IN-CAR USER INTERFACE DESIGN
Sept - Nov
This thesis topic explores how different instrument cluster designs impact the ability of a person to calibrate their trust while driving a Level 3, conditional autonomous vehicle.
The user study tested three instrument cluster designs with 15 participants. The recorded reaction times, trust scores, and workload scores did not show significant differences between the designs. However, certain elements and layout styles across the three designs were perceived as beneficial for appropriating user tests and responding faster to take over alerts/requests.
In a Level 3 Automation or Conditional Automation "a driver is a necessity but is not required to monitor the environment. The driver must be ready to take control of the vehicle at all times." The automation notifies the driver when he might need to take over control. This notice is famously called the takeover request (TOR). Current research on TOR typically assesses the quality of TOR with reaction times and workload. The appropriate usage of a system is owed to the calibrated level of trust which is the level of trust that reflects the system's capabilities and performance.
How can we visually communicate TOR so user might establish appropriate trust calibration with the autonomous vehicle?
The research question I was exploring answers to was brought down to three main topics of discussion: The role of humans in autonomous vehicles, Trust Calibration, and Design for Takeover. I read over 40 research papers that explored these topics inside and outside the context of autonomous vehicles.
The authors speak about the requirement for transparent communication between the system and driver facilitating decision-making with appropriate awareness, confidence, and trust. Measuring an abstract and complex emotional process such as trust was an interesting challenge to journey through. Human Factors and psychology experts through their research break down the relationship a person can build with a system. With this knowledge, they designed several tools to measure trust in the context of autonomous vehicles.
I analyzed some of the Level 2 and Level 3 features available in cars today. The car UIs of Tesla, Waymo, Cruise-General Motors, Audi A8, and Cadillac were evaluated against the following questions.
User Perception Research
Though for a large part, this is research on a technology that isn’t widely available to the public yet, I found it would be interesting to gather how people imagine this technology to look like. I created a design activity by breaking down the pieces of information that we expect to go on the instrument cluster (based on previous analysis of car interfaces and research on Level 3 autonomous vehicles). I had five participants place these pieces on an empty instrument cluster image based on where they would imagine seeing this information. By overlaying the pieces placed by different participants, it was clear to see where on the instrument cluster most users would prefer to see certain information.
As part of the design activity I also recorded the participant thinking out loud. This data revealed how the participants would wish to see the TOR appear.
Design Activity Takeaways
TOR needs to be indicated with complete change in display with the color red.
The view of the lane is beneficial. The map view can also add value.
They would like to see obstacles that appear around the vehicle in the lane view
Displaying current and expected actions of the vehicle in autonomous mode.