Social Robot

Navel

Navel is an AI enabled social robot. He is able to address each patient individually by remembering their names and previous conversations, and engage in various novel conversations because of its use of current AI technologies. While its built-in connection to AI aligns with some of our goals involved in the conversational aspects of the matching game, this functionality can be extended to any social robot in a more specific way as long as it has a microphone and speaker.

Miro-E

Miro is a small, cute animal-like social robot. It is designed to behave with animal qualities, and is able to react to various stimuli like sound and motion. For our purposes as a guide and conversational partner, Miro's more pet-like attributes make it less suitable than the alternatives.

Nao

NAO is a small humanoid robot that is built for interacting with people. It is able to talk, recognize faces, move around and gesture which makes it more engaging. That said, similar to Navel and that for our purpose we do not require locomotive functionality, we decided to skip it to reduce the complexity that needs to be considered.

Pepper

We want to use Pepper because it has a built-in screen where people with dementia can play the memory game easily. This makes things easier as well, since we only need one device to upload personal memories and the related context into the game. Pepper can then use this information to interact with the clients in an engaging and personal way. In addition to the screen, Pepper can talk and ask questions since it has speakers, which helps make the interaction more natural. Studies have found that life-sized humanoid robots are perceived by older adults as more socially engaging than smaller or animal-like robots [1]. 

Comparison table

Trade-offs:

• Humanoid vs. Pet-like
  • Humanoid robots have a stronger social presence, but take up more space and may fall into an uncanny-valley situation with their appearance. 
  • Pet-like robots have lower expectations of cognitive simulation abilities and are smaller, reducing the risks of injuries but their pet-like appearance may be misaligned for the tasks of playing an image based game.
• Display or No. Display: 
  • Having the display built-in makes everything more cohesive, with both the game and interaction partner being self-contained. Using a display also decreases running setup costs for new patients or games.
  • A display may not be the best medium for the game itself given the target group of PwDs who may not all be comfortable with technology.

Constraints:

• Mobility & physical footprint
  • Pepper is quite large and requires ample space for maneuvering around in the activity room
  • Getting Pepper to the room and out of it needs consideration and could be dangerous if Pepper tips over or someone trips
• Hygiene and Maintenance
  • With the expected interactions, Pepper needs to be adequately cleaning both on the general body and the screen.
  • In case of updates or damage, there would be need to be scheduled downtime where Pepper is unavailable
• Data and Privacy
  • Data needs to be transferred to and from Pepper to facilitate the games, and so requires a secure encrypted channel between (ideally) a locally hosted database and system

[1] Broekens, Joost, Marcel Heerink, and Henk Rosendal. "Assistive social robots in elderly care: a review." Gerontechnology 8.2 (2009): 94-103.