The current automation level must stay continuously visible

What it is

For as long as the vehicle is using any layer of driving automation, one glance must be enough to read who is driving now: the person, the system, or each taking a slice. That is not the one-frame announcement when a takeover completes. It is mode awareness across the whole trip. SAE J3016 levels describe function allocation, not the name on the badge; what people need is current authority, not the marketing label. After a menu, a navigation glance, or a conversation, the encoding still has to be there. It cannot flash only at engagement and then withdraw.

Why it happens

On a long trip nobody stares at a status lamp. Supervision is intermittent: most of the time the person looks at the road or at something else, and occasionally samples “still the same mode.” If status appears only at transitions, the sample often lands on a blank or on the previous task, and memory fills in—“I turned it on earlier.” Level 2 looks as if the system is driving while the rule still requires the person in the loop; that gap is exactly what a persistent indicator is for. Colors, light bars, and wheel halos that share a language with cabin ambience get read as mood lighting, not authority. Continuous visibility is therefore not decorative always-on. It is a stable sampling point for intermittent supervision.

Studying it

In a simulator, let automation run for a long stretch, insert a non-driving task, then—without a takeover—ask who currently controls steering and speed.

Independent variables: whether the indicator is persistent, whether it appears only at transitions, whether encoding is shared with ambient lighting or theme color, locus (cluster, light bar, center stack). Dependent variables: match between spoken report and true authority, time to notice a state change, false hands-off or false takeovers.

Ask too soon and people still remember the transition animation; too late and they have already guessed from the road. Hit the window where the indicator is supposed to still be there and the person has just returned from another task. Gaze only shows the region was looked at, not that it was read as authority; pair it with a spoken report.

Where it stops holding

Vehicles that never offer driving automation do not owe this indicator. Parking-camera and charging screens may yield the driving-status slot, but it must return as soon as the vehicle can move. A cabin with no driver seat in a driverless vehicle needs externally readable behavior, not a level bar for a driver. A robot’s idle / running / fault marks that are readable at a distance tell bystanders whether it is safe to approach; they do not replace the authority cue behind a steering wheel.

Applying it

• Put an authority indicator near the cluster or wheel that does not vanish when the theme changes; keep the glance target fixed.
• If lateral and longitudinal can automate separately, split the indicator; one lamp must not mean both “all on” and “only one on.”
• Keep the same encoding on uneventful stretches, not only at cut-in and cut-out; the transition animation cannot be the only source of state.
• Verify on a run that includes entertainment and no takeover: cover the displays mid-trip and ask who is driving. A wrong answer means the indicator never became a samplable object.

Related

• Within the group: K6.09.2 State changes need explicit confirmation · K6.09.3 Mode confusion is the highest-risk failure in this domain
• Adjacent: K6.08 Automated Driving Takeover Requests · K6.12 Automation State Expression and Mode Confusion · X3.01 Visible Current State
• Search terms: mode awareness · automation level display · SAE J3016 · who is driving