Forklift and pedestrian safety with AI video: a practical guide

What forklift and pedestrian safety covers

Forklift and pedestrian safety is the part of workplace safety that deals with lift trucks and people on foot working in the same space — in warehouses, distribution centers, production halls and yards. A forklift is heavy, can carry a load that limits the driver's view, and can turn and reverse in a small space. A person on foot is focused on their own task: picking, counting, carrying a part to a machine. The risk sits where their paths cross. It shows up in six recurring situations.

Blind corners and aisle intersections

Where a rack aisle meets a main aisle, high racking hides both sides from each other. A forklift turning out of the aisle and a picker stepping out from between the racks meet in the same second, with neither able to see the other until the last moment. Mirrors and floor markings help, but only when people slow down and look.

Docks and loading areas

At the dock, forklifts drive in and out of trailers, reverse with loads and turn in tight spaces, while truck drivers, checkers and dock staff walk the same strip of floor. Visiting drivers may not know the site's walkways at all. The dock edge and the space just behind a reversing forklift are where people on foot are hardest to see.

Pedestrians in vehicle lanes

Marked walkways and crossings only work when they lead where people need to go. When the shortest way to the break room, the toilets or a workstation runs across a vehicle lane, people take it. Each crossing outside the marked route puts a person on foot where drivers are not expecting one.

Speeding and dangerous maneuvers

Speed limits for lift trucks are set per site, and so are the rules for how to drive them: slow down at intersections, travel with the forks low, reverse when the load blocks the view, don't turn with a raised load. A forklift that takes a corner without slowing, reverses into a walkway without looking, or swings its load across a crossing turns into a near miss — or an accident — when a pedestrian steps out.

Near misses that go unreported

A near miss is the event where nobody was hurt but someone could have been: the pedestrian stepped back in time, the driver braked hard. Because no one was injured and nothing was damaged, it is easy to shrug off and never write down. Yet the near miss is the clearest early signal of where the next accident will happen.

Patterns by shift and zone

Vehicle–pedestrian risk is not spread evenly. It concentrates at particular intersections, dock doors and crossings, and at particular times: shift changes, peak dispatch windows, the end of a shift when everyone is in a hurry. A single near miss says little; the same kind of event at the same place and time, week after week, points at a cause in the layout or the process, not at one careless person.

Why traditional methods miss forklift and pedestrian risk

A site can have rules, markings, training and a safety team and still miss these moments. The gap is visibility: the risky moments last a few seconds, happen across a large floor and around the clock, and a near miss by definition ends without harm.

• Periodic audits see a moment, not the floor. A safety walk or inspection observes one area for a short time, and people behave differently when they know they are being watched. The near miss at the dock door on the night shift is not on the audit route.
• Near misses are not reported. Reporting takes time, nobody was hurt, and people may worry about blame — for themselves or a colleague. Good safety management treats these events as warnings to investigate, not luck to forget. That investigation can only start from events someone records.
• Manual CCTV review doesn't scale. The cameras may already be in place and recording. But checking them means knowing which camera and which minute to watch, so footage is reviewed after an accident — not to find the near misses that came before it.
• Rule-based detectors fire without context. A fixed zone or a line-crossing rule triggers when a person enters an area or a box gets too close to another box. It can't tell a forklift parked with the engine off from one turning into an aisle, or a picker working beside a stationary truck from one stepping into its path. The result is a stream of alarms for normal work and silence for the dangerous situation nobody drew a zone for — and a team that learns to ignore the alarms.

What these methods lack is the reading a safety professional makes when they see the scene: where the forklift is going, where the person is heading, and whether the two are about to meet.

How vision-language models make it solvable

Primarch builds on and enhances vision-language models (VLMs): models that read video and language together. That changes what can be checked on a busy floor. Instead of drawing boxes around objects frame by frame, the model reads the scene the way an experienced safety professional would — and does it on every camera, on every shift.

• It reads movement over time, not a single frame. A VLM reads a window of time, so the before and after of a movement are read together. That is what makes direction visible: a forklift leaving an aisle and a pedestrian walking toward the same intersection, or a truck reversing toward a person behind it.
• It reads the forklift's route and the pedestrian's direction together. The question is not whether a person and a forklift are in the same picture, but whether their paths are converging. That is how a near miss and a dangerous maneuver can be recognized before the intersection happens, rather than counted afterwards.
• It uses context instead of fixed rules. Context — a parked truck versus a moving one, a person working beside a forklift versus stepping into its path — is what separates normal work from risk. Reading the scene in context is designed to remove the biggest source of false alarms in rule-based systems: rules firing without context.
• It writes the finding in plain language, with a clip. The result is a sentence a supervisor can read — what happened, where, and why it matters — with a time-stamped evidence clip, so everyone discussing it sees the same thing.
• It chains events by shift, zone and time. Single events are recorded and alerted for review; chained together, they show where and when the same situation repeats, and help diagnose why: an attention lapse, or a process gap.

The useful question is not "was a person within a few metres of a forklift?" It is "were their paths about to meet, and how did the forklift move as they did?" The answer is in the seconds of video before the intersection.

Coverage depends on the cameras. A blind corner can be read only where a camera sees both approaches to it; a dock door, only where a camera covers the dock floor.

What to look for in a solution

Whichever vendor you talk to, these are fair questions to ask about a system meant to improve forklift and pedestrian safety with video:

• Works on your existing cameras. Can it connect to the IP cameras you already have, or does every intersection need new hardware?
• Reads interaction, not just proximity. Does it consider where the forklift is going and where the person is heading, or does it only measure distance or zone entry?
• Handles false alarms openly. Ask how it separates normal work beside a forklift from a real near miss, how findings are reviewed by a person, and how single events are separated from repeating patterns.
• Turns near misses into records. Each event should come with a time, a place, a plain description and an evidence clip, so it can be investigated and discussed like a reported one.
• Explains repeats. Can events be grouped by shift, zone and time, and does it help separate an attention lapse from a process or layout gap?
• Produces reports your safety team can use. Findings written in regulatory language, corrective actions with owners, and zone-level summaries for the safety committee and audits.
• Respects privacy. Is footage anonymized? Is there an on-premise option where footage never leaves the site? Does it meet GDPR and other data-protection requirements, and does it report situations rather than individuals?
• Can be piloted on your own floor. One zone, a few of your cameras, your own traffic — not a demo video.
• Gets alerts to the right role. An instant alert for the shift supervisor, a digest for the safety team, a dashboard for site management.

Example use cases

The walk-throughs below are illustrative examples of how a finding is built. They are not customer cases and contain no performance figures.

Example 1: a picker steps out at a blind aisle corner

1. 01

   Moment

   A distribution center, 10:25, where rack aisle 7 meets the main aisle. A camera covers both approaches to the intersection.

2. 02

   What the camera shows

   A forklift carrying a pallet travels down aisle 7 toward the main aisle. In the main aisle, a picker with a handheld scanner walks toward the same corner, looking at the screen. The racking hides each from the other. The forklift brakes hard at the corner; the picker steps back.

3. 03

   The finding + evidence clip

   "Near miss at the aisle 7 / main aisle intersection: the forklift's route and the picker's direction converged at the blind corner; the forklift stopped short and the picker stepped back." The clip covers the seconds before and after.

4. 04

   Who is alerted

   The shift supervisor receives an instant notification; the event goes into the safety team's weekly digest.

5. 05

   What the shift and zone pattern shows

   Similar near misses at this corner fall in the morning picking wave, when both forklift and picker traffic in the main aisle are at their highest.

6. 06

   Corrective action

   A recommendation written for the logistics manager: separate picking-wave pedestrian traffic from the forklift route at this intersection, and review the corner's visibility aids.

Example 2: a forklift reverses out of a trailer toward a truck driver

1. 01

   Moment

   Dock door 4 at a logistics warehouse, 15:40, during afternoon dispatch. A camera covers the dock floor in front of the doors.

2. 02

   What the camera shows

   A forklift unloads a trailer and reverses out onto the dock with a pallet. Behind it, a visiting truck driver walks along the dock edge toward the office with paperwork. The forklift continues reversing; the driver steps aside at the last moment.

3. 03

   The finding + evidence clip

   "Dangerous maneuver and near miss at dock door 4: the forklift reversed out of the trailer toward a pedestrian on the dock floor behind it." Clip attached.

4. 04

   Who is alerted

   The dock supervisor, instantly; the safety team in the weekly digest.

5. 05

   What the shift and zone pattern shows

   Flagged events at doors 3 and 4 fall in the afternoon dispatch window, and the pedestrians involved are walking between the dock and the dispatch office.

6. 06

   Corrective action

   Diagnosis: not an attention lapse but a process gap — visiting drivers' route to the office crosses the working dock. Recommendation: a marked waiting area and a separate walkway to the office for drivers, assigned to the dispatch manager.

Example 3: a shortcut across the vehicle lane at break time

1. 01

   Moment

   A production hall where forklifts bring raw material from the warehouse to the lines, 12:02.

2. 02

   What the camera shows

   Operators leaving the line for lunch cross the vehicle lane midway rather than at the marked crossing further down. A forklift with a load comes down the lane; one operator crosses in front of it and the forklift slows sharply.

3. 03

   The finding + evidence clip

   "Near miss in the vehicle lane: a pedestrian crossed ahead of a loaded forklift outside the marked crossing; the forklift slowed sharply." Clip attached.

4. 04

   Who is alerted

   The production shift supervisor, instantly.

5. 05

   What the shift and zone pattern shows

   Flagged events at this point of the lane cluster around break times on every shift.

6. 06

   Corrective action

   Diagnosis: a process gap, not individual carelessness — the marked crossing does not lie on the route to the canteen. Recommendation: move or add a crossing on the walking route, and schedule material runs outside break times where possible.

Example 4: shift change at the main gate

1. 01

   Moment

   The gate between the yard and the warehouse, 23:55, at the change from evening to night shift.

2. 02

   What the camera shows

   People arriving and leaving walk through the vehicle gate while a forklift moves loaded pallets from the yard into the warehouse. The forklift turns through the gate as a group of people walks in.

3. 03

   The finding + evidence clip

   "Near miss at the main gate: a forklift turned through the vehicle gate into the path of pedestrians arriving for the night shift." Clip attached.

4. 04

   Who is alerted

   The night-shift supervisor, instantly; the event also appears in the gate zone's weekly score.

5. 05

   What the shift and zone pattern shows

   Asked in plain language — "show near misses at the main gate in the last month" — the event memory returns a short list, all within a few minutes of a shift change.

6. 06

   Corrective action

   A recommendation for the site manager, with the regulatory reference: keep forklift movements out of the gate during shift changes and give pedestrians their own route in.

How Primarch's OHS Expert helps

This section is about our product. The OHS Expert is the Primarch expert module that reads the floor like a safety professional, on the cameras a site already has. For forklift and pedestrian safety it:

• Reads the forklift's route and the pedestrian's direction together, and recognizes the near miss and the dangerous maneuver before the intersection happens.
• Narrates each finding in natural language, with a time-stamped evidence clip, and recalls past repeats of the same scene from memory.
• Chains events by shift, zone and time and diagnoses why they repeat — an attention lapse or a process gap.
• Writes corrective actions in the language an inspector would use, assigns owners, and prepares regulation-referenced records, weekly zone safety scores and executive summaries.
• Needs no rule set. It comes pre-configured with OHS domain knowledge and regulation; site-specific thresholds are tuned together at onboarding, and situations nobody wrote a rule for are still caught through scene understanding.
• Flags speed violations, such as a forklift taking a blind corner without slowing.
• Recognizes pedestrians walking in a vehicle lane instead of the marked walkway, even when no forklift is nearby.

Because it reads the scene in context, it is designed to remove the biggest source of noise in classic systems — rules firing without context — and multi-stage validation and temporal noise suppression are added on top.

Putting it to work

1. 01

   Connect

   Connect your existing RTSP cameras; no new cameras or special hardware.

2. 02

   Pilot on your own floor

   Pilots typically start with a single zone and a few cameras — a busy intersection or a dock, say. Thresholds are tuned to your site together, and you review the findings with your safety team.

3. 03

   Roll out

   Scale facility-wide as the system proves out, with dozens of streams watched in parallel.

• Alerts, digests and dashboards. Instant notifications, periodic digests and role-based dashboards, so each finding reaches the person who acts on it.
• Ask the past in plain language. Findings are written into a queryable event memory: "Which near misses were recorded at dock door 4 last week?"
• Records you can defend in an audit. Every finding carries its time-stamped clip; reports carry the regulatory reference and the corrective action with its owner.
• Privacy by design. Footage is anonymized in a GDPR-compliant way, and the system is designed to report risky situations and process gaps, not individuals. With on-premise deployment, footage never leaves the facility.