Three things to take away
- Confirm that the event and footage match before discussing behaviour.
- Consider distance driven and task context rather than ranking raw alert counts.
- Keep the driver explanation, agreed action and follow-up together.
The situation: plenty of alerts, inconsistent conversations
A carrier operating highway and urban feeder routes receives close-following and distraction events. Different supervisors use different review methods, from a single screenshot to an alert-count ranking. Drivers receive mixed messages about what should change.
The pilot begins with one event type and a defined vehicle group. Record the event source, camera placement, access permissions, vehicle, task, time and clip reference. Do not present unvalidated distance or speed estimates as measured facts.
Detection output
Truck ahead detected
Labels and values illustrate the capability; review full footage before a finding.
Review: reconstruct the road context first
The reviewer checks the footage before and after the event, time alignment and available speed data. Sustained following, another vehicle cutting in and slow queues are different situations. An obstructed camera or missing clip should result in an insufficient-evidence status.
Add the driver explanation before recording a confirmed, excluded or unresolved finding. Detection is a screening signal that supports a review; it is not a substitute for complete evidence or an automatic disciplinary decision.
An evidence chain from alert to coaching
| Operating step | Action | Evidence to review |
|---|---|---|
| Triage | Check the device, task and clip time | Original event reference and playable footage |
| Human review | Consider road context and driver input | Reason for confirmed, excluded or uncertain finding |
| Coaching | Agree one practical behaviour | Discussion, action and follow-up date |
| Follow-up | Review recurrence with driving exposure | Comparable sample and unresolved actions |
Coaching: agree an observable action
For a confirmed coaching opportunity, a supervisor and driver review the clip and discuss one practical action, such as anticipating traffic, preserving space or avoiding non-driving tasks. Record the agreed behaviour and a follow-up date.
Do not design feedback that requires a moving driver to inspect a screen or join a long call. Apply the existing urgent safety process when needed; hold routine coaching at an appropriate stopped time and limit video access and retention to the operational need.
Follow-up: keep context in the comparison
Review confirmed-event proportions, repeated event types, completed coaching and valid distance driven together. Different routes, duties and camera conditions make raw counts a weak comparison.
The pilot should establish whether reviews are consistent and drivers understand the agreed actions. Without sufficient evidence, do not claim a reduction in crashes. Keep a review route available for disputed clips.
Questions, answered
Does a close-following alert prove a violation?
No. Review the road context, available footage, device conditions and relevant operating rules before reaching a finding.
Should drivers be ranked by total alerts?
Raw totals are not a sound comparison. Validate events and consider distance, duty type and road conditions first.
