Trust is highly relevant in human-robot interaction, particularly when it takes place in complex and dynamic social environments. Here, we give an overview of our research within the Real-World Lab Robotics-AI, an interand transdisciplinary research efort in which robots are embedded in society in long-term field research. We focus on two particularly challenging research sites, a kindergarten and an inclusive daycare, and reflect upon implications for researching and designing for trust in robots in this context.
Trust in robots is a concept as fundamental as it is elusive for Human-Robot Interaction (HRI) research
[
The potential benefits of integrating socially assistive robots into kindergartens are well documented
in the literature: social robots can deliver personalised and interactive learning experiences [
The possibility of conducting long-term research on socially assistive robotics in childcare settings ofered by the Real-World Lab Robotics-AI project provides us with a unique opportunity to contribute to the understanding of trust (and its evolution) on socially assistive robots “in the field”. In this position paper, we outline the specific objectives we aim for and the methodology we intend to pursue to achieve them.
The Real-World Lab Robotics-AI project involves multiple research sites, including a kindergarten and a
daycare, and follows the “Real-World Lab” paradigm [
One of the project partners is a kindergarten operated by a state-recognised provider of youth welfare services, which ofers 50 spots for children aged 3 years up to school entry (see Figure 1-left). The pedagogical focus of the kindergarten is on science and technology, as well as movement and sports. Additionally, the kindergarten provides a cross-group bilingual program through native English-speaking teachers. Through a years-long collaboration with KIT, the kindergarten regularly incorporates socially assistive robots (NAO and Pepper, shown in Figure 1-right) into their activities with children. Under teacher guidance, these robots engage children through playful activities, such as guiding them through language learning games and physical exercises.
Within the project we also cooperate with an inclusive daycare centre that cares for approximately 75 children. As an inclusive institution, the daycare is looking after both disabled and non-disabled children who spend their daily lives together. While this inclusive approach can lead to diferent development stages between children of similar ages, a wide age range - of one to six years - further introduces a plethora of diferent needs and desires managed by the caregivers. Caregivers put great efort toward the inclusion of all by e.g. including lessons in sign language in their interactions with the children equipping both non-verbal and verbal children alike with tools to communicate with them and amongst each other. The daycare also routinely incorporates a NAO robot to engage the children in physical activities or storytelling sessions.
Since trust (by all involved stakeholders) is a pre-requisite for a successful robot integration in any
educational context and since the “Real-World Lab Robotics-AI" project ultimately aims at developing
and deploying social robots that can support the activities and objectives of the partner research sites,
a key objective of the project is to understand how trust in robots evolves in the two complex social
environments. Referring to established findings and naming conventions on trust in HRI settings [
To address these research objectives, we have formulated the following key questions that guide our investigation: 1. To what extent does the social environment within a kindergarten, including peer and teacher interactions, afect children’s trust in social robots? 2. How do disabled children engage in interaction with social robots, and are there specific adaptations that can enhance trust? 3. How do perceived educational and developmental benefits influence children’s, educators’ and parents’ trust in social robots in early childhood education? 4. How does the nature and quality of interaction experiences between children and social robots influence the development of trust over time?
The striking diversity among our stakeholders suggests that a portfolio of research methods should be
favoured over a "one-fits-all" solution, to take into account and adjust to diferent abilities, roles and
requirements [
Alongside the problem of how to measure trust, stands the challenge of identifying what else we need to measure, since the real-world, peculiar and dynamic conditions of our research sites make it non-trivial to identify the environmental and social factors that need to be modelled to obtain an accurate picture of user trust.
To address our research questions and overcome the afore-identified challenges, our project will undertake an exploration of trust at the two research sites, examining each context both separately and comparatively. We envision to leverage the multidisciplinary expertise within the project consortium (including our practice partners) to utilise a mixed-methods approach attuned to each of the sites and capture the multifaceted nature of trust by combining day-to-day observations and robot logs with questionnaires and semi-structured interviews.
Given the embedded nature of our research approach, day-to-day observations of participant interactions with robots in combination with robot activity logs and metrics of use will lay the foundation for work; such real-time behavioural observations (conducted and analysed by qualified experts) will ofer insight into non-verbal indicators of trust, e.g., body language, and allow for identifying potential correlations with contextual, social and/or robot-related events.
These data will be supplemented by questionnaires and semi-structured interviews, regularly
exploring user trust levels. This will provide additional insight into subjective experiences and the underlying
reasons for user trust or distrust in the robots. Since valid and reliable questionnaire only exist for
adults directly interacting with robots [11], semi-structured interviews will be particularly useful with
all the adults indirectly involved in the human-robot interaction. To overcome the lack of validated
questionnaires for children [
Measuring trust in socially assistive robots is an open challenge due to the varying perspectives and expectations among diferent stakeholders, the long interaction required for the robot to reveal its benefits and the dynamic and unstructured environment in which the human-robot interaction occurs.
The Real-World Lab approach pursued in the “Real-World Lab Robotics-AI" project allows us to study the full complexity of human-robot trust beyond the dyadic relationship between humans and robots, particularly in the dynamically changing environments of the two diferent childcare settings.
Through the project we aim to develop accessible methods to explore what trust in socially assistive robots means to our diferent stakeholders groups, with a special emphasis on the needs of the children. We hope that by identifying and contextualising factors of relevance, real-world labs can provide deeper insights into the dynamics of trust in human-robot interactions in realistic social and assistive settings.
This work was funded by the Baden-Württemberg Ministry of Science, Research and Art (MWK), via the state digitalisation strategy digital@bw. [11] K. E. Schaefer, Measuring trust in human robot interactions: Development of the “trust perception scale-hri”, in: Robust intelligence and trust in autonomous systems, Springer, 2016, pp. 191–218. [12] K. Gerling, K. Hicks, M. Kalyn, A. Evans, C. Linehan, Designing movement-based play with young people using powered wheelchairs, in: Proceedings of the 2016 CHI Conference on Human Factors in Computing Systems, CHI ’16, Association for Computing Machinery, New York, NY, USA, 2016, p. 4447–4458. URL: https://doi.org/10.1145/2858036.2858070. doi:10.1145/2858036.2858070. [13] J. Falk, A. Blumenkranz, M. Kubesch, R. Vetter, L. Hofer, C. Frauenberger, Designing diverse pathways for participation, in: Proceedings of the CHI Conference on Human Factors in Computing Systems, CHI ’24, Association for Computing Machinery, New York, NY, USA, 2024. URL: https: //doi.org/10.1145/3613904.3642240. doi:10.1145/3613904.3642240.