Societal challenge

Our world population keeps growing. People are getting older, but the amount of caretakers keep declining. In the meantime facilities at home are slowly digitalizing and almost all touchpoints have their own “complicated” interfaces. Causing an information overload for elderly people. This all leads to an unknowing and lonely dependent position.

Project objectives

• Design an adoptable social entity;
• Supporting users in their daily needs (e.g. groceries);
• Supporting users in their emotional needs (e.g. involving family);
• Supporting users in their medical needs (e.g. taking medication on time);
• Supporting users in entertainment needs (e.g. jokes);
• North star: enable older people to live independent longer.

Research

I started with a Porter's five-forces analysis for more insight in direct and indirect competitors. But also to gain more insight in current alternatives against loneliness and for self-reliance. There were (at that time) around 20 substitutes within the healthcare robotics. All in different shapes forms and roles. However, it seemed that Alice as a humanoid would be unique in her specific role as a grandchild. After this I started mapping out different types of hardware enablers that are not yet being used in robotics to strive for a more unique creation - for example the use of facial video mapping. It also became more clear that there are many Artificial Intellegence use cases proving it's maturity.

We conducted a total of 21 semi-structured interviews. I did this based on predefined persona's: psychologists, hands-on caregivers, a health insurance company and of course with dependent elderly people. A shared theme amongst all were the many cutbacks in healthcare. Caregivers working in the field want more quality time over quantity. They feel that the primary motivation to do their job is disappearing: 'It is becoming a checklist rather than personal care and caring conversations'. While elderly people themselves don't feel like bothering others for attention: 'That's just not part of my upbringing.' They also don't connect to other elderly peers easily as they're quite critical about eachother. They do however feel lonely and feel happier after social interaction in whatever form.

To further collect qualitative data, we organized a number of focus groups in collaboration with a nursing home in Baarn, The Netherlands. During these 'Coffee and Cake sessions', the objectives were to validate previous user insights, gain new emotional insights, but also to just have fun and be creative all together by brainstorming all the functionalities the perfect robot should have. The elderly were invited to draw, use paint, clay, etc. and pitch their creations and ideas to us and eachother in the different sessions. Not only was this a lot of fun for the groups, it also led to interesting discussions amongst elderly groups on what 'must have' functionalities are. After our focus groups we not only collected qualitative data but also gained a lot of insight in functional expectations. Among other things it became clear that childish toys are not taken seriously and of course could never be a substitute for actual human care takers. However, a social robot that offers social interaction and extra assistance where needed is a very welcome entity.

How do you quantify all gained insights in a secure but effective way? Well, by hosting an event for 84 elderly people! In collaboration with The National Foundation for the Elderly, I moderated in a group of 84 potential Alice users. It almost went as it would with a TV format; different interactive elements with an audience and a moderator. While they enjoyed an extensive lunch, I went through various first concepts based on different functionalities (Alice on a screen on the wall, Alice as a voice-only and Alice as a portable figure). Next to validating our gained earlier insights, this evaluated whether we were on the right track in terms of concept directions. In addition, we went through about 20 strong statements to fuel a group discussion. The public could vote in favour or against each statement by raising their blue or yellow sign (we didn't want to use green and red as it could connotate right and wrong). After voting I would walk through the audience to ask questions. It became clear that Alice should be portable and easy to lift. Another common agreement was that Alice should be able to set out contact with other people (neighbors, friends, family).

We gained quite some insights after our research as described above. Of which the following six key insights:

• 🔑 - Proper timing

Technology is (finally) mature. Current hardware and software enables us to actually build the intended social robot for the mass.

• 🔑 - Human caretakers

The role of actual human caretakers simply can't be replaced, no matter how social and humanoid a robot would be.

• 🔑 - Intermediary role

Elderly people crave connection with others (e.g. neighbours, family, old friends) but: ''I don't want to bother them for attention''. Alice should connect to others.

• 🔑 - Conversational partner

Alice needs to be a personal buddy in the first place - being able to start a conversation, see what the user is doing by scanning her environment and respond to situations.

• 🔑 - Appearance

Every user has their own interpretation of a childlike dependent. This should be taken into account when designing and building hardware (e.g. custom sizes, colors, etc.)

• 🔑 - Privacy first

''Healthcare authorities may use my data to help me, but never for their own benefit. My personal info should stay my personal info.''

After this researching phase I started mapping out all processes to get a proper overview of the healthcare ecosystem. Based on this overview we now started seeing different valuestreams that Alice could add to the current as-is situation.

• 'Alice should be the social orchestrator within a complex healthcare ecosystem.'

Currently, an elderly person is the centerpoint of the healthcare ecosystem involving many different interests and contrasting information. Often even just sending information and not capturing the viewpoint of the one who it's all about: the elderly person. Our value proposition could have a proper problem-solution fit: creating one simple front source of information, a source of information that orchestrates complex streams behind the scenes, funnels information and translates back to the end user and at the same time can involve others on behalf of the same end user. In the end Alice would be the social orchestrator within a complex healthcare ecosystem. This is were I want to circle back to the research done by Professor J.F. Hoorn - proving the key role of human form within social robotics. It's almost stating the obvious: we all own a clear "interface" already that has been iterated on and developed for countless of years which is our own face. Hence, part of our Value Proposition is creating a social entity using the most familiar interface in the history of mankind: The actual human face.

In the visualization above the to-be healthcare ecosystem with all value streams. It was a V1, intented to be a living visual to be iterated on based on new insights. In this ecosystem, Alice runs on a cloud based server functioning as the big brain behind every Alice out there. This 'robot brain server' can gather data and learn from every conversation and situation from different users making Alice effectively smarter. As the social orchestrator within the healthcare ecosystem, Alice could for example receive information from other smart products such as an an empty fridge, ordering new groceries based on personal diets. Or sensing a cold temperature and asking if it's desired to turn up the heat as a favor for Alice. Another connected branche in this ecosystem would be to fulfull the user's emotional needs by stating for example: ''I feel like calling someone, do you want to call your niece Lisa?''. This action would be initiated based on an amount of time without any human interaction. This works the other way around as well, by sending a text message to Lisa asking if they feel like calling with grandma for couple of minutes as it would cheer her up. Another example of orchestrating would be to track medication intake to be on time and ordering new medication when needed.

Designing for robotics was a new type of project for me personally. Projects with robotics are very broad and involve a lot of green field thinking. It is therefor very tempting to get started aimlessly (out of passion). But without a clear policy and clear scope, me and all other experts are just doomed to fail. Which is why we created a roadmap with the first two phases including appraised streams: Research, Design, Development & Security, IP & Legal Identity, Business Development and Marketing Communications. My main role was within the Research and Design teams.

Without diving to deep into why you need to define a Minimum Viable Product, with a project this size it's definately a must have. To be able to test rapidly and gather real user feedback, you need the shortest route to go live. This by creating a product with just the minimal amount of functionalities to be viable. Not only is this a human centric approach, it's also very cost effective.

As Alice orchestrates all different user needs (daily needs, emotional needs, medical needs, entertainment needs), we chose to zoom in on the emotional user needs. Main reason for this is that emotional loneliness is a hot topic and our primary societal challenge. Secondly, we thought that the most unique value in the current market would be achieved by focussing on emotional needs. This way we learned fastest on what creating a social entity means. More specifically, we wanted to design at least one complete conversational use case that fulfills emotional needs.

This use case aims to especially reduce loneliness amongst likeminded peers - establishing a meaningful friendship. Sometimes even more. This meant that we needed Alice to be programmed to have a short conversational 'onboarding' to create an user profile with personal interests. After this Alice searches for matches in the back-end with other Alice's and saves them. When there is no human interaction for an X amount of time, Alice will ask the user to call with someone to socialize and just for fun. If the user is interested, Alice will initiate contact with another Alice (and her user with similar interests). Matching happens within the same region. It also never has to end in awkwardnes: Alice will fill in silent gaps by asking relevant questions or even by making a joke. This happens when it's silent for more then 5 full seconds. When the conversation is done, Alice will evualate afterwards, discussing three options with the user: 1. end the friendship there; 2. plan another call with the same person or 3. arrange a meeting in person. After which other processes within the ecosystem get involved. With this MVP we tested how social robotics can play a role in human conversations. Our MVP was ready when the conversation was working completely autonomously.

Thinking big and starting small is commonly recognized by now as the right approach when building products. You start with a proposition that is viable and feasible and gather user feedback for further development as quickly as possible, keeping the north star in mind. But there is a steady shift in what 'starting small' means. Beauty has importance. It moves consumers and evoques emotions. To add business value you need beautiful delivery as well. With this in mind we strived to move away from the MVP as fast as possible and create a loveable Alice by also accelarating on hardware development. It made most sense to focus on non-verbal communication (e.g. cute googly eyes), the sound and speed of her voice and the way Alice should feel in terms of used materials. We also tried to make Alice talk more informal and be funnier in her way of answering. All to make elderly users actually remember Alice as a social robot they love rather than like.

• Hardware

Testing Alice's hardware was mostly focused on autonomic survival. How fast would the battery be empty when we take out the power cable? How would our latest prototype handle falling on the floor an X amount of times? etc. The team started with simple survival tests of 1 hour building up to 4 hours, 8 hours, 24 hours and so on. The Alice project is still ongoing, which means testing and iterating is still taking place.

• Software

In terms of testing the software side of this project we started our testing in the beginning of this project with analog roleplay to test conversations with users in person. We did this by asking one of our lovely female colleagues to come with us to the sessions - only to hide 'behind the curtains' with a microphone. Her voice then would be the voice coming out of the speaker of Alice. In this way we could experience at first hand how elderly people react to Alice as a social robot and how simple day-to-day conversations would go. After this phase, real conversations with Alice herself were held. Together with either me or another representative of the project to help a hand were needed. One of the developers within our team would come along to solve fixable issues on location. In the end it would be testing with Alice and her user alone while we remotely track and follow along.

A large project with equal sized challenges. Sr. Manager Marly Kiewik did an amazing job proving her extensive experience. I did have my share of managerial responsibilities next to being one of the designers. My learning is that there are a lot of different streams involved in setting up such an ambitious robotics project: Research, design, development and security, IP & legal identity, business development and communication. Many different disclipines that always have to work in parallel but don't always work together in person. The project got delayed sometimes when teams weren't on the same page and/or speed. My take away would be to plan more frequent alignment sessions (but shorter) instead of ellaborate alignment sessions once every few weeks.

A project that was close to the heart as I feel a lot for this specific vulnerable targetgroup. Which made me do this project with lots of passion. I also learned a lot, high fived many smart people and I want to wish all the best to the current project team in further developing Alice. Interested in the project? Reach out to Marly Kiewik (Director at Deloitte Consulting) or Franklin Heijnen (Creative Director at Deloitte Digital).