MineSense
Real-time ore-grade sensing mounted on the excavator bucket, telling ore from waste rock on the spot
MineSense is a mining-tech company in Vancouver, Canada, doing something very concrete: mounting sensors directly on the mine's excavator buckets and conveyor belts to measure ore grade the instant the ore is scooped up, then deciding in real time whether that scoop goes to the processing plant or the waste-rock pile.
Features and application scenarios
How does a traditional open-pit mine tell ore from waste rock? It relies on a grade model drawn from drilling samples months earlier, plus manual judgment after blasting. The problem is blasting displaces and mixes rock, so the boundaries the model draws often don't match what's actually dug up — the result is throwing valuable ore away as waste (ore loss) or sending a pile of worthless rock into the plant to waste electricity (dilution). MineSense's ShovelSense puts XRF sensors into the bucket's tooth row, and every scoop's loading directly reads out the content of elements like copper and zinc, and with its own algorithms gives a sorting decision within seconds — the truck knows where to go before it even starts moving. Another product line, BeltSense, mounts on the conveyor belt to do the same real-time analysis.
This kind of tool's value isn't that "the AI is amazing" but lies in the mine's economic structure — the processing plant is the most energy-intensive stage of the whole mine, and sending one fewer truck of waste rock into it is real savings in electricity and grinding media. For Taiwanese readers, MineSense is more like a textbook for understanding "how industrial AI makes money": it replaces no one, just swaps a decision that used to be guessed by experience with a number measured on the spot. Taiwan has no large metal mines, but the same logic fully applies to raw-material sorting in cement, glass, and scrap-metal recycling.
Main features
- ShovelSense: XRF sensors mounted directly on the electric/hydraulic excavator bucket, measuring ore grade per scoop
- BeltSense: real-time elemental analysis on the conveyor belt, for feed-end grade monitoring
- Real-time ore/waste sorting decision, producing a route instruction before truck loading is complete
- Integration with the mine's existing fleet-management system (FMS), pushing decisions directly to drivers
- Accumulated per-scoop measurement data can be used to recalibrate the ore-body grade model
- Mainly applied to base-metal open-pit mines like copper and zinc
Common uses
- Real-time ore/waste sorting at open-pit copper mines to reduce waste-rock feed into the plant
- Recovering marginal-grade ore that would have been misjudged as waste
- Recalibrating the existing ore-body grade model with measured data
- Conveyor-belt feed-grade monitoring and anomaly alerts
Key Features
- ShovelSense: XRF sensors mounted directly on the electric/hydraulic excavator bucket, measuring ore grade per scoop
- BeltSense: real-time elemental analysis on the conveyor belt, for feed-end grade monitoring
- Real-time ore/waste sorting decision, producing a route instruction before truck loading is complete
- Integration with the mine's existing fleet-management system (FMS), pushing decisions directly to drivers
- Accumulated per-scoop measurement data can be used to recalibrate the ore-body grade model
- Mainly applied to base-metal open-pit mines like copper and zinc
Pros
- Measurement happens at the 'decision point' rather than in after-the-fact assays — a rare real-time quality among such solutions
- The value chain is very clear: less waste rock into the plant = direct savings in grinding energy and cost
- The sensor hardware is designed to withstand the extreme impact of a mine bucket, not lab-grade equipment
- The data produced can in turn improve ore-model precision, getting more accurate with use
Cons
- Applicability limited to specific metal-ore types, not a general-purpose tool
- Requires installing hardware on heavy machinery, so adoption involves downtime and mine-engineering coordination
- Pricing not public; it's an enterprise-grade project needing sales negotiation
- Taiwan has almost no applicable scenario, so it's a knowledge reference rather than a usable tool for general readers
Use Cases
- Real-time ore/waste sorting at open-pit copper mines to reduce waste-rock feed into the plant
- Recovering marginal-grade ore that would have been misjudged as waste
- Recalibrating the existing ore-body grade model with measured data
- Conveyor-belt feed-grade monitoring and anomaly alerts
Editor's Note
Editor's note: This is the kind of tool that makes you think 'so this is where AI makes money.' It's completely unsexy, with no chat interface, but the business logic is rock-solid — the electricity to grind a ton of waste rock at the plant is real money, and grinding one less ton saves one ton. I especially like their decision to mount the sensor on the bucket teeth: that's the most brutal spot in the mine, and their willingness to put a precision instrument there shows they truly understand what the customer's work site looks like.
FAQ
Is MineSense software or hardware?
Both, and the hardware is essential. ShovelSense is an XRF sensing module actually mounted on the excavator bucket, and the software turns the sensor signal into grade interpretation and sorting decisions. You can't just buy the software.
Can Taiwan use it?
In practice it's hard. Taiwan has no large metal open-pit mines, and MineSense's target customers are large copper-zinc mines in South America, North America, and Africa. Taiwanese readers are better off treating it as a case of 'industrial sensing + AI decisions.'
How is it different from ordinary ore-model software?
An ore model is an 'advance prediction,' estimating where ore is underground based on drillhole data from months ago; MineSense is 'on-the-spot measurement,' measuring directly the moment it's scooped. The two are complementary, not a replacement relationship.
Related AI Tools
MOVUS
Australia's AI machine health monitoring that goes beyond alerts to tell you what to fix first
Joulea
Combining drone thermal imaging and AI to pinpoint building envelope heat loss.
Enersee
AI-powered energy management platform from Belgium designed to detect hidden power anomalies across multi-location buildings.
Nantum AI
Commercial Real Estate AI that transforms building data into actionable energy and ESG insights.
PassiveLogic
Autonomous building control platform that runs buildings using physics-based world models
Luxonis
OAK series edge AI depth cameras that run computer vision directly on the device