The estimation of the pose of a differential drive mobile robot from noisy odometer, compass and beacon distance measurements is studied. The estimation problem, which is a state estimation problem with unknown input, is reformulated into a state estimation problem with known input and a process noise term. A heuristic sensor fusion algorithm solving this state-estimation problem is proposed and compared with the extended Kalman filter solution and the Particle Filter solution in a simulation experiment. https://doi.org/10.4018/IJAIML.2020010101 https://www.linkedin.com/in/john-bolte-0856134/
Introduction: Visually impaired people experience trouble with navigation and orientation due to their weakened ability to rely on eyesight to monitor the environment [1][2]. Smartphones such as the iPhone are already popular devices among the visually impaired for navigating [3]. We explored if an iPhone application that responds to Bluetooth beacons to inform the user about their environment could aid the visually impaired in navigation in an urban environment.Method: We tested the implementation in an urban environment with visually impaired people using the route from the Amsterdam Bijlmer train station to the Royal Dutch Visio office. Bluetooth beacons were attached at two meters high to lampposts and traffic signs along a specified route to give the user instructions via a custom made iPhone app. Three different obstacle types were identified and implemented in the app: a crossover with traffic signs, a car parking entrance and objects blocking the pathway like stairs. Based on the work of Atkin et al.[5] and Havik et al. [6] at each obstacle the beacon will trigger the app to present important information about the surroundings like potential hazards nearby, how to navigate around or through obstacles and information about the next obstacle. The information is presented using pictures of the environment and instructions in text and voice based on Giudice et al. [4]. The application uses Apple’s accessibility features to communicate the instructions with VoiceOver screenreader. The app allows the user to preview the route, to prepare for upcoming obstacles and landmarks. Last, users can customize the app by specifying the amount of detail in images and information the app presents.To determine if the app is more useful for the participants than their current navigational method, participants walked the route both with and without the application. When walking with the app, participants were guided by the app. When walking without the app they used their own navigational method. During both walks a supervisor ensured the safety of the participant.During both walks, after each obstacle, participants were asked how safe they felt. We used a five point Likert scale where one stood for “feeling very safe” and five for “feeling very unsafe”.Qualitative feedback on the usability of the app was collected using the speak-a-lout method during walking and by interview afster walking.Results: Five visually impaired participated, one female and five males, age range from 30 to 78 and with varying levels of visual limitations. Three participants were familiar with the route and two walked the route for the first time.After each obstacle participants rated how safe they felt on a five point Likert scale. We normalized the results by deducting the scores of the walk without the app from the scores of the walk with the app. The average of all participants is shown in figure 2. When passing the traffic light halfway during the route we see that the participants feel safer with than without the app.Summarizing the qualitative feedback, we noticed that all participants indicated feeling supported by the app. They found the type of instructions ideal for walking and learning new routes. Of the five participants, three found the length of the instructions appropriate and two found them too long. They would like to split the detailed instructions in a short instruction and the option for more detailed instructions. They felt that a detailed instruction gave too much information in a hazardous environment like a crossover. Two participants found the information focused on orientation not necessary, while three participants liked knowing their surroundings.Conclusion and discussion: Regarding the safety questions we see that participants felt safer with the app, especially when crossing the road with traffic lights. We believe this big difference in comparison to the other obstacles is due to the crossover being considered more dangerous than the other obstacles. This is reflected by their feedback in requesting less direct information at these locations.All participants indicated feeling supported and at ease with our application, stating they would use the application when walking new routes.Because of the small sample size we consider our results an indication that the app can be of help and a good start for further research on guiding people through an urban environment using beacons.
Urban environments are full of noise and obstacles, and therefore potentially dangerous and difficult to navigate for the visually impaired. Using Bluetooth beacons and a smartphone app we guide them through these environments by providing the information needed for that specific location. We present the preliminary results concerning the usability of our approach.
In de MKB-maakindustrie zijn aanzienlijke kosten gemoeid met het intern transport van goederen en materialen. Het belang van gebruik van mobiele robots voor interne logistieke processen neemt dan ook razendsnel toe. Men wil betaalbare mobiele robotsystemen die betrouwbaar kunnen werken en navigeren binnen de unieke omgeving van de eigen werkvloer. Voor gerobotiseerde oplossingen wil en kan de MKB-maakindustrie echter niet afhankelijk zijn van één leverancier. Er is immers geen omgeving die geheel is toegespitst op en aangepast aan het gebruik van robots zoals in grootschalige logistieke bedrijven. Daarom is inzet van combinaties van robots van verschillende leveranciers met elk verschillende functionaliteiten noodzakelijk. Probleem is echter dat roboticaleveranciers specifieke verkeersmanagementsystemen (fleetmanagement) leveren en ook de interactie van robots met ERP (Enterprise Resource Planning) systemen, liften, deuren etc. op hun eigen wijze implementeren. Er bestaat geen generiek of geïntegreerd fleetmanagement systeem, waardoor de diverse typen robots nu letterlijk onafhankelijk van elkaar opereren. Dit resulteert in inefficiënt robotverkeer met een groot risico op verkeersproblemen, hinder voor personeel en dure parallelle koppelingen met interfaces (met deuren, liften etc.). Dit leidt tot verwarring, onzekerheid en potentiële veiligheidsproblemen bij werknemers op de werkvloer. Ambitie van het project Let’s Move IT is om verschillende fabricaten en typen mobiele robots (met elk hun eigen logistieke taken) beter met elkaar te laten communiceren en samenwerken. In het project wordt daartoe gewerkt aan integraal verkeersmanagement, modulaire interfaces en slimme gecombineerde omgevingsinterpretatie. Zo kunnen logistieke robots veilig, flexibel, robuust en adaptief opereren in een steeds veranderende productieomgeving in aanwezigheid van mensen. Het project is een samenwerking van Fontys Hogescholen, Haagse Hogeschool en NHL. Participerende (MKB-)bedrijven zijn werkzaam als ontwikkelaar, system integrator, toeleverancier en eindgebruiker van mobiele robotsystemen. Daarnaast zijn coöperatie Brainport Industries en Metaalunie nauw betrokken. In het project zal bestaande kennis toepasbaar worden gemaakt en zal nieuwe kennis worden ontwikkeld ten behoeve van het verkeersmanagement van meerdere fabricaten mobiele robots tegelijkertijd. Verder zal verankering van kennis en kunde in onderwijs en lectoraten plaatsvinden en een vergroting van de kwaliteit van docenten en afstudeerders. Er zullen circa 18 (docent-)onderzoekers van de hogescholen en circa 100 studenten betrokken worden, die in de vorm van studentenprojecten, stages en afstudeeronderzoeken werken aan interessante vraagstukken direct uit de beroepspraktijk.
Lack of physical activity in urban contexts is an increasing health risk in The Netherlands and Brazil. Exercise applications (apps) are seen as potential ways of increasing physical activity. However, physical activity apps in app stores commonly lack a scientific base. Consequently, it remains unknown what specific content messages should contain and how messages can be personalized to the individual. Moreover, it is unknown how their effects depend on the physical urban environment in which people live and on personal characteristics and attitudes. The current project aims to get insight in how mobile personalized technology can motivate urban residents to become physically active. More specifically, we aim to gain insight into the effectiveness of elements within an exercise app (motivational feedback, goal setting, individualized messages, gaming elements (gamification) for making people more physically active, and how the effectiveness depends on characteristics of the individual and the urban setting. This results in a flexible exercise app for inactive citizens based on theories in data mining, machine learning, exercise psychology, behavioral change and gamification. The sensors on the mobile phone, together with sensors (beacons) in public spaces, combined with sociodemographic and land use information will generate a massive amount of data. The project involves analysis in two ways. First, a unique feature of our project is that we apply machine learning/data mining techniques to optimize the app specification for each individual in a dynamic and iterative research design (Sequential Multiple Assignment Randomised Trial (SMART)), by testing the effectiveness of specific messages given personal and urban characteristics. Second, the implementation of the app in Sao Paolo and Amsterdam will provide us with (big) data on use of functionalities, physical activity, motivation etc. allowing us to investigate in detail the effects of personalized technology on lifestyle in different geographical and cultural contexts.
Lack of physical activity in urban contexts is an increasing health risk in The Netherlands and Brazil. Exercise applications (apps) are seen as potential ways of increasing physical activity. However, physical activity apps in app stores commonly lack a scientific base. Consequently, it remains unknown what specific content messages should contain and how messages can be personalized to the individual. Moreover, it is unknown how their effects depend on the physical urban environment in which people live and on personal characteristics and attitudes. The current project aims to get insight in how mobile personalized technology can motivate urban residents to become physically active. More specifically, we aim to gain insight into the effectiveness of elements within an exercise app (motivational feedback, goal setting, individualized messages, gaming elements (gamification) for making people more physically active, and how the effectiveness depends on characteristics of the individual and the urban setting. This results in a flexible exercise app for inactive citizens based on theories in data mining, machine learning, exercise psychology, behavioral change and gamification. The sensors on the mobile phone, together with sensors (beacons) in public spaces, combined with sociodemographic and land use information will generate a massive amount of data. The project involves analysis in two ways. First, a unique feature of our project is that we apply machine learning/data mining techniques to optimize the app specification for each individual in a dynamic and iterative research design (Sequential Multiple Assignment Randomised Trial (SMART)), by testing the effectiveness of specific messages given personal and urban characteristics. Second, the implementation of the app in Sao Paolo and Amsterdam will provide us with (big) data on use of functionalities, physical activity, motivation etc. allowing us to investigate in detail the effects of personalized technology on lifestyle in different geographical and cultural contexts.
