The interest in social robots has grown dramatically in the last decade. Several studies have investigated the potential markets for such robots and how to enhance their human-like abilities. Both of these subjects have been investigated in this thesis using the company Furhat Robotics, and their robot Furhat, as a case study. This paper explores how machine learning could be used to classify dialogue acts in human-robot conversations, which could help Furhat interact in a more human-like way. Dialogue acts are acts of natural speech, such as questions or statements. Several variables and their impact on the classification of dialogue acts were tested. The results showed that a combination of some of these variables could classify 73 % of all the dialogue acts correctly. Furthermore, this paper analyzes the market for social robots which are used for education, where human-like abilities are preferable. A literature study and an interview were conducted. The market was then analyzed using a SWOT-matrix and Porter’s Five Forces. Although the study showed that the mentioned market could be a suitable target for Furhat Robotics, there are several threats and obstacles that should be taken into account before entering the market. / Intresset för sociala robotar har ökat drastiskt under det senaste årtiondet. Ett flertal studier har undersökt hur man kan förbättra robotars mänskliga färdigheter. Vidare har studier undersökt potentiella marknader för sådana robotar. Båda dessa aspekter har studerats i denna rapport med företaget Furhat Robotics, och deras robot Furhat, som en fallstudie. Mer specifikt undersöker denna rapport hur maskininlärning kan användas för att klassificera talhandlingar i människa-robot- konversationer, vilket skulle kunna hjälpa Furhat att interagera på ett mer mänskligt sätt. Talhandlingar är indelningar av naturligt språk i olika handlingar, såsom frågor och påståenden. Flertalet variabler och deras inverkan på klassificeringen av talhandlingar testades i studien. Resultatet visade att en kombination av några av dessa variabler kunde klassificera 73 % av alla talhandlingar korrekt. Vidare analyserar denna rapport marknaden för sociala robotar inom utbildning, där mänskliga färdigheter är att föredra. En litteraturstudie och en intervju gjordes. Marknaden analyserades sedan med hjälp av en SWOT-matris och Porters femkraftsmodell. Fastän studien visade att den ovannämnda marknaden skulle kunna vara lämplig för Furhat Robotics finns ett flertal hot och hinder som företaget måste ta hänsyn till innan de tar sig in på marknaden.

This study intends to further investigate the fitness for robot assisted learning in language cafés. Through conducting dialogues with two learners of swedish, moderated by the robot Furhat and later interviewing the participants, the aim is to identify what areas work well and what needs to improve upon with robot assisted learning in general, and Furhat in particular. This study identifies several areas of improvements, and make some possible suggestions for solutions. / Denna studie avser att undersöka lämpligheten i robotassisterad inlärning på språkcaféer. Genom dialog med två SFI-studenter, modererat av roboten Furhat följt av en intervju med de båda deltagarna, är målet att identifiera vilka områden som fungerar bra idag, och vilka som behöver förbättras inom robotassisterad inlärning som område, och konkret med Furhat. Denna studie identifierar flera områden för förbättringar, och ger förslag på åtgärder.

In the last decades the average life expectancy of humans has increased significantly. There are more old people than ever before. At the same time there is a big staff shortage at nursing homes. A future study will examine the effect of introducing a socially intelligent robot called Furhat in such a home. In this work a so-called Wizard-of-Oz control system is developed which enables remote control of the normally autonomous Furhat. This control system will later be used in the future study. The Wizard-of-Oz control system is developed with the intention of reaching the lowest possible response time between the control system and Furhat to minimize the risk of unnatural conversation due to long wait times between actions. It is found that a response time at or above 500ms can have a clearly degrading effect on a conversation. By following code-standards with a focus on developing fast code an average response time in the range of 35-245ms depending on the action taken is reached.

Det svenska företaget Furhat Robotics har skapat en robot kallad Furhat vilken är kapabel till att interagera med människor i språkcafé-liknande miljöer. Syftet med den robotledda konversationen är att utveckla deltagarnas språkkunskaper, vilka efter varje konversation får svara på en enkät om vad de tyckte om samtalet med Furhat. Ur detta har frågan huruvida det är möjligt att förutspå vad deltagarna tyckte om samtalet baserat på konversationens struktur uppstått. Syftet med denna rapport är att analysera huruvida det är möjligt att kvantifiera konversationerna och förutspå svaren i enkäten med hjälp av maskininlärning. Det dataset som rapporten baserar sig på erhölls från tidigare studier i Kollaborativ Robotassisterad Språkinlärning (Collaborative Robot Assisted Language Learning). Resultaten visade på ett RMSE högre än variansen för medelvärdet av enkätsvaren vilket indikerar att den framtagna modellen inte är särskilt effektiv. Modellen presterade dock bättre i vissa förutsägelser då varje enskilt enkätsvar förutspåddes var för sig. Detta antyder att modellen skulle kunna användas till vissa frågeformuleringar / The Swedish company Furhat Robotic have created a robot called Furhat, which is able to interact with humans in a language café setting. The purpose of the robot led conversation is for the participants to develop their language skills. After the conversation the humans will answer a survey about what they thought about the conversation with Furhat. A question that has arisen from this is if it is possible to predict the survey answers based on just the conversation. The purpose of this paper is to analyze if it is possible to quantify the conversations linked to the survey answers, and by doing so be able to predict the answers in new conversations with a machine learning approach. The data set being used was obtained from an earlier study in Collaborative Robot Assisted Language Learning. The result returned a RMSE that was greater than the variance of the average conversation score which indicates that the model is not very effective. However, it excelled in some predictions trying to give scores to each separate survey answer, indicating that the model could be used for certain question formulations.

I detta arbete genomfördes en enkätundersökning för att se hur upplevelsen av en furhatrobot ändrades efter att ha implementerat felhantering och mer användarinitiativ. Genom enkätformatet Godspeed undersöktes kategorierna människolikhet och upplevd intelligens. Roboten står placerad på ett museum och har som syfte att problematisera kring dagens teknologiska utveckling. Arbetet utfördes genom att modifiera en existerande skill för att undersöka hur ovan nämnda aspekter påverkar besökares upplevelse av roboten. Enkätstudien visade ej på någon större skillnad i hur roboten uppfattades efter modifikation. Framtida arbete skulle kunna undersöka om mer användarinitiativ i form av möjlighet att avbryta roboten eller småprata med den leder till en ökad upplevd intelligens eller människolikhet.

In recent years, Swedish school children have experienced a decline in reading motivation. This is no less true for middle school students both in and outside of school. This thesis aims to find ways in which robot book-talks can be used in a school setting to motivate middle schoolers to read. With a Research through Design approach, the study also aims to find recommendations for designers in the future. These goals were achieved through a design process in three phases, in which the author developed and tested book-talks between fourth graders and the robot Furhat from Furhat Robotics, using the Wizard of Oz (WOZ) technique. In the final evaluative phase, the students had book-talks with Furhat in two different modes, over two sessions. In Passive mode, the robot asked general questions and made little effort to make sure the students were concentrating on the task. This mode served as a baseline for the “novelty effect”. The other – Active mode – built on the ideas created throughout the process. In this mode, the robot asked questions directly related to the students’ books and made more effort to guide the students through their task. Results show that while the robot interaction was appreciated by all partaking students, it was not possible to prove if it spurred genuine reading motivation. Nevertheless, results showed that the interaction could potentially fulfil the basic psychological needs of competence and relatedness. Along with autonomy, both needs are fundamental to motivation and well-being, according to Self-Determination Theory. The book-talks’ effect on students’ autonomy could not be tested properly, as there were too few copies of the books which the students could read for the book-talks. The study also showed signs of the “novelty effect”, as some students were interested in reading more so that they could talk to the robot again. Three main takeaways for future designers were found as a result of the study: using Self-Determination Theory to create UX goals and to design; make designs that encourage people to interact with people; both follow and guide the users throughout the WOZ interaction, to create moments where the students can experience relatedness and show competence.

The work presented in this thesis comes in pursuit of the ultimate goal of building spoken and embodied human-like interfaces that are able to interact with humans under human terms. Such interfaces need to employ the subtle, rich and multidimensional signals of communicative and social value that complement the stream of words – signals humans typically use when interacting with each other. The studies presented in the thesis concern facial signals used in spoken communication, and can be divided into two connected groups. The first is targeted towards exploring and verifying models of facial signals that come in synchrony with speech and its intonation. We refer to this as visual-prosody, and as part of visual-prosody, we take prominence as a case study. We show that the use of prosodically relevant gestures in animated faces results in a more expressive and human-like behaviour. We also show that animated faces supported with these gestures result in more intelligible speech which in turn can be used to aid communication, for example in noisy environments. The other group of studies targets facial signals that complement speech. As spoken language is a relatively poor system for the communication of spatial information; since such information is visual in nature. Hence, the use of visual movements of spatial value, such as gaze and head movements, is important for an efficient interaction. The use of such signals is especially important when the interaction between the human and the embodied agent is situated – that is when they share the same physical space, and while this space is taken into account in the interaction. We study the perception, the modelling, and the interaction effects of gaze and head pose in regulating situated and multiparty spoken dialogues in two conditions. The first is the typical case where the animated face is displayed on flat surfaces, and the second where they are displayed on a physical three-dimensional model of a face. The results from the studies show that projecting the animated face onto a face-shaped mask results in an accurate perception of the direction of gaze that is generated by the avatar, and hence can allow for the use of these movements in multiparty spoken dialogue. Driven by these findings, the Furhat back-projected robot head is developed. Furhat employs state-of-the-art facial animation that is projected on a 3D printout of that face, and a neck to allow for head movements. Although the mask in Furhat is static, the fact that the animated face matches the design of the mask results in a physical face that is perceived to “move”. We present studies that show how this technique renders a more intelligible, human-like and expressive face. We further present experiments in which Furhat is used as a tool to investigate properties of facial signals in situated interaction. Furhat is built to study, implement, and verify models of situated and multiparty, multimodal Human-Machine spoken dialogue, a study that requires that the face is physically situated in the interaction environment rather than in a two-dimensional screen. It also has received much interest from several communities, and been showcased at several venues, including a robot exhibition at the London Science Museum. We present an evaluation study of Furhat at the exhibition where it interacted with several thousand persons in a multiparty conversation. The analysis of the data from the setup further shows that Furhat can accurately regulate multiparty interaction using gaze and head movements. / <p>QC 20121123</p>