A portal-mounted LiDAR + camera pair identifies and classifies approaching vehicles at plant gates, weighbridges and in-plant intersections — truck, wheel loader, forklift, AGV — in real time. Drives gate control, traffic-light interlocks and routing. Works at night, in fog and in dust. Custom-built for your fleet and your existing gate infrastructure.
Vehicle gate detection AI is a perception pipeline mounted at industrial plant gates, intersections and weighbridges that uses LiDAR and a calibrated camera to identify and classify approaching vehicles (truck, wheel loader, forklift, AGV) in real time and drives gate-control, weighing and in-plant routing decisions.
The pain is well-known to every plant operator. Existing gate automation relies on RFID tags, ANPR cameras and induction loops — and each one fails in a different condition. RFID assumes every vehicle carries a working tag, which falls over the moment a subcontractor, a hire truck or a maintenance vehicle turns up. ANPR cameras lose the plate in dirt, road spray, low sun or driving snow. Induction loops miss small vehicles, forklifts on rubber tyres and pedestrians altogether. The plant ends up paying a manual gate operator or living with a delayed gate, and the queue at peak shifts turns into measurable lost throughput.
Our approach fuses two sensors at the gate. A 3D-LiDAR delivers geometry-based detection that does not care about paint, lighting, dirt or weather — a truck has the silhouette of a truck whether it is clean, muddy or covered in road salt. A calibrated camera adds visual confirmation and, where the camera angle allows, fuses with existing ANPR for plate identification. A custom classifier — trained on your fleet, your wheel loaders, your forklifts, your AGVs — returns a per-vehicle class plus track ID, written directly into the gate PLC or the weighbridge controller. Pedestrians are detected as a separate safety class.
This is not an off-the-shelf SaaS. Every plant has its own gate layout, its own fleet, its own weighbridge interlock logic. We treat every engagement as a discovery + fixed-scope build, descending from our hub service Industrial Perception AI. Typically a retrofit: we add the sensor portal next to your existing gate, RFID reader, ANPR camera and weighbridge — we do not replace what already works. The code and the model are handed over to you at the end. Related capability: truck terminal LiDAR volume measurement.
Built on the same perception stack we use for our LiDAR wagon-detection and truck-terminal volume pipelines — sub-5 cm geometry, two-second response.
One LiDAR plus one calibrated camera mounted on a portal, mast or existing gate structure at the plant entrance, weighbridge approach or dispatch shed. Detection range 3–30 m, throughput beyond 300 vehicles per hour. IP65+ housings, dust-tolerant optics, runs in fog, snow, night and direct sunshine — no maintenance call when the weather turns.
The fusion classifier separates the scene into your vehicle classes — truck, wheel loader, forklift, AGV, passenger car, pedestrian — plus an "unknown" fallback. Class-aware rules let the gate open for trucks at the goods entrance, route loaders to the dispatch shed, hold the gate when a pedestrian is in the lane. Models retrained on your labelled data, not on a public benchmark.
Classification results write into the gate PLC, the weighbridge controller and the yard-management system over OPC UA, REST, MQTT or digital I/O. The gate opens, the weighbridge triggers, the event is logged with class and image evidence, the ticket hands off to ERP. Fuses with existing RFID and ANPR — we strengthen what you have, we do not replace it.
We keep the architecture boring on purpose. Three loosely coupled stages, each one independently testable, each one swappable when the sensor stack changes. Built on the same stack we use across all our perception work: PCL, Open3D, OpenCV, PyTorch.
The LiDAR delivers a 3D point cloud of the approach lane — hundreds of thousands of points per frame at portal speed. The camera adds a synchronised colour image, lens-calibrated and time-stamped. Per-vehicle UUID assigned on first contact and carried through the full track. Raw data writes to local storage for audit trail and re-labelling.
The point cloud goes through ground removal, voxel down-sampling and a PointNet-style classifier trained on your fleet. The camera image goes through a CNN trained on the same labels — and, where the geometry allows, into the existing ANPR pipeline for plate readout. The heads fuse into a per-track class plus optional plate ID. Pedestrians are detected as a dedicated safety class with a hard interlock.
Per-vehicle output: class, track ID, plate (if read), timestamp, audit image. Writes to the gate PLC or weighbridge controller over OPC UA — or to the MES, ERP, yard-management system, whatever the plant already runs. Every event is logged with image evidence; the dashboard shows live vehicle positions across the gate area.
All three stages run on an industrial PC in the gate cabinet. No cloud dependency, no external API, no licence dial-home. The code is yours at handover.
Every gate pass gets a structured record: class, track ID, plate (if read), timestamp, audit image, gate decision. Drops into ERP, into yard management, into the security audit trail. Searchable per shift, per supplier, per vehicle type — and useful for the incident report nobody wants to write from CCTV scrubbing.
The gate opens — or stays closed — based on classification plus your business rules: trucks at the goods entrance, loaders at the dispatch shed, only forklifts inside the warehouse aisle. Queue logic handles back-to-back arrivals without bouncing. The manual gate operator becomes an exception handler, not a full-time role.
Live view of which vehicles are where in the plant — gate counts, weighbridge throughput, dispatch-shed approach queue, forklift movement in the warehouse aisles. Same temporal model we use on stockpile monitoring — applied to vehicle tracks instead of material piles.
Every plant has its own fleet, its own gate layout, its own existing RFID, ANPR and weighbridge infrastructure. A generic classifier solves the generic case; your case is not the generic case.
Your fleet is not the public benchmark. Your wheel loaders, your forklifts, your AGVs, your subcontractor trucks — we calibrate the model on your vehicles under your gate lighting and dust conditions, and we re-calibrate when the fleet changes or a new subcontractor comes on site.
We add the sensor portal next to your existing gate motor, boom barrier, RFID reader, ANPR camera and induction loop. The mechanical gate stays, the existing automation stays, the cabling stays. We add a perception layer that takes the failure cases off your plate — dirty plates, missing tags, small vehicles.
Most plant gates already run a weighbridge controller, an RFID reader, an ANPR camera and a gate PLC. We fuse the classification output into what is already there — we do not replace your gate-management system. OPC UA, REST, MQTT, digital I/O and analog outputs all supported.
You own the source code, the model weights and the labelled dataset at handover. We document the system, train your team, and walk away clean. No black box, no monthly per-vehicle licence, no service contract you cannot exit. See our FAQs on IP and engagement model for the standard terms.
The engagement-model questions we hear from every plant operator considering a custom perception build for the gate. Need something more specific to your site? Ask us.
Gate layout, fleet description, existing RFID / ANPR / weighbridge setup, sample point clouds — anything you have. We come back within two business days with an honest first assessment and a fixed-price scope for the Discovery workshop.