Researchers at the University of California San Diego have identified several characteristics and design guidelines for robots to support people living with dementia and their caregivers.

The list was derived from interviews and hands-on workshops at adult day centers for those with dementia. Researchers interviewed 13 family caregivers for people with various types of dementia, five social workers who led the day centers, and three geriatric nursing students who work at the centers.

“Caregivers conceived of robots not only managing difficult aspects of caregiving, but also for supporting joyful and fun activities,” said Laurel Riek, Ph.D., a professor of computer science at the university and the resulting paper’s senior author.

Robots, according to the study, should:

  • Help redirect conversations when repetitive questioning becomes burdensome.
  • Be integrated into everyday objects with which the people who have dementia are already familiar, or borrow features from those objects. For example, one family caregiver wanted her husband to get messages through the TV, which he spends a lot of time watching.
  • Be able to adapt to new situations and to the behavior of the person living with dementia. This is particularly important, according to the study authors, because dementia is a progressive disease, and each stage brings new challenges for caregivers. Also, patterns of progression vary from person to person and as a result are almost impossible to predict.
  • Be able to learn from end-users and customize and personalize their interaction and responses.
  • Have human-like components. The machines, for example, could use a real human voice or face to help persuade someone to do something or help build trust.
  • Interact with humans via voice activation, much like a smart speaker. More specifically, caregivers wanted the robots to use voices with which their loved ones would be familiar, such physicians or other caregivers. Caregivers also wanted robots to be capable of facial recognition.

The researchers presented their findings at the Human Robot Interaction conference in March in South Korea.