PAL Robotics’ social robot ARI is taking part in the SHAPES project. The implementation of the first pilot (phase 2) took place in Clinica Humana, a private clinic with more than 7-years of expertise in chronic patient management and providing hospital care to seniors’ residences, communities, and homebound patients with a strong technological component in telemedicine, as well as in Ca’n Granada, a residence in Mallorca with independent, accessible homes, adapted for people with reduced mobility and fully equipped.

During these pilots, based on user requirements, ARI was programmed to carry out the following actions:

– Temperature monitoring

– Visualizing the agenda of the day

– Sending alerts to caregivers, or sign up for care-home events such as movies

– Video calling

– Playing games (puzzles, solitaire, or finding matching pairs)

– Information about the care-home facility (a request from phase 2 mock-ups)

– Filling in the weekly menu (a request from phase 2 mock-ups)

– Voice-based assistant

– Emotion recognition via TREE Technology

– Detecting and monitoring falls

The primary purpose of Phase 3: hands-on training of SHAPES Pilots was to collect feedback. The experiment was carried out with five target users (all of whom were over the age of 65). In general, ARI raised positive interest, however, some people did not feel comfortable with the robot eye movement, so adjustable eyes will be integrated for the next phase.

Caregivers mentioned the need for pop-ups to confirm actions. It was also suggested that the robot should have a more adaptive interface to increase its accessibility, such as for example, different font sizes for the touchscreen, adjustable volume, and mechanism to compensate for the height difference between people sitting down or in wheelchairs and the robot’s touch-screen. It was also stated that ARI should be more personalized to the user, which means that the robot should remember the configuration chosen by the user (volume, font size).

Another point to discuss was that fall detection was thought to be difficult to evaluate, mainly due to navigation restrictions, so it will not be used in the following phases. It was also indicated that the registration and authentication procedure to the SHAPES platform had two points of improvement. On one hand, the need for a forgotten password functionality and on the other hand a faster face recognition image.

Modifications to the robot include volume control icons and an additional Android tablet at the back of ARI. If you want to learn more about the ARI robot used for the pilot, you will find more information in this video.

ARI has two behaviors that were developed using Behavior Trees. The first behavior involves users initiating interaction with a static robot at the reception or docking station, whereas the second behavior involves the robot actively looking for a user to initiate interaction. For the second behavior, when the user presses the touch-screen the robot starts identifying the face using face recognition. If you want to learn more about behaviour trees you will find more information here.

Between February and March 2022, phase 4 is taking place, where the mentioned improvements will be tested with five target users in Ca’n Granada and one caregiver for two weeks.

In phase 5, scheduled for summer 2022, ARI robots will be used in completely autonomous mode with 10 service user participants and up to 2-3 carers: one month in individual homes and five months at the care home. ARI will also be ready to start with pilots in Italy and Greece very soon as the prototype is nearly completed.

Finally, at the HRI 2022 Conference at PAL Robotics we had the opportunity to present a late-breaking report ‘Towards Using Behaviour Trees for Long-Term Social Robot Behavior’ and a short-paper at the workshop called TRAITS, hosted by Antonio Andriella with a presentation by Sara Cooper, titled “Lessons Learnt From Deploying an ARI robot in residential care about the results of this phase.

To find out more about ARI, visit the PAL Robotics wesite.