As the first project in the Interaction Design Studio at Carnegie Mellon University, we were tasked to conceptualize and integrate an intelligent voice assistant in existing product ecosystem.
Our team designed Dottie, an AI voice assistant for Pennsylvania Department of Transportation (PennDOT). We identified opportunities where voice assistant could intervene to provide personalized, seamless & assistive experience across different devices, including mobile, desktop, ipad, and kiosk screens, specifically for people who need to apply for drive license and the DMV staffs.
Welcome
Speaking
Listening
Welcome
Processing
Idle
Success
Error
I just moved from another states, I am not sure what should I do in this case....
Using Natural Language processing, Dottie provides Individualized requests for prescreening and process overview before the users start the application
I can't find the required documents, the information on the website is overwhelmed...
Organizing complex information into simple card design and actionable suggestions, Dottie provides a clear way to review process overview, document checklist, location detail, alternate documents List, and so on
I don't know what is my next step...
Dottie provides a cross-device stepper with linear overview and non linear detail steps to help the users keep track of their application process more easily
My parents don't speak English so visiting the DMV is always an intimidating experience for them...
Understanding users language preference, Dottie offers real-time language translate at the counter for both applicants and DMV staffs
We started researching a range of companies across the globe and listed out what features and opportunities a CUI has. We then categorize these features to see which can be beneficial for each company. We as a team chose the Pennsylvania Department of Transportation (PennDOT) DMV because we see the potential for a CUI to ease the ambiguity, offer accompanies, and save time for the users during the driver's license application process. Also, PennDOT DMV encompasses a wide range of users, and applying for a driver's license is an activity that has a clear end-to-end journey with steps and users can benefit from suggestions and guidance.
To understand the current DMV experience with the existing technologies - Website and Check-in Kiosk, we research on the website and went on a field trip to identify the current user flow, pain points, and opportunities for conversational assistant. This helped us be able to think about why and how a CUI could be beneficial for DMV users.
interviews with people who have experiences applying, renewing, or transferring their driver's license in PA.
The participants are age from 16 to 57, which covers a wide ranges of users.
Base on our field observation at the DMV offices and interviews with the users, we individually created storyboards that covers the needs and brainstormed how CUI could be helpful throughout the process.
Through the research, we identified several pain points and opportunities that application and DMV staffs could benefit the most from a conversational assistant.
To identify the most plausible and valuable features that a conversational interface can provide in a DMV experience, we first defined the target audience we are designing for. From the stakeholders' map, we identified different groups involved in the process and decided to focus on Driver’s License Holders and Applicants since they are the most frequent users of DMV services.
To find out the situations in which our CUI would be beneficial to passengers, we drew storyboards encompassing key scenarios, possible intervention approaches, dialogue between applicants and CUI, modality, and applicants' emotional needs. This enabled us to move forward to the ideate and design phase.
The name Dottie comes from the abbreviation DOT (Department of Transportation).
We wanted Dottie to be an accompaniment for applicants throughout the application process. As DMV experience usually includes uncertain, situations, strict rules, and various groups of people, we wanted Dottie to be friendly, trustworthy, and approachable.
To create a visual representation of Dottie, we assembled all our references on a whiteboard and collectively brainstormed ideas. Our brainstorming primarily revolved around key concepts like roads, wheels, and driving, as they align with the concept of DMV services.
Final form of Dottie consists of on dot and line, which represent the wheel and the road. We keep the visual simple and clean considering a clear visibility on the small mobile screen and smoother transitions between different states
We opted for the Barlow and Inter fonts to maintain a clean and straightforward visual design, facilitating easy scanning. We retained the primary colors of blue and green from the current visual system, making adjustments to enhance contrast and readability. Additionally, we introduced rounded-shaped icons to align with the personality
We identified states needed and mapped out them on the active-passive and positive-negative matrix. This process assisted us in determining the appropriate actions for each state
.gif)
Prescreening & Process overview
Stepper
Location suggestion
Reliable alternative options
Document checklist
Appointment notification
Check-in
Stepper & Document checklist
Waiting queue number
-1.png)
Queue number
.png)
Language assistance
Regarding conversational design, I've gained insights into the importance of simplifying information and offering actionable recommendations. Additionally, I've discovered the value of presenting information in a more digestible format, beyond solely relying on plain text.
A diverse range of users encourages us to consider and address various perspectives when brainstorming solutions. For instance, the inspiration for the language assistance feature came from an interviewee who shared their immigrant parents' challenges with language barriers at the DMV.
Within this project, we've created designs for mobile, iPads, desktop, and kiosks. I learned that maintaining design consistency is important when implementing features that span across these diverse devices. Furthermore, I learned the importance of considering scalability on various devices for better readability when designing the stepper feature.
Since I am new to conversational design, I initially found it challenging to determine the scope of work and the best way to delegate tasks and work collaboratively. Fortunately, we've established an effective approach where we distribute tasks based on our interests and strengths and collectively engage in discussions to decide scenarios and narratives. Subsequently, we focus on individual contributions for more detailed aspects such as UI and motion design but maintain a continuous feedback loop by returning to discuss our direction.