Project homeAnagha - Self Reflection
Week 1: Introduction to Socio-Cognitive Engineering
The teams were formed and a potential application for a robot intervention system to assist individuals with dementia was discussed. The group was introduced to three available robots for the project: Nao, Pepper, and Miro. My team and I chose to focus on mealtime, specifically during dinner, when individuals with moderate-stage dementia may experience confusion, anxiety, and irritability due to Sundown Syndrome. To develop our use-case scenario, we created three personas - Georgina (the individual with dementia), Eleana (the formal caretaker), and Sam (a family member). After careful consideration, we determined that the Pepper robot was the most suitable for our use case, as it can narrate stories and display pictures on a screen attached to its torso. Despite both Nao and Pepper being humanoid robots capable of engaging in conversations, we selected Pepper because our use case didn't require much movement from the robot. Additionally, the height of the Pepper robot was more appropriate for standing in front of or around the individual with dementia while narrating a story.
Week 2: Dementia, Memory, and Music as an Intervention
As engineering students designing a robot for patients with dementia, a guest lecture on the medical aspects of the disease proved informative. We learned about changes in the brain during aging, types of dementia, diagnostic methods, interventions, and treatments. Lifestyle changes can improve the quality of life in the long run, something we may overlook. This understanding helps us design a better solution for the target audience.
The "SCE Method" involving stakeholders and multidisciplinary design helps refine the core functions and design rationale. Cognitive abilities decline with age, and an individual's lifestyle contributes to the rate of decline. Cognitive reserve, the inherent quality of the brain, explains how susceptible the individual would be to functional impairment in case of disease or accident. It can be increased with healthy lifestyle choices. Dementia shows a significant decline in cognitive domains and interferes with daily lifestyle. Pharmacological treatments delay symptoms but not for the long term. Interventions aim to maintain cognition and regulate mood, preventing the individual from feeling "more disabled" than they are, and maintaining self-sufficiency and independence. Current interventions for maintaining cognition involve stimulation, training, and rehabilitation. Music, with its potential for engaging different brain networks, can be used in rehabilitation, mood regulation, social interaction, and even movement rehabilitation.
Our project involves personas, stakeholders, and problem and design scenarios. We designed the scenario for the evening when the patient faces Sundown Syndrome, mood swings, and loses motivation to eat. Our design scenario facilitates conversations between family members while the robot narrates stories and prompts interaction to help the patient complete meals and enjoy them. We decided to employ music in our use case for a more immersive environment for the patient during story narration. We also explored using the patient's favorite songs during narration to create opportunities for reminiscing.
Week 3: PAL
In week 3 we learned about PAL ( Physically Assisted Learning). PAL is a robot intervention system designed to help children with Type 1 diabetes manage their blood glucose levels. The system uses collaborative learning, a theory that suggests individuals learn better when actively interacting with the information, to teach children how to become self-sufficient in managing their diabetes. The robot interacts with the child through games and conversations, reminding them to check their blood glucose levels regularly, administer insulin, and cope with T1DM. PAL uses an adaptive robot that tailors tasks in games to the child's learning progress, motivating them to perform the activity. Evaluation results showed that children preferred playing with the adaptive robot more than the non-adaptive variant, as it helped them learn better. The project exemplifies the partnership between humans and robots, where both work together to achieve objectives and goals.
This week, the focus was on defining use cases, environments, stakeholders, values, and personas for the project. We finalized personas, stakeholders, environment, design, and problem scenarios for the project, using supporting literature and research in their design scenarios. We decided to use the Pepper robot for their project. While our use case is not intended for learning, creating a more collaborative and conversational robot for the project could motivate the patient to look forward to their daily activities.
Week 4: First Presentation
My teammates, Hrishitha and Manali, presented our project on an intervention to assist a person with dementia, Georgina, during mealtimes. We introduced a robot that would act as a storyteller, engaging all the direct stakeholders present in easy-going and nostalgic conversations around the story it narrates. The presentation received feedback and questions that were used to improve the design for implementation. Our approach was unique in that we targeted patients at home with family, which provided more freedom to involve family members, but the disadvantage of this was that it was more expensive than in a care home. We learned about use case design and the team design pattern, which helped us provide a detailed overview of the project and visualize how it would work. We also explained their reasoning for choosing certain features and not choosing others.
Week 5: Design Specifications
In this week's learning, the focus was on the design of a task level for a robot, collaboration, and evaluation of the prototype robots. We learned about task-level design (TLD) and how it plays a vital role in the functioning of a robot. The TLD involves defining what the robot should be doing, the use case, and the motivation effect behind the robot's action. The use case is particularly important, as it sets the context for the robot and determines the actors involved, where the robot will be working, and when it should intervene. The motivation effect helps justify the robot's actions by showing what effect the actions will have and is supported by relevant theory and empirical studies.
We were introduced to team design patterns (TDP) and interaction design patterns (IDP) and how they relate to the TLD. TDP provides an abstract representation of the TLD, including the problem, solution structure, solution description, human and robot requirements, and consequences. IDP, on the other hand, focuses on the interaction level design, including how the robot will perform the requirements.
The evaluation of prototype robots is essential to determine the effectiveness and safety of the robot. Different factors must be considered when evaluating prototype robots, including practical issues, ethics committee permission, formative vs. summative evaluation, measures and their levels, and tools. We used the Godspeed questionnaire, which measures how human-like and likable the robot is, as well as custom questions pertaining to the use case.
We also learned about ontologies and how they can represent users and their properties in one diagram. An ontology is an explicit representation of knowledge, including entities used in the system, their structure, classes, instances, relationships, and properties. Ontologies can be used to represent patients with dementia, their relationships, and properties such as their age and gender.
Finally, we performed a pilot study of our project using Pepper at the Insyght Lab to test the basic connections and familiarize ourselves with the robot. We encountered some issues with the tablet not working, which we later realized was due to a connection issue with InteractiveRobotics. We tested voice and touch inputs and concluded that voice input may not be so accurate, so we went ahead with a backup touch input instead.
Week 6: Implementation and Initial Testings
We focused on finalizing the system design and implementing it on the Interactive Robots platform for evaluation. We decided to test the robot in two scenarios, one where it encouraged conversation and the other where it only narrated a story. We came up with two short stories and tested the motions and flow of the story on the virtual robot before booking a slot for testing on the Pepper robot. During the testing, we modified the prompts and triggers to ensure the flow of conversation was not interrupted. We had a successful testing session with the final code, where the (person who played the) nurse clicked on the correct button on the screen to make choices since voice input was unreliable.
In addition,we also learned about the importance of inclusive design, which considers special cases like minorities and persons with disabilities. There was an emphasis on the need to gather users' needs before the start of the design and to think outside the box to cater to a larger population. During testing, we tested both story versions with all participants involved and focused on whether the prompts made logical sense in the context. The session gave chances for the family members to interact with each other, and the nurse would prompt Pepper to continue with the story by patting its head. Overall,we were successful in implementing their design and ensuring the inclusivity of our project.
Week 7: Evaluation
We explored the concept of human agent teamwork and how shared memory models and transactive memory systems are used to achieve common goals. We understood that performing a joint activity requires inter-predictability, common ground, and directability, and the theory of mind plays a role in attributing agency characteristics to self-moving objects. We also describe how situation awareness and coordination in teams are crucial for task management in high-risk domains.
We aimed to evaluate the effectiveness of a conversational robot in improving the mood and enjoyment of patients with dementia during meal times. The hypothesis was that a more interactive robot would create a more immersive and enjoyable storytelling session, which would motivate patients to finish their meal enthusiastically. The study involved 14 participants and was conducted using a within-study design, where each participant performed a story session with the robot and then reported our evaluation through a questionnaire.
The questionnaire used for evaluation was based on the Godspeed questionnaire, modified to test the patient's mood and enjoyment of the story. The study included two types of storytelling sessions: an experiment scenario, where the robot narrated the story and asked questions to spark conversations, and a control scenario, where the robot narrated the story and enacted conversations via voice modulations to portray different characters. The participants took on the role of the patient, while the team members played the roles of the formal caretaker and family member.
The results of the study showed that the conversational robot significantly improved the patient's mood and perceived it to be more natural and responsible, and they liked it more than the non-conversational robot. The patients also reported enjoying the conversational robot's storytelling more than the non-conversational robot's. Furthermore, all three of the study's added questions were significant, proving that conversational robots improved the patient's mood in comparison to non-conversational robots.