 Yeah Okay Do you hear me? I think yes as We are speaking about Some Additional hardware automation in usability research It should be noted that obviously measurements of some biometric parameters in the usability are No way the new approach there were extremely expensive devices affordable by large companies and laboratories and Obviously even if you were an An employer of such large company or if you were working in such laboratory you were unable to use Large amount of such devices in parallel because well you had few of them and not tense And Last several years We have seen new wave of Really powerful biometric devices Which are user grade and Extremely cheap mass market production makes them cheap and of Course they are targeted at entertainment and fitness applications as their primary goal But still they are producing measurements of biometric parameters With enough precision which is well It's a really attractive idea to use these data for something useful related to usability and Let us speak about how Can we do it? So the typic scheme of testing with using of some biometrics looks like this you have a user a user is typing text moving his well mouse or Doing some other things with the software you track time Possibly you track errors which user makes while working you Monitor heart rate. It's the easiest parameter to track Also, you may track attention level with the Consumer grade encephalographer or you can Even use an eye tracking technologies because computer games have reached this level of Biometrics eye trackers for gamers are not so widely used as joysticks and some other Hmm game parts, but still they are produced and they are in our group of biometric devices We are going to discuss What can be measured For the sake of usability research galvanic skin response unfortunately this parameter is not measured by consumer grade gadgets typically and It's a pity Because well, it's the thing which is really easy to measure and Rather informative, but we have heart rate a Lot of fitness trackers are monitoring it Possibly we can use to fitness trackers or some other tools to Monitor blood pressure, but it's rarely used We have enough chip EEG devices. I mean an electro encephalography waves measuring We have enough tools to measure kinematic activity and as I have already mentioned we have gaze detection What did we use ourselves? In some usability tests express tests comparisons For electro encephalography devices The simplest One is a Neuro skymind wave. It has a mindset modification, which is Non-durable Just the construction was not very successful. You will break it rather soon less durable electro encephalography devices are Is so-called necomimia ears, which are well cat ears Placed on your head for the entertainment purposes, which stands still when you are Concentrated on something they were really popular. They used the same technology of The Neuro sky we are using for research Yeah, but well We do not discard them here and more comprehensive EEG devices from emulative Frankly speaking these are more popular among researchers epoch and epoch plus the newer one and The most new motive inside little bit simpler But still highly comfortable fitness trackers our own experience Well Fitbit charge from our own point of view Fitbit Heart rate measurements where the most precise we were using in our research unfortunately open source tools are not very good in getting data from newer models of Fitbit fitness trackers and really easy to get data from our Xiaomi me bands and MS fit which is Practically the same device from the data acquisition point of view If speaking about it records We have successfully used to be wrecks and to be foresee both are targeted at gamers I will speak about this a little bit later Testing schemes which we can use The simplest one is an individual testing mode So you have a computer with some software you are Make user to do any actions you write down logs Rather typical logs from the testing program may be logs from his biometric devices the ones you have esquired and we're able to put on user and The more interesting where it is parallel testing mode You just run the same set of software and More or less same set of devices on several users at once you collect logs from all of them and You may run all Data acquisition scripts in parallel by starting script Or you can make something more comprehensive. We have created our own open source UX dump suit to carry on such parallel testing Which is well the link is on the bottom but Aza you are able to have extremely cheap me bands or something like this and This will allow you to have representative number of tests in a reasonable time the main problem of UX research Is typically a small set of users you are tested well, obviously some more traditional ways like questionaries and Video recording and so on are really time-consuming and If you are doing some biometric testing in parallel You get a lot of data in a short period of time, which is really attractive of course, well nothing to speak about the devices themselves Just how row hat rate or row EEG signal from some frequency band are looking just an illustrative slide The most interesting part is getting data from the device Devices are cheap and affordable. You have a lot of them. Maybe But now you remember that their primary goal Is nothing about UX research They show some pulse or anything on the screen, but Typically not share this data with you Several approaches are available to get Something from them the ideal Variant is using universal API like Qt Bluetooth or blues. Most of them are Working with a Bluetooth protocol Unfortunately, it's as rare case as it only possible API from the device vendor two examples are present here he'd fit bit web API which is placed on a separate slide because It's something terrible and some special SDK like For example, Toby ice rakers one, which is also terrible, but it is not placed on a separate slide Or You can use special open source tools, which were written by Not by the developers of the original device, but are able to get data from it a Lot of them exist and that's the most preferable variant if you have choice for example might wave Headset Is sharing comfortably sharing data. We are the puzzle box synapse project Initially NeroSky wanted some money for the simple suit, which was just able to save raw data, but I Don't think they sell it anymore because good open source tools have superseded it Obviously the same situation is true for fitness trackers and the last one is file obstruction It's well, not real-time data acquisition, but it also is a good thing to use I would say that the other are really good pulse matters. We were using our Garmin fitness trackers and We didn't manage to get data from them in real-time But if you put them into USB slot, they are pretending to be a flash drive with logs so it's Even simpler than we wanted it to be for our aim goals Little bit more about the vendors provided a P Here you see the strange scheme of dealing with Fitbit fitness tracker. It's is precise well few words about the precision of fitness trackers you may have heard that Fitness trackers are not very precise in measuring heart rate but They are not very good for Active physical trainings because when you wave your hands they may lose contact But actually our users are not so physically active and if there is contact There is a good probability that this contact with the optical sensor will not be lost But still fitness Fitbit fitness trackers are the better ones We were able to use to get data in real-time or almost in real-time But the scheme is terrible at first you use open source Galileo tool to download encrypted data, which you cannot decrypt yourself then Galileo sends this batch of data to the fitness web Trackers vendor website Fitbit.com and then you use your own application Which is authorized to get data from the website and the worst is Toby SDK There is old SDK, which is unavailable for downloading Which works with all the eye trackers and there is a new SDK, which is available to downloading and Which used which is named to be usable with all eye trackers But actually usable only with the last model Oh, sorry So You still have obstacles but It's not too difficult to overcome them So you can calculate also duration of the actions You can calculate number of errors heart rate attention Level a little bit more about the attention level of the headset it may be Precalculated metrics from the vendor which are used to measure mind concentration in games or you can calculate them yourself Precalculated are better Because they are faster Calculated by yourself are more scientific a little bit Gaze detection a little bit more after you get coordinates of gaze fixation from your Eye tracker The best tool from our point to deal with this data is a new Octave Which Octave is actually a fork of Matlab and You can use three lines of code to plot of these gaze map or you can use a little bit more Comprehensive script the whole script is on the slide to get nice hitmaps Which as you of course know Show which parts of the screen were mostly looked during the test Types of test tasks Serious of different type operations in one program user has a list of tasks to do and By this way you can compare Two versions of one program or you can compare maybe two different programs of the same area of application and You can use different approach one sequence of routine operations for example a lot of clicks or Some key types or some other actions Few examples really few of them One is comparison of the office interfaces Here you see three office suits Two versions of LibreOffice with the top panel and side panel and ribbons from one of the early versions of Microsoft Office with ribbons How the final results would look like for example in our case For over complicated interfaces like office suits are half not for Emina If you have simple interface ribbons are good for nothing if you have over complicated interface then more than half of your users work faster with ribbons and Smaller parts work faster with top or side panel also You can compare How The state of the user is different when he works with Not the best suits for himself Red background means that These users are better with top panel blue background means that these users are better with ribbons and you can see that physical load is higher and Mind concentration may be higher and so on and you can compare errors The other example of The second approach This is example of comparing on-screen keyboards with the hardware keyboard of some asus small asus typical asus laptop You can see the same approach as first of all you compare speed you compare errors and you can see if someone had higher Errors rate lower speed, but higher mind concentration in this case perhaps this person was hardly walking with the less comfortable tool and If the speed and mind concentration are High, it's okay. If the speed is low and the mind concentration is high. It's the worst case for example The last slide I think is it okay well So the conclusions are the following ones consumer grade biometric turned out to be enough major to produce us data usable in UX UI comparisons Even some vendor driving metrics are good if you compare Several suits then Something good is written here about biometric indicators indicators that well they are good. I think and Obviously we can Use these to save our time. Well, I'll save our time So, thank you Yeah Who can Good idea the better objects are users which Which have the same level of ignorance in Programs you are testing often then Of course, you should take into account external destructors You should randomize the order of suits You should do some other things which are rather typical for UX research and What you can simplify along this approach you can Compare the results of the same user with several types of software in this case His well state or skills or NSNLs are relatively the same Yeah, of course But if you do traditional UX research, you have the same problems and you still have to take these factors into attention Yeah, thank you