The Future of Autonomous Vehicles
Speculative Research Project on AV Trust and Safety w/ Johns Hopkins APL
SPONSOR
ROLE
TIMELINE
JHU APL
UX Researcher • Team of 9
January - May 2025 (16 weeks)
Problem
Autonomous vehicles may be technically capable, but public trust remains fragile — especially in dense, unpredictable environments like college campuses.
Constraint
We were not designing policies or infrastructure alone. We needed to design a believable user-facing experience that helps people understand and trust AV behavior in real time.
Outcome
We designed a campus-based autonomous shuttle experience, supported by an in-vehicle interface and a design fiction vignette, to explore how transparency and communication shape trust during uncertainty.
PROJECT OVERVIEW
The Johns Hopkins Applied Physics Lab is a not-for-profit UARC (University Affiliated Research Center) focused on solving complex challenges and developing novel solutions for public service and innovation. As autonomous vehicles become increasingly prevalent, ensuring the safety and trust of all road users will be crucial. This project focuses on understanding how pedestrians, cyclists, and other vulnerable road users might interact with autonomous systems in the year 2040, especially on college campuses, while also examining the coexistence of autonomous and human-driven vehicles. A key objective is to foster user trust in autonomous decision-making and mitigate risks associated with distracted or unpredictable behaviors. Through user-centered strategies, the project seeks to inform future policies, technological development, and experience design to promote a safer, more efficient autonomous transportation ecosystem.
PROBLEM STATEMENT
How might we support perceptions of safety and decision-making of campus road users in an autonomous future?
While autonomous vehicles are advancing rapidly in capability, trust often breaks down in moments of ambiguity. When a vehicle pauses, reroutes, or behaves unexpectedly, users are left guessing. Even safe decisions can feel unsafe when their intent is unclear. The challenge wasn’t autonomy itself, but rather interpretability. If people can’t understand what an AV is doing and why, confidence erodes and adoption stalls. Designing for trust meant designing experiences that make AV behavior legible in real time.
RESEARCH
Exploring the current state of autonomous vehicles and how pedestrians and drivers navigate shared space on college campuses
In order for us to make an opinion on the state of the future, it was important that our team had an understanding of the current state of road usage in heavily trafficked areas, such as a college campus.
Our research questions included:
When do people feel the most uncertain around autonomous vehicles?
What information do users want in the moment — and what overwhelms them?
How does context (campus vs. city streets) change trust expectations?
To answer these questions, we conducted desk reviews on public trust and perceptions on AVs, existing policies and regulations, and human–vehicle interaction, interviews with subject matter experts and road users, and scenario observation and analysis focused on campus environments.
Key Research Insights
INSIGHT #1
Users were comfortable when AV behavior was predictable. Trust dropped sharply when vehicles paused, rerouted, or behaved unexpectedly.
Implication: Design should focus on edge moments, not just ideal flows.
INSIGHT #2
Users did not want full system logic or sensor readouts. They wanted plain-language explanations that answered “what is happening” and “why.”
Implication: Transparency must be interpretable, not exhaustive.
INSIGHT #3
Campus users expected AVs to behave cautiously and communicate clearly due to dense pedestrian presence and social norms.
Implication: Trust signals must adapt to environment, not remain static.
Across early milestones, our work moved from understanding system-level challenges to identifying interaction-level opportunities. Initial explorations focused on mapping stakeholders, policies, and AV ecosystems. While this provided valuable context, it became clear that these layers alone did not explain why people felt uneasy around autonomous vehicles.
As we synthesized interview data, scenarios, and literature, a consistent pattern emerged: discomfort wasn’t rooted in abstract concerns about autonomy; instead, it was triggered by specific moments of uncertainty. Pauses at crosswalks, unexpected reroutes, or slow approaches to pedestrians created tension, even when behavior was technically correct.
Concept Development & Prototyping
Showcasing the Future of AVs and their Integration into College Campuses
With a grounded understanding of both AV technology and campus behavior, we moved into future-facing design. This milestone focused on three threads: policy ideation, speculative vehicle design, and storytelling.
We first facilitated a policy ideation workshop to explore “what if” scenarios across pedestrian safety, VRU (vulnerable road user) protection, emergency response, large campus events, and infrastructure. We then grouped these ideas into recommendations for four stakeholder groups: vehicle manufacturers, AV service users, lawmakers, and campus transportation teams.
In parallel, we sketched and prototyped a future-state AV designed for a campus environment:
Exterior: SUV-like form factor with LiDAR + 360° cameras, turquoise lighting to distinguish AVs from traditional vehicles, and front/back displays plus projected crosswalks to clearly signal intent.
Interior: Reconfigurable seating, folding tables, wheelchair ramp access, and an interface designed for calm, limited cognitive load rather than flashy features.
Interface: Hazard pop-ups, ADAS visualization (showing how the AV “sees” the environment), concise navigation, and lightweight controls for speed, climate, and media.
To bring these ideas to life, we created a design fiction vignette framed as “A Day in the Life of an AV.” Rather than centering solely on one user persona, the vignette follows the AV as it serves multiple riders throughout a single day highlighting how design features and policies converge in everyday scenarios. Each story highlights a different aspect—like access ramps, charging hubs, or pedestrian signaling—showing how the AV system responds to real constraints and behaviors on a college campus.
OUTCOMES
Designed interaction patterns that make autonomous vehicle intent visible and understandable in the moment, reducing uncertainty during pauses, reroutes, and edge cases
Reframed trust as an experience design problem by using plain-language explanations and real-time cues to reassure users during moments of ambiguity.
Explored how autonomous systems should adapt communication and behavior to dense campus environments, where social norms and shared space shape trust expectations
FINAL PROJECT REFLECTION
This project demonstrates that trust in autonomous vehicles cannot be solved by engineering alone. Even technically correct behavior can fail at the experience level if users cannot understand what the system is doing in real time. That gap between system intelligence and human interpretation is where adoption breaks. By focusing on interpretability, real-time intent communication, and context-aware behavior, this work reframes trust as a design responsibility, not just a safety metric. It shows how interaction design can actively shape public acceptance of autonomy, especially in high-visibility environments like college campuses where social dynamics and unpredictability are constant.
Rather than treating transparency as a data problem, this project positions it as a communication problem. That shift is critical for AV teams building products that need to coexist with pedestrians, cyclists, and first-time users. Designing legible autonomy is not just a nice quality but a prerequisite for large-scale adoption.
To move this work toward real deployment, the next phase would focus on validating interpretability patterns in live or simulated driving environments, ensuring that intent cues are accurate, timely, and aligned with vehicle decision-making. These transparency behaviors would then be adapted across different contexts (campus, urban, transit) to support scalable, environment-specific communication rather than a single fixed interface.
Dive Deeper
For this project's final deliverables, our team compiled the following products.
.png)
