Click to watch the AVAI Trailer
Role: Lead UX Researcher & Designer
Deliverables: 15-minute research finding showcase presentation
Background: A grant research project for the future of mobility communications and the introduction of technologies into communities with trust and confidence.
Solving for
How do we build trust between autonomous vehicles and pedestrians in a dense city?
Vehicle and bicyclist accidents have increased over the years. Poor road-sharing communications are the leading concerns. With the future of autonomous cars near, communication for mixed-technology traffic is even more important.
Manage designers and data scientists
Develop objectives
Lead Research & synthesis
Present findings to stakeholders
Responsibilities
Bumps & Pivots
Initially, the objective was to research a future solution for personally owned autonomous vehicles, but in discovery, it was learned that there are over one million registered vehicles in Chicago. To stay within the grant limits, I pivoted the audience based on research data.
Chicago buses encounter bikes most often and at the most dangerous times of the day.
1,560,417 registered personal passenger vehicles
1,864 CTA buses in operation for 129 routes
19,237 CTA bus trips a day
Predictive Communication Patterns
A communication of action intentions based on a prediction algorithm of bicyclists’ paths.
The buses have 360-degree screens with live confidence tracking lights that cluster at the location of any human agent near the bus. This is reflective of the “communicating intent” when we look to make eye contact with other commuters.
There is an advanced display lights on all sides of the bus for the predictive path. This shows where the bus believes the bicyclist is going to go. This mimics the “second nature” wave we give other commuters to proceed into the intersection or go ahead.
Diagram of AVAI bus
A future with autonomous vehicles is closer than we think. Companys are already testing their level 4-5 robo taxis in many large cities. Combined with personal vehicles that have ADAS levels 2-3, we are already experiencing mixed technology roadways.
When the public perceives self-driving technology in a negative light, companies will have a hard time introducing it to our communities.
The research showed that bicyclists can be the most unpredictable but also the most vulnerable agents on the road. With limited time, the focus was narrowed to research a specific future problem.
How can bicyclists and autonomous buses safely and confidently communicate on Chicago streets?
“It would help a lot if cyclists would follow the rules of the road”
Research documentation
Do you see me? I see you.
No or unclear communication led to accidents (both with drivers and bicyclists)
Bicyclists are the most unpredictable and most vulnerable
There is a need for smart infrastructure implementation
An insight takeaway was to solve the simple communications that are taken for granted.
Waving others on the road to go head-on or making eye contact for that confidence before crossing the street is second nature. Both drivers and bicyclists mentioned that they feel most confident and safe when they know others see them.
Artificial Intelligence
Recommended hand signal for stopping
The AI will learn from suggested bicycle safety hand signals however based of the generative research it was learned that not all bicyclists follow the recommended hand safety communications.
Not wanting to only train the AI on hand signals but also body language in the event a signal isn’t given. Taking a page from Tesla’s “Shadow Mode” AVAI will partner with Divvy to collect data right from the source, Chicago bicyclists.
Acknowledging that training predictive AI-based on images and Divvy collected data alone would take hundreds of hours.
Simulations scenarios for AVAI
AVAI would use simulation training with our specific target audience of bicyclists and even motorcyclists. Who moves in similar ways as bicyclists and have same recommended hand safety signals.
How can this translate and create an intuitive user experience for the Chicago bicyclist riding with autonomous CTA buses?
I come back to the second nature and communication intent of the glance and wave. The autonomous bus has its predictive model of where the bicyclist is going, but the bicyclist has no idea what the bus is going to do. Participants mentioned this pain point: not trusting vehicles without human drivers because they don’t know where the bus will be moving to.
How do we create confidence-inducing UX design?
This is how, one part of the experience is intelligent live pedestrian tracking light on the buses. They cluster near the location of a human agent and dissipate when they leave. The bus will show its AI pedestrian predicted travel path with an advanced lighting puddle light of where it thinks the bicyclist will go.
Here, an AV bus and bicyclist are approaching a traffic light. The bicyclist wants to make an unsafe maneuver by turning left and crossing in front of the bus. The bus sees what the rider is doing and knows it’s unsafe but it acknowledges the bicyclist’s intentions, slows for the riders safety and shows the predicted path. If the bicyclist was in immediate danger the bus would stop.
Introducing new technologies into our communities would included building confidence with an advanced knowledge campaign. There would be a social media and poster campaign in the city to teach the residents the language of the autonomous CTA buses.
AVAI introduction poster