In-vehicle displays are generally designed for one driver, or for none in particular. Neither fits a shared car with an amateur driver in it: the settings that make a car safer for a newly licensed driver are precisely the ones an experienced driver would find restrictive, so a single fixed configuration cannot serve both. CARID starts from identification — establish who is driving, then apply the profile that suits them — which turns safety configuration from a setting someone has to remember into a consequence of starting the car.

Participants

10 participants in the usability evaluation, ranging from 18 to 57 years old — a spread chosen so the interface was tested by drivers on both sides of the experience gap the system is meant to close.

Study Design

Five research goals framed the work: uncovering user motivations, understanding what drivers need from verification, identifying pain points, determining how much tutorial support the system required, and assessing the intuitiveness of the three core functions — the ID scanner, the alert system, and customisation.

Design proceeded through personas, storyboards and hierarchical task analysis before any interface was drawn, then into paper prototypes and rapid prototypes built in Axure. The evaluation had participants complete research tasks and surveys while we recorded task completion time and error counts, so that self-reported satisfaction could be read against observed performance rather than in place of it.

Participants rated satisfaction at 4 out of 5. Editing the notification radius for the dangerous-driver alert took 57.1 seconds on average, with 1.1 errors per participant — slow and error-prone for what should be a routine adjustment.

The pattern behind those numbers was navigational, not conceptual: users struggled initially, then reported satisfaction once they had found their way around. Nobody misunderstood what the features did; they could not reach them efficiently. The specific culprits were redundant access screens and the decision to keep system customisation outside the main menu, both of which added clicks without adding clarity.

Measuring time and errors alongside a satisfaction score is what kept this project honest. A 4-out-of-5 rating on its own reads as a success; a 57-second task with more than one error per person on the same feature says something is wrong. Users had adapted to the navigation and then rated their adapted experience — satisfaction captured the end state, and the performance measures captured the cost of getting there.

It also changed how I think about menu depth. The redundant screens each looked defensible in isolation, and only became a problem when counted as clicks against a task a driver might perform in a parked car with limited patience.