TeslAR Vision

Overview

TeslAR Vision is an augmented reality heads-up display concept designed to help new drivers navigate UK roads with more confidence. It sets out to answer the simple question of how do you bridge the gap between the safety net of a driving instructor and the reality of driving alone for the first time?

The result is a HUD system that projects navigation, hazard warnings, road signage, and landmark information directly onto the windscreen and reducing how often a driver has to look away from the road to make a decision.

The problem

New drivers crash more, and it's rarely about recklessness its more about reaction time. Novice drivers take longer to recognize road situations and respond to them, and in the UK, over half of all road casualties in 2023 were car occupants, with the number still climbing year over year.

We looked at existing solutions which are heads-down displays, smart glasses, VR training, and we found each solved part of the problem while introducing new ones: distraction, eye strain, unrealistic simulation. A heads-up display, projected directly into the driver's natural line of sight, consistently performed better in prior research: fewer navigational errors, faster hazard detection, no need to look away from the road.

That became our starting point.

Who we designed for

We built two personas to represent the range of "new" driver: Abigail, a 19-year-old art student who's technically passed her test but panics in complex city driving, and Dan, a 32-year-old who's driven for years but just never on the left, and never in the UK. Different backgrounds, same core problem: neither trusts their instincts on unfamiliar roads yet.

Designing the system

TeslAR Vision leans on Tesla's existing sensor architecture, with eight cameras, radar, GPS, and its onboard neural network used to detect road signs, hazards, and vehicles in real time, then decide what's worth surfacing to the driver and when.

We focused the HUD on a handful of core moments:

• Navigation & road sign awareness: signs get echoed on the HUD, paired with a subtle audio cue, so drivers don't miss what they haven't learned to recognise yet
• Roundabouts: real-time lane guidance based on 3D mapping, showing exactly which exit and lane to take
• Proximity & lane-change warnings: distance-based guidance for parking and motorway driving, with haptic feedback on the wheel if a lane change looks unsafe
• Adverse weather: thermal and LIDAR-assisted detection for hazards hidden by fog, rain, or snow
• Points of interest: an optional "explore" mode that narrates landmarks as you pass them, useful for drivers new to an area

Position and legibility mattered as much as the features themselves. Prior research pointed us toward keeping alerts centered in the driver's direct line of sight, driving-related info (speed, time) on the driver's side, and adaptive brightness, the HUD needs to hold up in 8,000-nit sunny glare and a 2,500-nit night drive without becoming a distraction either way.

Testing it

We ran an open-discussion study with 11 drivers who were a mix of new UK licence holders and experienced drivers unfamiliar with UK roads, walking them through four video scenarios, then followed up with A/B/n testing across nine component variants: graphic style, colour, motion, sound, and interaction method.

A few things stood out:

• Drivers strongly preferred subtle glow effects over motion on traffic rather than animated elements which read as distracting rather than helpful
• Less is more on information density: at roundabouts, participants actively didn't want to read text, they much preferred a single clear graphic
• Blue was the clear favourite for wayfinding arrows as it didn't strain the eye and stood out from white road markings
• Voice interaction beat every other input method for anything hands-free, like saving a landmark

Those results directly shaped our final HUD design: a semi-transparent, glow-based interface with minimal on-screen text, blue directional cues, and voice as the primary way to interact with anything beyond driving.

Final prototype

What we'd change

Some honest limitations: we never tested this in an actual moving vehicle, so we don't fully know how it performs under real driving load. Road signage and laws vary by country, which limits how portable the system is as designed. And with only 11 participants, our results skew toward a small, specific sample.

If we took this further, we'd want proper AR hardware trials, expanded testing across more driver profiles, and deeper integration with vehicle-to-vehicle communication for hazard detection beyond what a single car's sensors can see.