Designing trustful and empowering human-robot interactions using a living lab approach
Résumé
he notion of trust, not yet enough scientifically defined, that a machine can inspire explicitly or implicitly has been much studied (Schaefer et al., 2016; Chang & Joseph, 2020). For fragile users in isolated situations, it has been shown that by manipulating very simple robot parameters, it is possible to trigger strong attachment and desirability of the machine (Aubergé et al., 2014). Two research questions arise: 1) What are the robot parameters that develop this trust? 2) Does this trust have beneficial or adverse consequences for these users and the entire ecosystem that cares for them? In this presentation, we address the first question using a living lab approach (Pino et al., 2013). Methods and Materials: For our research, we used the first drawings and prototypes of the new Miroki robot currently under development. This robot is both a social and a logistical robot, serving care professionals as well as patients and their caregivers. Over the past 2 years, we have carried out the following studies in hospitals and nursing homes: 1) Three workshops with 15 older adults and 6 semi-directive interviews with experts in hospital innovation to gather their initial reactions to this robot and discuss their perception of the benefits and risks associated with its use. 2) Two workshops with a total of 67 hospital caregivers to better understand the notion of trust they might have in such a robot. 3) An experiment with 30 caregivers in a scenario for helping older adults to hydrate themselves (a scenario in which we introduced the robot as an object or a subject, varying its degree of expressivity). 4) A study with 48 people (10 older adults, 10 informal caregivers, 14 hospital caregivers and 14 other hospital staff) to ensure that the robot's development was guided by users' needs, while aligning the objectives of the innovation project with the requirements of the geriatric hospital environment. Results and discussion: Some of our results confirm the main findings of the scientific literature, in particular concerning 1) the importance of a humanoid / animaloid design as an important factor in this acceptability (Barnes et al., 2017) compared with purely industrial robots, 2) the strong link between the robot's perceived usefulness in its use cases and its level of acceptability (Lee et al., 2018), 3) the link between the quality of the robot's interaction and its acceptability to its end-users (Whelan et al., 2018). Other results show the influence of parameters little studied so far on the perception of the robot, such as the influence of introducing the robot as an object (machine) versus a subject as well as its level of expressiveness. Thus, the robot Miroki 1) introduced as a subject (rather than an object) with a high degree of expressiveness seems to be perceived as more flexible and inspiring more trust 2) with a high degree of expressiveness (rather than a low one) and presented as an object seems to be perceived as more tender. How these parameters influence user perception over time will require further research. The impact of these parameters on user perception and interaction with the robot over time will be addressed in further studies