In steel mills where temperatures soar and heavy coils weighing tens of tons move constantly, manual crane operation exposes workers to extreme risk and limits productivity—operators need at least four hours of rest per shift, and a single malfunction can trigger millions in losses. To eliminate these hazards and unlock continuous throughput, an AI‑enabled unmanned EOT crane system was deployed using ruggedized industrial computers, 3D vision, and predictive maintenance algorithms. Since 2018 the solution has completed over 60,000 trips, lifted more than 300,000 coils, and now runs 24/7 across multiple facilities. The result is a safer, more efficient operation that scales beyond the pilot plant and keeps production lines running around the clock.
Case Study Source: NEXCOM International Co., Ltd.
Problem Statement
Manual operation of EOT cranes in a high‑temperature, hazardous steel mill limited throughput, required operator rest, and exposed staff to risk. Any crane failure could halt the production line and trigger heavy financial losses, while frequent on‑site checks added further safety concerns.
Goal
Create an unmanned, AI‑enabled EOT crane system that runs continuously, positions coils precisely, integrates with warehouse logistics, and reduces downtime through predictive maintenance.
Challenges
Manual crane driving in a hot, risky environment with operators needing at least 4 hours of rest per shift.
A single crane malfunction could stop the line and cost up to millions of dollars in losses.
Harsh conditions (high temperature, shocks, vibration) made regular on‑site measurements hazardous.
Steel coils are extremely heavy, weighing tens of tons each, demanding precise positioning and safe handling.
Reliable data capture and control were required despite electrical noise and surges in an industrial setting.
Actions
Equipped unmanned EOT cranes with a rugged, fanless industrial computer (NISE 3910E) rated for wide temperature, shock and vibration, and isolated COM ports to withstand surges.
Integrated high‑resolution PoE cameras and a 3D scanner; combined 2D images with reconstructed depth to generate accurate storage coordinates.
Added GPU acceleration via PCIe to boost AI workloads, including image processing, deep learning, and object recognition for machine vision.
Adopted an OT/IoT software stack (supporting OPC UA and Docker) to deploy and manage microservices seamlessly from cloud to edge.
Implemented AI scheduling, optimal lift path planning, and queue optimisation with accurate time predictions to align driver arrivals with lifting tasks.
Developed a smart lifting clamp that identifies coil ID, locates the coil’s centre, and uses deep learning to detect personnel and obstacles with active safety protection.
Key Results
Impact
Improved worker safety by removing operators from high‑temperature crane cabs and cutting risky on‑site inspections.
Greater operational efficiency through overnight re‑arrangement of coils and integration with driver check‑in for faster dispatch.
Higher resilience and fewer production stoppages thanks to health‑based maintenance and real‑time anomaly alerts.
The Challenge
A steel mill faced a productivity bottleneck. Human operators had to manually control overhead cranes in sweltering, dangerous conditions. The intense heat forced each driver to take at least 4 hours of mandatory rest every shift, severely limiting output.
The stakes were extraordinarily high. If a crane broke down, the entire production line would grind to a halt. A single failure could trigger losses running into millions of dollars. Meanwhile, routine equipment checks meant sending workers into hazardous zones filled with extreme temperatures, vibration and physical shocks.
The steel coils themselves presented another layer of complexity. Each one weighs tens of tonnes, so even minor positioning errors could prove catastrophic. To make matters worse, the mill’s electrical environment—plagued by surges and industrial noise—made reliable sensor data difficult to capture.
The Solution
The answer was to remove people from harm’s way entirely. Engineers designed an autonomous crane system powered by artificial intelligence and computer vision.
At its core sat a fanless industrial PC (the NISE 3910E) built to survive temperature extremes, vibration and electrical interference. High-definition cameras and a 3D scanner worked together to map the warehouse in three dimensions, pinpointing exactly where each coil should go.
A graphics processor handled the heavy lifting on the computational side, running deep-learning models that recognised objects and planned optimal routes. The software architecture used modern cloud-to-edge tools—OPC UA and containerised microservices—to keep everything running smoothly across the network.
Perhaps the cleverest element was the intelligent lifting clamp. It could read each coil’s ID tag, calculate the centre of mass, and scan continuously for people or obstacles. If anyone strayed too close, the system would halt immediately.
The AI also tackled logistics. It scheduled lifts, calculated queue times, and coordinated with truck drivers so they’d arrive precisely when their coil was ready—no wasted journeys or idle waiting.
What It Delivered
Impressive Scale
Since going live in 2018, the system has handled more than 60,000 crane movements and shifted over 300,000 steel coils. A second crane later joined the operation, fully automating a warehouse storing roughly 20,000 metric tonnes.
Round-the-Clock Operation
The cranes now work 24 hours a day. Overnight, they quietly reorganise inventory and prepare loads for the morning shift—tasks that would have been impossible with human crews.
Smarter Maintenance
Sensors continuously monitor motors, gearboxes and brakes. Machine-learning algorithms spot early warning signs of wear or failure, scheduling repairs before breakdowns occur. This predictive approach has slashed both downtime and repair bills.
Proven Scalability
Word spread quickly. By 2019, the company had sold 12 systems to other steel plants. When the pandemic hit, remote commissioning meant installations could continue without site visits.
The Real-World Impact
Worker safety improved dramatically. Operators no longer endure scorching crane cabs or venture into dangerous zones for routine checks.
Efficiency soared. Coils are rearranged during the night, and lorry drivers check in digitally so their loads are waiting when they arrive. Production stoppages became rare, thanks to the health-monitoring system catching faults early.
It’s a compelling example of how AI and industrial automation can solve genuinely difficult problems—not by replacing skilled workers out of convenience, but by removing them from genuine danger whilst lifting performance to levels manual operation could never achieve.
Case Study Source: NEXCOM International Co., Ltd.
These industry AI case studies featured on our site are based on publicly available sources and are presented for informational and educational purposes only; we do not claim ownership of these case studies or affiliation with the companies mentioned, and attribution is provided where applicable.
