Crusher Foreign Material Detection AI

Bring real-time intelligence to every tonne that reaches the crusher with Roboflow Vision AI for foreign material detection. Built for mines, quarries, and aggregate plants where one shovel tooth or length of drill steel in the feed can crack a mantle, jam a jaw, rip a belt, or take the primary down for a shift.

Catch tramp metal and non-metallic foreign material on the feed:

• Detect bucket and shovel teeth, adapters, drill steel, rock bolts, mesh, grinding media, and liner bolts on the feed conveyor or in the dump pocket with a model trained on your own footage, including wet ore, dust, and low light.
• Flag timber, rubber, conveyor belt scraps, and plastic that a metal detector does not see, so they are pulled before they blind a screen or jam a chute.
• Separate foreign material from ordinary ore and rock by shape and texture, so alerts fire on a tooth and not on every dark rock on the belt.

Find oversize and bridging before the crusher chokes:

• Detect oversize slabs and boulders in the dump pocket or on the apron feeder and alert the rock breaker operator before the jaw bridges.
• Watch the crusher throat and chute for blockage and build-up so the control room sees a choke forming, not a stopped crusher.
• Extend the same models to secondary and tertiary crusher feeds, where a missed tooth from the primary does the most damage to cones.

Trigger the right response and keep the record:

• Send a stop, divert, or magnet trigger to the belt PLC so the conveyor halts or the load is rejected before the item reaches the crusher.
• Run alongside metal detectors and tramp magnets as a supplementary layer that covers non-metallic material and the angles those devices do not see.
• Log every event with a clip, belt, timestamp, and the haul truck or shovel that loaded it so maintenance can trace which bucket is losing teeth.

Bring intelligence to every crusher feed today. Stop tramp metal and foreign material from becoming cracked mantles, ripped belts, and lost production.

‍

Frequently asked questions

What is crusher foreign material detection with Vision AI?

Crusher foreign material detection with Vision AI uses cameras over the dump pocket, apron feeder, and feed conveyor and a trained computer vision model (opens in a new tab) to find material that should not go through the crusher: bucket and shovel teeth, drill steel, rock bolts, liner bolts, timber, rubber, and oversize rock. When the model finds an item, it can stop the belt, trigger a divert or magnet, or alert the operator, and it logs the event with a clip. It is the mining counterpart to foreign object detection for manufacturing, and it pairs with rock fragmentation analysis and conveyor belt monitoring on the same cameras.

Can Vision AI find a shovel tooth on a fast, dusty conveyor?

A missing ground engaging tool is one of the costliest events on a processing plant, and it hides in a stream of ore that looks similar at speed. A detection model trained on your own belt footage learns tooth, adapter, and drill steel shapes against your ore, and a camera with a fast shutter and good lighting over the belt keeps motion blur low enough to see them.

Does crusher foreign material detection support MSHA and ISO 45001 programs?

Crusher foreign material detection supports the conveyor and crusher safety expectations of MSHA 30 CFR Part 56 and 57 (the US safety and health standards for surface and underground metal and nonmetal mines) and ISO 45001 (the occupational health and safety management system standard) by reducing manual belt interventions and blocked-crusher clearing, and by keeping a record of every event.

Roboflow serves as the detection engine, and your plant, maintenance, and safety teams own the stop and divert logic, the lockout procedure for removing tramp material, and the sign-off on how the belt control responds.

Can it integrate with our belt controls, SCADA, and maintenance system?

Yes. Roboflow writes detection events to a PLC tag over OPC UA, Modbus TCP, or MQTT, so Allen-Bradley, Siemens, and ABB controllers can stop the conveyor, fire a divert gate, or pulse a tramp magnet. Events and clips can be sent over REST (opens in a new tab) or written to a database for SCADA and historians such as Ignition, AVEVA PI, and Citect, and to maintenance systems such as SAP PM, IBM Maximo, and Pronto, so a missing tooth becomes a work order on the right shovel.

‍