No description available.

No description available.

No description available.

Det här examensarbetet bygger vidare på en tidigare konstruerad prototyp, ett interaktivt konstprojekt där två plattor, placerade på en båge, med vattenmunstycken i mitten styrs med en joystick. Plattorna följer joystickens rörelse med hjälp av servomotorer.Målet i detta examensarbete var att koppla åskådarens smarta mobiltelefon till konstprojektet. Åskådaren ska kunna styra konstprojektet genom att vinkla sin mobiltelefon. Även vattenflödet ska kunna kontrolleras och all interaktion ska ske via trådlös kommunikation.Målet har uppnåtts genom att användaren installerar en applikation som skapats för operativsystemet Android. Applikationen skickar information om mobiltelefonens aktuella vinkelposition till en server via ett trådlöst nätverk. Servern skickar i sin tur vidare denna information till en Wifi-modul kopplad på ett arduinokort som ställer in plattornas servomotorer i rätt läge. Rapporten beskriver hur tekniken för att nå målet fungerar och illustrerar även hur prototypens design och funktionalitet har förändrats för att få ett mer effektfullt intryck. / In this thesis we further develop an existing prototype. The prototype is an interactive art project where two plates with water nozzles in the middle are placed at each end of an arc. These two plates can be controlled directly with a joystick. The plates are following the motion of the joystick using servo motors. Our main goal is to allow spectators of the art project to remotely control the plates and the water flow by using their own mobile phone.Our goal has been achieved by creating an Android application that the user can download and install directly on their mobile phone. The application automatically sends information about the angular position of the phone to a server. Then the server forwards this information to a Wifi-module connected to an Arduino board which is moving the servo motors, attached to the plates, in the correct position.This thesis describes the technique used within the project and also illustrates how the design and functionality of the prototype has been changed to make a more striking impression on the audience.

Human activity recognition (HAR) is a growing area within machine learning as the possible applications are vast, especially with the growing amount of collectable sensor data as Internet of Things-devices are becoming more accessible. This project aims to contribute to HAR by developing two supervised machine learning algorithms that are able to distinguish between four different human activities. We collected data from the tri-axial accelerometer in two different smartphones while doing these activities, and put together a dataset. The algorithms that were used was a convolutional neural network (CNN) and a support vector machine (SVM), and they were applied to the dataset separately. The results show that it is possible to accurately classify the activities using the algorithms and that a short time window of 3 seconds is enough to classify the activities with an accuracy of over 99% with both algorithms. The SVM outperformed the CNN slightly. We also discuss the result and continuations of this study. / Mlinsklig aktivitetsigenkanning (HAR) lir ett vlixande omrade inom maskininllirning da de mojliga applikationerna lir stora, speciellt med den vlixande mangd insamlingsbar sensordata da Internet of Things-enheter blir mer atkomliga. Detta projekt siktar pa att bidra till HAR genom att utveckla tva algoritmer som kan urskilja mellan fyra olika mlinskliga aktiviteter. Vi samlade in data fran den treaxlade accelerometern i tva olika smarta telefoner medans dessa aktiviteter utfordes, och satte ihop ett dataset. Algoritmerna som anvlindes var ett faltande neuralt nlitverk och en stodvektormaskin, och de applicerades separat pa datasetet. Resultaten visar att det lir mojligt att med slikerhet klassificera aktiviteterna genom att anvlinda dessa algoritmer och att ett kort tidsfonster med 3 sekunder av data lir tillrlickligt for att klassificera med en slikerhet pa over 99% med bada algoritmerna. Stodvektormaskinen presterade nagot blittre an det neurala nlitverket. Vi diskuterar liven resultatet och fortsatta studier. / Kandidatexjobb i elektroteknik 2021, KTH, Stockholm

Internet of Things devices, such as smartphonesand smartwatches, are currently becoming widely accessible andprogressively advanced. As the use of these devices steadilyincreases, so does the access to large amounts of sensory data.In this project, we developed a system that recognizes certainactivities by applying a linear classifier machine learning modelto a data set consisting of examples extracted from accelerometersensor data. We obtained the data set by collecting data from amobile device while performing commonplace everyday activities.These activities include walking, standing, driving, and ridingthe subway. The raw accelerometer data was then aggregatedinto data points, consisting of several informative features. Thecomplete data set was subsequently split into 80% training dataand 20% test data. A machine learning algorithm, in this case,a support vector machine, was presented with the training dataset and finally classified all test data with a precision higher than90%. Hence, meeting our set objective to build a service with acorrect classification score of over 90%.Human activity recognition has a large area of application,including improved health-related recommendations and a moreefficiently engineered system for public transportation. / Internet of Things-enheter, så som smarta telefoner och klockor, blir numera allt mer tillgängliga och tekniskt avancerade. Eftersom användningen av dessa smarta enheter stadigt ökar, ökar också tillgången till stora mängder data från sensorer i dessa enheter. I detta projekt utvecklade vi ett system som känner igen vissa aktiviteter genom att tillämpa en linjär klassificerande maskininlärningsmodell på en uppsättning data som extraherats från en accelerometer, en sensor i en smart telefon. Datauppsättningen skapades genom att samla in data från en smart telefon medan vi utförde vardagliga aktiviteter, så som promenader, stå stilla, köra bil och åka tunnelbana. Rå accelerometerdata samlades in och gjordes om till datavektorer innehållandes statistiska mått. Den totala datauppsättningen delades sedan upp i 80% träningsdata och 20% testdata. En maskininlärningsalgoritm, i detta fall en supportvektormaskin, introducerades med träningsdatan och klassificerade slutligen testdatan med en precision på över 90%. Därmed uppfylldes vårt uppsatta mål med att bygga en tjänst med en korrekt klassificering på över 90%. Igenkänning av mänsklig aktivitet har ett stort användningsområde, och kan bidra till förbättrade hälsorekommendationer och en mer effektiv kollektivtrafik. / Kandidatexjobb i elektroteknik 2020, KTH, Stockholm

No description available.

Background: Less than half of Americans meet the recommendation of leisure-time physical activity (PA) of 150 minutes (CDC, 2012). A number of efficacious programs have been developed, and those that are based on theory are more effective. However, it is often difficult to determine the mechanisms of effect through meditational analyses. This is particularly an issue when a theory-based intervention is effective but theoretically hypothesized meditational relationships are not found. One reason for the lack of relationship could be the methods used to assess different theoretical constructs. The Theory of Planned Behavior (TPB) is one model used in the exercise and PA research domain which generally employs traditional fixed-graded measures of proposed theoretical mediators (e.g., strongly disagree to strongly agree response sets). More recent research provides initial evidence that using continuous-open scaling (e.g., ratio-level scaling; days or time/per week rather than agreement scales) has shown the superiority in measurement accuracy demonstrated by stronger relationships between the theoretical constructs and behavioral outcomes when compared to other scaling types. However, continuous open scaling has only been applied to correlational studies and there is no evidence that this scaling procedure results in measures that are sensitive to change or are related to both self-reported and objectively measured PA. Primary Aims: The primary aims of this study were to determine 1) the sensitivity to change of TPB constructs using different response sets and self-report and accelerometer assessed PA, and 2) if TPB constructs measured using the different response-sets have differential prediction of PA measured using self-report and accelerometry. Methods: Forty-six young adults were recruited to complete 13-item measure of TPB constructs using both fixed-graded and continuous-open scales as well as Godin's Leisure Time Exercise Questionnaire at 3 time (T1=Baseline, T2=End of week 1, T3=End of week 2) points over 2-weeks. Potential participants were excluded if they engaged in PA of 150 minutes or more per week. Inclusion criteria were the ability to perform moderate intensity PA and aged 18-25 years old. The order of different scales for the TPB constructs was randomly and evenly assigned within each condition. Participants were asked to wear an accelerometer for 2 weeks; one week prior to the action planning activity and one week after. To determine the sensitivity to change of the measures, participants were randomly assigned to either complete a personal action plan for physical activity (AP) or not (Control). Due to the exploratory nature of the pilot we set the significance level for all tests at p<0.10. Results: In general responses to the continuous open versus fixed closed items, at each time point, resulted in significantly (p<.05) lower perceptions of attitude (instrumental attitudes T1=4.4, T2=4.2, T3=4.3 versus time1= 6.2, T2=6.4, T3=6.3; affective attitudes T1=3.6, T2=3.5, T3=3.9 versus time1= 5.0, T2=5.0, T3=4.8), subjective norm ( T1=3.4, T2=3.3, T3=2.4 versus time1= 5.3, T2=5.2, T3=5.3), perceptions of control ( T1=3.2, T2=3.7, T3=3.9 versus time1= 4.6 T2=4.9, T3=5.2), and intention (T1=1.5, T2=1.8, T3=2.4 versus time1= 4.8, T2=5.1, T3=5.2). In regards to sensitivity to change continuous open and fixed closed measures of instrumental attitudes, subjective norms did not demonstrate significant changes as a result of action planning. Affective attitudes measured by the continuous-open scale, but not when measured by the fixed-closed scale, showed change over time regardless of condition. Perceived behavioral control measured using the continuous-open scale increased for AP participants by approximately 0.5 compared to control participants change of approximately 0.1 (p=.09). A similar pattern was found with intention in that changes in the continuous-open scale were significant (AP=0.9; control=0.2; p=0.07). No other scales showed significant sensitivity to change. Self-reported PA increased significantly for AP participants (81-16 minutes per week of PA) when compared to control participants (87 +/- 19 to 75 +/- 17 minutes per week of PA; p<0.1). Same pattern of differences was shown between AP (65 +/- 13 to 107 +/- 15 minutes per week of PA) and control (70 +/- 14 to 65 +/- 16 minutes per week of PA) participants (p<0.05). Conclusions: Continuous open scaling have significant correlations with all constructs along with affective attitude and intention being correlated with the actual reported exercise behavior over fixed graded scaling. This data sheds further insight into the different response sets of the TPB in application to exercise domain within a sedentary, young population. The lack of a significant difference may be due to the small sample size. Further research should investigate the role of the personalized action plan utilizing a larger sample size and the correlation of the TPB with intention and actual exercise behavior within an intervention. / Master of Science

Detta arbete undersöker om mätning av vibrationer kan användas som övervakning av slitdelar för jordbearbetande redskap som använder sig av pinnar för att bearbeta jorden som till exempel pinnkultivatorer, kombinationskultivatorer och harvar. Syftet med denna typ av övervakningen är primärt att använda på framtidens autonoma maskiner som kräver uppkopplade sensorer för att inte arbeta med defekta redskap. Arbetet analyserar vibrationsdata som är insamlad med hjälp av accelerometrar. En serie mätningar på pinnar med hela respektive slitna spetsar har genomförts. Datan har sedan analyserats genom att undersöka förekomst av dominerande frekvenser samt deras amplitud. Arbetets resultat indikerar att slitagedetektering av harvspetsar genom att studera vibrationer i systemet är möjligt. För att kunna avgöra om kvantifiering av slitagegraden är möjligt krävs dock ett betydligt mer omfattande genomförande med bland annat ett större antal mätningar och sensorer samt en mer djupgående analys. / This work investigates whether vibration measurement can be used to monitor wear parts for tillage implements that use tines to work the soil, such as tine cultivators, combination cultivators, and tine harrows. The purpose of this type of monitoring is primarily to be used on future autonomous machines that require connected sensors to avoid working with defective implements.  The work analyzes vibration data collected using accelerometers. A series of measurements on tines with intact versus worn points has been conducted. The data has then been analyzed by examining the presence of dominant frequencies and their amplitudes.  The result of the work indicate that wear detection of harrow points by studying vibrations in the system is possible. However, to determine whether quantification of the wear degree is feasible, a significantly more extensive implementation is required, including a larger amount of measurements and sensors, as well as a more in-depth analysis.

Yes / To investigate factors associated with movement behaviours among White British (WB) and South Asian (SA) children aged 6-8 years during school terms and holidays. Cross-sectional. Three primary schools from the Bradford area, UK. One hundred and sixty WB and SA children aged 6-8 years. Sedentary behaviour (SB), light physical activity (LPA) and moderate-to-vigorous physical activity (MVPA) measured by accelerometry during summer, winter and spring and during school terms and school holidays. Data were analysed using multivariate mixed-effects multilevel modelling with robust SEs. Factors of interest were ethnicity, holiday/term, sex, socioeconomic status (SES), weight status, weekend/weekday and season. One hundred and eight children (67.5%) provided 1157 valid days of data. Fifty-nine per cent of children were WB (n=64) and 41% (n=44) were SA. Boys spent more time in MVPA (11 min/day, p=0.013) compared with girls and SA children spent more time in SB (39 min, p=0.017) compared with WB children in adjusted models. Children living in higher SES areas were more sedentary (43 min, p=0.006) than children living in low SES areas. Children were more active during summer (15 min MVPA, p<0.001; 27 LPA, p<0.001) and spring (15 min MVPA, p=0.005; 38 min LPA, p<0.001) and less sedentary (−42 min and −53 min, p<0.001) compared with winter. Less time (8 min, p=0.012) was spent in LPA during school terms compared with school holidays. Children spent more time in MVPA (5 min, p=0.036) during weekend compared with weekdays. Overweight and obese children spent more time in LPA (21 min, p=0.021) than normal-weight children. The results of our study suggest that significant child level factors associated with movement behaviours are ethnicity, sex, weight-status and area SES. Significant temporal factors are weekends, school holidays and seasonality. Interventions to support health enhancing movement behaviours may need to be tailored around these factors.