Levels of Automation and Transparency: Interaction Design Considerations in Assistive Robots for Older Adults

Abstract

It is important to encourage older adults to remain active when interacting with assistive robots. This study proposes a schematic model for integrating levels of automation (LOAs) and transparency (LoTs) in assistive robots to match the preferences and expectations of older adults. Metrics to evaluate LOA and LoT design combinations are defined. We develop two distinctive test cases to examine interaction design considerations for robots working for this population in everyday tasks: a person-following task with a mobile robot and a table-setting task with a robot manipulator. Evaluations in user studies with older adults reveal that LOA and LoT combinations influence interaction elements. Low LOA and high LoT encouraged activity engagement while receiving adequate information regarding the robot's behavior. The variety of objective and subjective metrics is essential to provide a holistic framework for evaluating the interaction. Index Terms-Assistive robots (ARs), human-robot interaction, interaction design, level of automation (LOA), level of transparency (LoT), older adults, socially ARs. I. BACKGROUNDT HE GLOBAL population of older adults (aged 65+) is increasing rapidly without commensurate growth in people that can support them [1]. This shortage is creating an eldercare gap in which the scarcity of caregivers, social support, and healthcare professionals has left many in this group facing many barriers in aging [2]. Assistive robots (ARs) can help reduce these barriers, facilitate independence, and promote more successful aging (e.g., [3]-[5]). While there has been progress in AR design and development for many daily applications [6],

Methods to assess the effectiveness and acceptance of information and communication technology-based assistive technology for older adults: a scoping review

“…Appropriate quality: two or three additional criteria were missed [24,33,34,36–38,42,44,48,53,57–59].…”

“…(1) Aspirational quality: none or only one of the additional criteria was missed [40,41,[45][46][47][48][49][50][51][52]54,55,61]. (2) Appropriate quality: two or three additional criteria were missed [24,33,34,[36][37][38]42,44,48,53,[57][58][59]. (3) Acceptable quality: four or more additional criteria were missed [22,35,43,56,60].…”

“…In some cases, the convenience sampling technique was refined based on the size and composition of the sample population. In five studies [33,44,47,51,58], participants were selected from a large population through advertisements, the social care network and the databases of health insurance companies. The samples included patients from one or more clinics in two reports [36,55], social service clients in two reports [39,51], nursing home care recipients in three studies [42,57,59], voluntary applicants in two reports [49,52] and those recommended by the other participants in four studies [24,48,50,58].…”

A Literature Survey of How to Convey Transparency in Co-Located Human–Robot Interaction

“…Here, human operators need to receive feedback from the remote robot via UIs since no inherent information from face-to-face interaction is available [28]. The robot needs to communicate information on its perception of the environment to the human in order to inform them of possible changes or obstacles that could inhibit the otherwise given robot capabilities [6,19,29]. As teleoperation within transparent HRI has already been thoroughly covered, we exclude this area from our study.…”

“…3-point Likert [29] "The expression of the robot was perceivable" "The expression of the robot was easy to understand" 5-point Likert [96] " "The movements of the robot become more predictable through the projection of a projector." 5-point Likert [77] Objective Measures Refs.…”