Anke Hilgers, and Jan Borchers. 2009. TactileMotion Instructions for Physical Activities. In
Proceedings of the SIGCHI Confer-ence on Human Factors in Computing Systems (Boston, MA, USA) (CHI ’09) . ACM,New York, NY, USA, 2243–2252. https://doi.org/10.1145/1518701.1519044 [42] Haska Steltenpohl and Anders Bouwer. 2013. Vibrobelt: Tactile Navigation Sup-port for Cyclists. In
Proceedings of the 2013 International Conference on IntelligentUser Interfaces (Santa Monica, California, USA) (IUI ’13) . ACM, New York, NY,USA, 417–426. https://doi.org/10.1145/2449396.2449450 [43] Kathryn Terzano. 2013. Bicycling safety and distracted behavior in The Hague,the Netherlands.
Accident Analysis & Prevention
57 (2013), 87 – 90. https: obileHCI ’20, October 5–8, 2020, Oldenburg, Germany Paweł W. Woźniak, Lex Dekker, Francisco Kiss, Ella Velner, Andrea Kuijt, and Stella F. Donker //doi.org/10.1016/j.aap.2013.04.007 [44] Velko Vechev, Alexandru Dancu, Simon T. Perrault, Quentin Roy, Morten Fjeld,and Shengdong Zhao. 2018. Movespace: On-body Athletic Interaction for Runningand Cycling. In
Proceedings of the 2018 International Conference on AdvancedVisual Interfaces (Castiglione della Pescaia, Grosseto, Italy) (AVI ’18) . ACM, NewYork, NY, USA, Article 28, 9 pages. https://doi.org/10.1145/3206505.3206527 [45] Dick De Waard, Ben Lewis-Evans, Bart Jelijs, Oliver Tucha, and Karel Brookhuis.2014. The effects of operating a touch screen smartphone and other com-mon activities performed while bicycling on cycling behaviour.
Transporta-tion Research Part F: Traffic Psychology and Behaviour
22 (2014), 196 – 206. https://doi.org/10.1016/j.trf.2013.12.003 [46] Wouter Walmink, Danielle Wilde, and Florian ’Floyd’ Mueller. 2013. DisplayingHeart Rate Data on a Bicycle Helmet to Support Social Exertion Experiences.In
Proceedings of the 8th International Conference on Tangible, Embedded andEmbodied Interaction (Munich, Germany) (TEI ’14) . ACM, New York, NY, USA,97–104. https://doi.org/10.1145/2540930.2540970 [47] Jacob O. Wobbrock, Leah Findlater, Darren Gergle, and James J. Higgins. 2011.The Aligned Rank Transform for Nonparametric Factorial Analyses Using OnlyAnova Procedures. In
Proceedings of the SIGCHI Conference on Human Factorsin Computing Systems (Vancouver, BC, Canada) (CHI ’11) . ACM, New York, NY, USA, 143–146. https://doi.org/10.1145/1978942.1978963 [48] Jacob O. Wobbrock, Meredith Ringel Morris, and Andrew D. Wilson. 2009. User-defined Gestures for Surface Computing. In
Proceedings of the SIGCHI Conferenceon Human Factors in Computing Systems (Boston, MA, USA) (CHI ’09) . ACM,New York, NY, USA, 1083–1092. https://doi.org/10.1145/1518701.1518866 [49] Paweł Woźniak, Kristina Knaving, Staffan Björk, and Morten Fjeld. 2015. RUFUS:Remote Supporter Feedback for Long-Distance Runners. In
Proceedings of the17th International Conference on Human-Computer Interaction with Mobile Devicesand Services (Copenhagen, Denmark) (MobileHCI ’15) . ACM, New York, NY, USA,115–124. https://doi.org/10.1145/2785830.2785893 [50] Kazuhisa Yamashita. 2000. Brake lever stroke adjusting mechanism. US Patent6,112,614.[51] Desheng Zhao, Jeannie Su Ann Lee, Chek Tien Tan, Alexandru Dancu, SimonLui, Songjia Shen, and Florian ’Floyd’ Mueller. 2019. GameLight - Gamificationof the Outdoor Cycling Experience. In
Companion Publication of the 2019 onDesigning Interactive Systems Conference 2019 Companion (San Diego, CA, USA) (DIS ’19 Companion) . ACM, New York, NY, USA, 73–76. https://doi.org/10.1145/3301019.3325151 [52] Kathryn Zickuhr. 2012. Three-quarters of smartphone owners use location-basedservices.