Manali - Self Reflection

**Week 1**

We were introduced to the coursework project that we would be working on: Building a robot to support patients with dementia.

4

5

The concept of human centered design was introduced along with its international standards, which was central to the project: how to build robots which are beneficial and focused towards the (human) users? The system should maximize the extent to which the system can be used efficiently, effectively and with satisfaction. The system should consider the social and physical environment of use, the stakeholders and user experience.

6

7

We learned about dementia and the affordances of a robot as a social actor in patients' lives. To further elaborate, two previously studied use cases were introduced:

8

9

~1. REJAM: A robot for elderly patients with dementia in a care home, which used music therapy to facilitate processes of emotions and cognition. This robot was deployed in Pieter Van Foreest, a care home for the elderly in Akkerleven. Apart from playing music, the robot was played games, acted like a companion and gave individual attention to each person. All activities encouraged reminiscence and helped patients connect with others in the care home. This was a new concept to me: it was nice to see robots taking on the role of encouraging people to socialize, play games and evoke memories and emotion.

10

11

2. PAL: A robot for children with diabetes, which supports them in day to day activities like reminding them to take medication, and checking blood sugar levels. It helped the parents be at peace about the sugar level and health of their children.

12

13

For the project, there was a guest lecture which clearly explained how to define social actors. It was interesting to see how giving a simple hobby to each actor makes them seem more human to the developers. We defined the actors for our use case, and made a storyboard for the use case we wanted to focus on. We picked the use case of a patient with dementia suffering from sundown syndrome, which reduces their motivation to have meals. The robot will narrate a story to engage the patient while they complete their meal.

14

15

Different robots available were introduced: Pepper, Nao and Miro. We decided to use Pepper as it went hand in hand with the design we had in mind: narrating stories, having a tablet to show images, and the required gestures and movements.

**Week 2**

Since our project pertained to a human centric approach of a robot for patients with dementia, it was quite fitting to have a guest lecture which made us aware of the medical aspects of dementia: as engineers, this is a side which we might not know. We learned about the changes in the brain while ageing, and the possible decline in memory and cognitive function. We learn about the different types of dementia, and the ways of diagnosing it. We gain an understanding of the current interventions and treatments: these could possibly be included in the robot as well for higher effectivity. Music too plays a major role by activating parts of the brain which affect emotion, attention and physiological aspects.

20

21

Even without the perspective of this course, it was quite enlightening to see how lifestyle changes can improve the quality of life in the long run: something which most of us ignore.

22

23

All in all, it was a very informative session from the perspective of understanding the target audience and their needs better, and the biological outlook. This will particularly help in designing the solution better, instead of shooting shots in the dark of what we, as engineers, think will help.

24

25

In the project, we finalized the personas, stakeholders and the problem and design scenario. We designed the problem scenario to be in the evening, where the patient suffers from Sundown Syndrome, and hence faces mood swings when the sun goes down, and loses motivation to eat due to this. The design scenario is centered around getting the patient to complete meals while enjoying them, also facilitating conversations between family members: this is done while the robot narrates stories, providing prompts for interaction.

26

27

This incorporates the "SCE Method" which considers the involvement of stakeholders, and a multidisciplinary (biological, engineering, musical) and iterative design process to perfectly refine the core functions and design rationale.

**Week 3**

In this week, we dive deeper into PAL, the robot which assists children with diabetes, and see how this was approached. It gave an insight into how we must define our use cases, environments, stakeholders, values and personas. They use the zone of proximal development to facilitate collaborative learning in children. It is very clearly shown how the demands, human factors and technology are all in sync with each other, as should be in any project. The project is viewed as a partnership between humans and robots, where they both work together to achieve objectives and goals.

33

34

We finalized our personas, stakeholders, environment, design and problem scenario for the project in this week, for the presentation in Week 4. We also looked for supporting literature and research which we used in our design scenarios. We decided to use the Pepper robot too.

**Week 4**

I gave the first presentation for my team, and described the premise of the project we were working on: the personas, stakeholders, environment and problem scenario. It was intriguing to see what the other groups had been working on, and I did notice that ours was the only group which targeted patients at home, with family, and not in a care home. This gave us a whole new scenario we could work with, and more freedom to involve family members. The downside of this, as we realized, was that it will be costlier to deploy the robot for a single patient, compared to that in a care home.

40

41

We were taught the design of use cases, along with the team design pattern (TDP), which I feel is the heart of the project. It gives a detailed overview of the whole project and provides a visualization of how the team envisions the project to work. We give a reasoning behind why each feature was chosen, what features were debated and not chosen, and the reasoning behind it.

**Week 5**

We learn about team design pattern (explained above) and interaction design pattern (IDP). The latter shows how the designers expect the different actors to interact with each other in the context of the robot. The IDP shapes the TDP. It is shown that postures, gestures and expressions of the robot can have an impact on the user: it is essential because it makes the robot more human like, and the users feel more connected to it.

47

48

We learn how to evaluate our prototype robots: there are different factors to consider:

49

50

~1. Practical issues: We could not evaluate with actual patients, and did not have the time to build all the features we wanted, like a story template, or a robot which recognized the patient's mood and decides the story.

51

52

2. HREC: To evaluate prototypes with participants, we must have permission from the Ethics committee, which we had for this course.

53

54

3. Formative vs Summative Evaluation: For the project, we go ahead with a summative final evaluation due to limited time to perform a formative evaluation.

55

56

4. Measures and their levels: We used the Godspeed questionnaire as it seemed relevant to measure how human like and likeable the robot was in the use case. We added custom questions pertaining to the use case. This used the interval level of measurement.

57

58

5. Tools: Qualtrix for the survey

59

60

We also gained an insight into ontologies: what they mean and how they can represent the users and their properties in one diagram. It is an explicit representation of knowledge: the entities used in the system, what they are and their structure. Ontologies represent classes (patient with dementia), their instance (Georgina), relationship (mother of Sam) and properties (80 years old).

61

62

This week we performed the pilot study of our project. We worked with Pepper at the Insyght Lab to test the basic connections and familiarize ourselves with the robot. The tablet did not work, which we later realized is an issue in the connection with InteractiveRobots. We tested voice inputs and touch inputs, and came to the conclusion that voice input may not be so accurate, so went ahead with a backup touch input instead.

**Week 6**

We understood the meaning of an inclusive design, and how we can use it in our projects as well. An inclusive design is one which not only considers the "average users" but also special cases, like minorities and persons with disabilities. We were encouraged to look into our own projects and see how we can better our design and make it inclusive. We must not assume anything about users, but quite the opposite, think outside the box and see how the design can cater to a larger population, those including people with disabilities.

68

69

Secondly, we must gather users' needs BEFORE the start of the design, and not assume what they may or may not need, because more often than not, designs turn out to be redundant for the users themselves.

70

71

This week, we tested the interactive version of the robot with the final code. We had two story versions: Thanksgiving and a Picnic. Both of them were tested with all participants involved: nurse, family member and the patient. We tested the prompts and focused on whether they made logical sense in the context. Since voice input was not reliable, the nurse had the job of clicking on the correct button (on screen) to make choices. It gave chances for the family members to interact with each other, and when the interaction was done, the nurse would pat Pepper on the head, prompting it to continue with the story. This session was a success in terms of testing out the code that was written.

**Week 7**

We learn about the concepts of human agent teamwork, and the different team aspects. We learn about the requirements of performing a joint activity: interpredictability, common ground and directability. The theory of mind plays a role here, which states that humans attribute agency characteristics to self moving objects. Human Agent teamwork uses shared memory models and transactive memory systems, where both humans and agents have certain knowledge, and they share it to achieve a common goal. This is similar to human-human teams as well. We learn about situation awareness in high risk domains, and how the Shared Memory Model and Transactive Memory System, help in gaining better awareness. We gain an insight into coordination in teams and task management: the ways in which teams work together to fulfil a goal.

76

77

In this week, we carried out our evaluations of the robot with 14 participants. It was quite an experience to see how different people have unique reactions to the robot. I also participated in other group evaluations, and it was interesting to see how everyone used different features of Pepper to achieve different goals.

78

79

After the evaluations, I carried out the statistical test to determine significance and also helped in creating the presentation.

**Week 8**

My teammates presented our final presentation this week. It consisted of a recap, the use cases (TDP and IDP), the evaluation method and the final results. I believe it was a fulfilling team experience, where everyone played their role for the final outcome.

Wiki source code of Manali - Self Reflection