20th WORKSHOP “FROM OBJECTS TO AGENTS”
PARMA, June 26 - 28, 2019
Assessing Usability of a Robotic-based AAL
System: a Pilot Study with Dementia Patients
Claudia Di Napoli Emanuela Del Grosso
Elena Salvatore Federica Garramone Gabriella Santangelo
Silvia Rossi Giovanni Ercolano
TALK OUTLINE
➤
Motivations (
“User-centered Profiling and Adaptation for Socially Assistive
Robotics” UPA4SAR Project
)
➤
The AAL Robotic System
•
Layered Architecture
➤
A service-oriented approach
•
Abstract & Concrete workflows
•
Concrete Services
➤
Pilot Study
•
Patients classification
•
UTAUT-based Usability Test
MOTIVATIONS
➤
Cognitive and mental stimulation
•
Key activities for MCI or dementia to
prevent cognitive reserve
•
Burden on caregivers/relatives
•
Not fixed over long periods of time
➤
Socially Assistive Robotics relying on social rather than
physical interaction for ageing society
•
feasible for vulnerable group of users?
•
feasible in not high-technological environments?
THE AAL ROBOTIC SYSTEM LAYERS
➤
Data Computational Model
•
patient information including daily routine, personality
profile (NEO-PI3), cognitive profile (ACE-R)
➤
Assistive Workflow Management Middleware
•
for instantiating and executing assistive plans
•
… by time-based scheduling of component services
➤
Daily Assistive Actions
•
provided by devices as Android applications
•
communicating through Web Socket Protocol using Json
THE AAL ROBOTIC SYSTEM LAYERS
➤
Smart Environment
•
Sanbot Elf: humanoid robot with
a tablet and an RGB-D camera
•
iBeacons: for indoor positioning
through Bluetooth Low Energy
signal
•
Polar M-600: wearable
smartwatch
A SERVICE-ORIENTED APPROACH
➤
Assistive plans as
workflow of abstract services
to
:
•
represent the
heterogeneous tasks
involved in
entertainment plans in a
uniform way
•
decouple
the required
functionalities
for an assistive plan
from the ways
they are performed
•
adapt/change
part of the plan as soon as dynamic events
are detected
A SERVICE-ORIENTED APPROACH
➤
Abstract workflows according to the
patient’s daily routine
Time%Range%
Daily%Ac/vity%
07:00$%$07:30$ Wake%up$ 07:30$%$08:30$ Breakfast$ 09:30$%$11:00$ Entertainment$Cogni:ve$ 11:30$ Take$medicine$ 12:30$%$13:30$ Lunch$ 14:00$%$16:00$ Res:ng$ 16:00$%$18:00$ Go$out$ 18:30$%$19:30$ Entertainment$Physical$ 20:00$%$21:00$$ Dinner$A SERVICE-ORIENTED APPROACH
➤
Types of concrete services
•
Monitoring services
•
for checking user state: Hearth Rate Detection, Emotion
Recognition
•
for checking user activity: Pose Detection, Activity
Classification, Dialogue Check, In Room Detection
•
Navigation services
•
Look User, Find User, Approach
•
Interaction services
•
Dialogue Suggest, Video Entertainment, Audio Entertainment,
A SERVICE-ORIENTED APPROACH
➤
Concrete workflows instantiation according to
•
the patient’s cognitive and personality profile
•
the schedule of the daily routine
•
the available concrete services
A PILOT STUDY
➤
A preliminary study to assess the acceptance degree of the robotic system
➤ 4 patients classified according to ACE-R and NEO-PI3
➤ in a controlled environment
A PILOT STUDY
➤
Results of patients classification
Subjects/
Features Subject 1 Subject 2 Subject 3 Subject 4
Gender Male Female Female Male
Age 74 75 71 75
Years of
education 8 5 8 18
Language Normal Low Normal Normal
Attention Normal Low Normal Normal
Fluency Low Low Normal Normal
Memory Low Low Slightly Low Slightly Low
Neuroticism Low High Normal Low
Openness Normal Low High Normal
Depressive
A PILOT STUDY
➤
A UTAUT-based usability test is used to evaluate 12 constructs on a Likert type
scale with range 1-5
➤
Score greater than 3 is considered a positive perception
➤
Facilitating conditions
and
Social influence
not significant in a not domestic
A PILOT STUDY
➤
All other constructs > 3
➤
Patients with the same cognitive decline show different
interactions with the robotic system:
•
low education level and depressive symptoms and high
neuroticism impact negatively the interaction with the
robotic system
•
high education level and low neuroticism result in a more
CONCLUSIONS
➤
Personalization based on:
•
Cognitive profile (education, memory, language,
…
CDR)
•
Personality traits (neuroticism, openness, extraversion,
…
NEO-PI)
•
Daily Routine (daily activities carried out throughout the day)
•
Patient’s preferences (music, TV programs,
…)
➤
Adaptation based on:
•
the patient’s state (physical and emotional)
•
the patient’s current activity (detected by devices)
CONCLUSIONS
➤
Concrete services are selected according to:
➤
the way a monitoring activity should be carried out, represented in terms of QoS
parameters evaluated against patient’s profile
➤
the available technology with the possibility to add/delete service providers
➤
Future works
➤