Case Study | 3D Vision Solves Zinc Ingot Depalletizing and Stacking Challenges

In the front-end processes of battery manufacturing and non-ferrous metal smelting, zinc ingot depalletizing and feeding are core steps to ensure continuous production line operation. Traditional manual depalletizing not only involves high labor intensity and elevated working temperatures but also faces significant bottlenecks in long-term efficiency and stability due to the coexistence of heavy zinc ingots and high-frequency equipment.

Recently, Transfer Tech implemented a 3D vision-based intelligent zinc ingot depalletizing system for a major domestic battery manufacturer. Targeting the special conditions of the furnace feeding station, the system overcomes industry pain points such as high-frequency electromagnetic interference and irregular material stacking, achieving fully automatic zinc ingot depalletizing and precise feeding.

1. Core Challenges

The station interfaces with the furnace feeding process. A forklift delivers a full stack of zinc ingots beneath the camera's field of view, and the 3D vision system guides a gantry robot to pick each piece and place it onto a transition table. The on-site conditions are complex, with core difficulties concentrated in three dimensions:

• Unstable material state: Two specifications of zinc ingots are involved, and incoming materials often exhibit irregular stacking such as overlapping, side overlapping, and tilting, making fixed-position programming infeasible.

• Strong electromagnetic interference environment: The station is adjacent to high-frequency furnace equipment, and the complex electromagnetic environment can interfere with vision communication and imaging, posing a key technical barrier for vision system deployment.

• Dual requirements for precision and cycle time: The solution uses a gantry robot with vacuum grippers, requiring stable output of precise position and orientation to ensure successful grasping, while also matching the production rhythm of the furnace feeding.

2. Solution

Transfer Tech adopted an integrated solution combining a 3D smart camera and vision software, with specific adaptations and optimizations for the on-site conditions.

Intelligent Algorithm Adapts to Complex Stack Patterns

The vision system can identify the front and back sides of zinc ingots and output precise position and orientation information for each piece. For irregular stacks such as overlapping and side overlapping, the system employs a top-priority grasping strategy, automatically planning the optimal grasping sequence to guide the robot to pick the top ingots first, logically avoiding collisions and grasping failures.

When the entire stack is fully depalletized, the system automatically outputs a completion signal, coordinating the front-end forklift replenishment and back-end feeding processes, achieving full automation of the station workflow.

Native Support for Dual Material Specifications

The solution natively supports the two sizes of zinc ingots on site, enabling recognition switching without hardware modification. It also reserves expansion capability: when new zinc ingot models are added, they can be quickly adapted through sample training, reducing the cost of production line upgrades.

Anti-Interference Design Ensures Communication Stability

To address electromagnetic interference from the high-frequency furnace, the solution includes specific optimizations in hardware wiring and signal transmission, using shielding protection and dedicated communication links to ensure complete and lossless visual data transmission, mitigating operational risks from the electromagnetic environment at the foundational level.

3. Technical Highlights

After deployment, on-site operation verification confirmed that all performance indicators meet the requirements:

• Recognition accuracy: The vision system's own recognition error is ±1 mm, and the overall guided grasping accuracy is ±4 mm, fully meeting the alignment requirements for vacuum gripper grasping.

• Cycle time: The cycle time for imaging, recognition, and data output for a single zinc ingot is 4 seconds, stably matching the production rhythm of the furnace feeding.

• Grasping success rate: The vision recognition success rate reaches 99.9%, reliably handling complex stacking states such as tilting and overlapping.

This intelligent depalletizing system frees operators from the high-temperature, high-load feeding station, eliminates cycle time fluctuations caused by manual operation, and ensures continuous and stable operation of the smelting process.

More importantly, this project validates the feasibility of deploying 3D vision in complex scenarios such as irregular material stacking. For industries with similar pain points, such as metallurgy and battery manufacturing, this field-verified solution can be quickly replicated and adapted, providing a reliable visual perception solution for heavy material depalletizing scenarios.