Pia
An intelligent AI for the Pittsburgh International Airport designed to provide a seamless, hassle-free travel experience.
SCOPE
Carnegie Mellon University
Interaction Design Studio
4 Weeks, 2021
TEAM
Matt Muenzer
Youngryun Cho
Wei-Cheih Wang
Devika Pillai
ROLE
Concept & ideation
Identity design
User interface design
SOFTWARE
Figma
Voiceflow
Premiere Pro
Aftereffects
THE CHALLENGE
Design an experience that showcases novel interactions between human(s) and a virtual assistant.
THE OUTCOME
Pia is a conversational assistant for the Pittsburgh international airport providing a connected, single-stop solution for all travel queries. As a data-rich assistant, she connects the different pain points and facilitates a seamless experience for both the airport and its passengers.
My role involved design research, concept ideation, identity design, and UX design.
FEATURES
Intelligent transit suggestions
Transit options customised to match flight time and available realtime transit data.
Proactive prompts
From carrying essentials, saving your parking information to helping you leave on time.
Parking and payment assistance
Parking options with visibility on available spots and walking time to reach the gate. Voice secured access to saved payment options.
Location based service discovery
Explore dining, shopping and other service experiences near your departure gate or specified location.
Multilingual support
Recognize languages based on the starting prompt and assist.
Staff assistance
Reach out to the airport staff from any kiosk at the airport for easy assistance.
Navigation and transport guidance
Find the quickest way to leave the airport.
RESEARCH AND FRAMING
Why we chose an airport, what our insights were and how we defined our core solutions.
CONTEXT AND CLIENT
Flying commercially can sometimes be described as nothing more than a series of inconvenient events. From late or canceled flights, long lines, crowded terminals, appalling food, and a series of long procedures and steps right from leaving our homes to the point of boarding our flight, the current traveler experience can be complicated and disconnected. Pain points were considered for experiences before the airport, at the airport, as well as for post-travel engagement.
Pain point mapping
Field visits and contextual inquiry
.png)
Learning about the existing interface
INSIGHTS
Transit planning is a common painpoint and the parking services are outdated and time-consuming.
People dread time at the airport and need suggestions into the services offered and things they can do.
Digital interfaces lack navigational support, leaving passengers unable to explore the space without walking.
Airports are often understaffed. Travelers are unsure about how to seek help during novel situations.
The app and kiosks are weak touchpoints covering functionality better provided by airlines.
International passengers require multilingual support across signage, interface, and verbal assistance.
DESIGN OPPORTUNITIES
Commute
Parking, Ground Transport, Travel Time
Procedures
Check-in, Delays, Luggage, Security, Navigation
Discovery
Dining, Shopping, Play Areas, Other experiences.
USER JOURNEY MAPPING
With a better understanding of scenarios for our solution, we mapped a user journey that highlighted the context, emotions, and suggested interface modals.
CRAFTING THE IDENTITY
Tapping into the existing visual affordances of mobility and intelligence to design how Pia looks, sounds and moves.
Logo and form
Considering the symbol represents not just a static logo but an on-screen intelligence that communicates through its shape and sound, it was important to consider how each of our forms moved. Our final form was chosen for invoking qualities of navigation and movement while providing versatile transformations.
Palette
Crisp and confident. Dominant on blue to stay close to aviation and sky shades.
Typography
Cabin to tap into classic wayfinding and informational aesthetics. Rubik to keep it conversational and soft.
Icon style
Two-tone icons from the Google library were stylized to reflect the translucency and opacity play from the logo.
Rubik
Cabin
Tone of voice and speech
We chose a voice that best suited the personality traits that laid out.
Motion states
Designing to assist for non-verbal human interaction was a key consideration for the motion design. We extended the conventional limits of conversational AI to account for other naturally occurring states in our conversations.
Resting
Awake
Listening
Listening (Paused)
Speaking
Processing
Success
Error
INTERFACE DESIGN
.png)
.png)
MOBILE
The graphical interface was retained for easy visibility of all services. The voice assistant is provided as a primary affordance on the home screen.
Home
Directory
Alerts
Conversation Overlay
VUI Overlay
Asking Pia a question opens up a conversation overlay. The states of the VUI are matched by supporting GUI patterns on the screen.
Resting
Listening
Processing
Speaking
Prompting
.png)
Parking Card
Transit Options
Restaurant Listings
Navigation
Suggestions
KIOSK
VUI Overlay
Similar to the mobile, the voice assistant's overlay follows interface patterns for each state on the airport kiosk.
Resting
Prompting
Speaking (Collapsed)
Listening
Processing
Speaking
Kiosk hardware
Since our solution involved a kiosk, we wanted to leverage the different ways in which the hardware could support the user experience.
PIA BEYOND PITTSBURGH
How can Pia support operational staff and optimize passenger experience across airports?
.png)
Bridge flight and passenger data to drive savings and reduce footprint.
.png)
Increase retail, parking, and service visibility to enhance engagement.
.png)
Use shared data & predictive AI to improve the travel experience.
LEARNINGS AND TAKEAWAYS
Design for Proactiveness
While considering scenarios and use cases it was important to factor in user needs beyond the immediate user request to craft associated prompts that would help provide a stronger experience.
Reconsider your GUI
Since this was a voice interface, it was extremely important to design screen patterns that gave appropriate feedback, helped the user move forward, and forgave mistakes.
AI + Human
While preparing the scripts, we considered where the user was delegating, controlling, and collaborating. This made me think about the responsibility involved in designing AI interfaces.
Scroll to Top