Designing for Autonomous Vehicles
Exploring a new way of communication between pedestrians and self-driving vehicles
Role: Lead Product designer
Brief: Design for Human Robot Interaction
7 weeks group project
T H E C H A L L E N G E
How Can We Communicate Intent with Autonomous Vehicles?
For years, drivers and pedestrians have relied on simple gestures like nods or waves to communicate, especially when crossing the road. But as fully autonomous vehicles become more common, drivers will no longer be actively engaged—they may be reading a newspaper or watching a movie. This shift raises a crucial question: how will pedestrians, who have long depended on eye contact and gestures, understand a vehicle’s intent—particularly in situations without traffic signals?
O U R G O A L
System for Communicating a Vehicle’s Intent to Pedestrians
Our team aimed to design a system for self-driving vehicles that clearly signals their intent to pedestrians—whether the vehicle is stopping or has stopped for them to cross, or not.
Our key scenarios include situations where a self-driving car needs to communicate with a pedestrian attempting to cross when:
1) There is no traffic signal and no crosswalk, 2) there is no traffic signal, but a crosswalk is present.
T H E D E S I G N
A System Using Traffic Light Colors to Communicate Intent
We conducted a survey to understand how people interpret colors like green and red on a vehicle. Based on the results, we designed a system that uses these colors to clearly communicate the vehicle’s intent to pedestrians.
1) When the vehicle notices a pedestrian and recognizes person’s intention to cross the road, it blinks green light a couple of times and is slowing down.
2) After blinking green, it then remains green. The blinking green indicates that the car has seen the pedestrian. The pedestrian can start crossing once sees the green light.
3) Once the vehicle has come to a stop, the countdown bar in green is displayed, indicating how much time pedestrians has before the vehicle will start moving again.
4) After countdown, the light turns red to communicate to pedestrians not to start crossing now because the vehicle will start moving.
. . .
K I C K O F F
Defining Human-Robot Interaction
Our main task was to design interactions between humans and robots. In our Human-Robot Interaction (HRI) course, a robot isn’t just a humanoid machine—it’s any intelligent, autonomous physical system that impacts its surroundings. This includes Roombas, flying drones, and autonomous cars.
After extensive ideation to define our scope, we chose to explore communication between pedestrians and autonomous vehicles, focusing on future scenarios that are both innovative and feasible.
Our approach to Human-Robot Interaction Defenition
BRAINSTORMING
We brainstormed various approaches, including using anthropomorphic features like vehicle "eyes" and applying Disney animation principles such as squash and stretch or exaggeration to convey intent.
While we were excited by these bold, imaginative ideas and wanted to explore their possibilities, we also recognized that a vehicle changing its shape or color in real-world scenarios might not be practical for actual road use.
DESIGN PRINCIPLES
To evaluate the initial areas and the future concept, we established following design principles:
Communication should be universal regardless of languages and cultures.
Communication should have a low learning curve.
Communication should be socially acceptable so that it can be applied in reality.
*The word communication represents our design solution.
T H E R E S E A R C H
Survey to Validate the Concept
As we refined our concepts based on design principles, we found that all of them were reinforced by using colors already associated with traffic—red, green, yellow, and white. To validate this approach, we conducted a survey to quickly understand:
How people associate different colors on a vehicle with specific meanings.
Whether people perceive the colors as communicating the vehicle’s state or directing the pedestrian’s actions.
SURVEY RESULTS
We learnt that most people perceived the communication as what they should do and not the state of the vehicle. Also, Green was the strongest for ’Cross’ while Red was the most intuitive for ‘Don’t cross’. Here’s link to full survey.
C O N C E P T D E V E L O P M E N T
User Test to Validate Communication Methods
We had identified the right colors, but we weren’t sure if people would intuitively understand how they conveyed intent. To test this, we quickly built a video prototype to see if participants could interpret the meaning of the colors. While most initially found it unusual—since they weren’t accustomed to seeing these colors on a vehicle—they ultimately understood their intent.
To further immerse participants in a real-world crossing scenario, we built and tested a 3D prototype. We compared two scenarios: one with no communication of intent (the current state) and one using our proposed system. The study included a mix of qualitative and quantitative questions. Full test protocol can be found here.
“ I was not sure about it but, as I encounter this over time, I will be able to cross confidently.”
“ I picked it up from the convention of traffic lights.”
“Even though I cannot understand what exactly that means, I think the visual cue helps to decide whether to cross or not.”
USER TEST RESULTS
Of the five participants, all reported feeling more confident when the vehicle communicated its intent compared to when there was no communication. One participant didn’t fully understand the signals but still felt more assured with the lights, as it gave the impression that the car was "alive" and aware.
Additionally, the countdown green pattern was the most intuitive indicator for crossing—participants immediately understood its meaning upon seeing it. Full result here.
On a scale of 1 - 5, 1 is not at all confident and 5 is super confident.
G O I N G F O R W A R D
Reflection & Future Work
Our testing was conducted indoors using a small toy-sized prototype, but real-world scenarios involve many additional physical and emotional factors. These factors vary depending on the type of person, vehicle, environment, and more. Additionally, vehicle movement plays a crucial role in how people make crossing decisions, so this aspect requires further testing.
To enhance ecological validity, we would ideally test our system using a real vehicle. However, this approach is costly and complex. As a next step, we plan to build a virtual environment incorporating our proposed solution to observe how people react in a more immersive setting.