Automatic part unloading: which operations it removes and which remain ART-148

Three different levels that are often given the same name

### 1. Automatic loading

The system takes a sheet from storage or a stack and loads it into the machine. It reduces manual movement of raw material but does not necessarily handle cut parts.

### 2. Automatic sheet unloading

The system removes a processed sheet, often together with its skeleton and parts, and moves it to a separation position or buffer. Bystronic describes ByTrans Modular as loading and unloading automation; Salvagnini states that ADLU automates sheet feeding from a stack and the unloading of cut sheets. This reduces intermediate handling, but parts may remain in the sheet.

### 3. Automatic part removal and sorting

A picking system recognizes parts or receives their positions from the program, grips them and places them at defined destinations. BySort is offered for automatic removal and sorting of finished parts; AMADA TK II is intended for collection, sorting and stacking. Salvagnini MCU is described as an automatic sorting device for parts of different shapes, sizes and weights within a particular configuration.

These levels have different costs, footprints, integration requirements and remaining manual work. If the need is only to release the pallet quickly, a part picker may be excessive. If the main loss is manual separation of five orders, a simple unloader will not solve the task.

The “removes — reduces — leaves” matrix

| Operation | Sheet unloading | Part picking and sorting | What to verify | |---|---|---|---| | loading a new sheet | only with a loading module | depends on the system | format, weight, material, storage | | removing the processed sheet | generally yes | yes or integrated | cycle time, buffer space | | releasing the pallet | generally yes | yes | synchronization with the machine | | separating individual parts | no or partly | for suitable geometries | minimum and maximum size, weight, remaining attachment | | stacking parts | no | yes, within defined zones and rules | height, stability, surface | | sorting by order | no | possible | number of zones, program data | | removing the skeleton | yes or separately | often a separate flow | format and handling | | small scrap | not always | not always | risk of falling or being gripped | | inspecting edge and geometry | no | usually no | separate quality gate | | confirming the complete set | no | partly through data | actual quantity and exceptions | | transport to bending | no | only with additional integration | containers, automated vehicle or conveyor, route |

This table is a functional framework, not a specification for a particular model. The supplier should complete it for every offer using documentation and test results.

How to compare an unloader and part picker across a part portfolio

One demonstration nest does not prove that a system suits the product range. Add an application matrix to the technical request and assess each part class separately.

| Class | Success criterion for a sheet unloader | What a part picker must additionally prove | What remains manual or separate | |---|---|---|---| | large flat parts | sheet removed, pallet released | part safely separated and placed in a defined stack | surface inspection and stack transfer | | many small parts | cut sheet moved to a buffer | minimum geometry, gripping and correct destination confirmed | unsupported items and small scrap | | long narrow parts | no damage while the sheet is transferred | part remains stable and correctly oriented in the stack | exceptions related to rigidity or centre of gravity | | mixed orders | sheet does not block the machine | every part is linked to the right order and zone | unidentified or doubtful items | | sensitive surface | sheet route creates no contact damage | gripping and stacking leave no unacceptable marks | final quality gate |

For every row, the supplier should state a documented limit rather than simply “supported,” as well as the required gripper, number of zones, dependence on CAM data and the failed-pick scenario. Verify the record at the factory acceptance test (FAT). If part of the portfolio fails, do not hide it: record the share of orders, manual route and effect on the expected result.

The comparison may show that a simple unloader already removes the primary pallet blockage and that sorting is not a system constraint. In another plant, the machine keeps cutting while a downstream operation waits for a set, making a part picker potentially more valuable. Decide using complete-set output, labor time, exceptions and floor area rather than the largest number of functions in a commercial product name.

The application matrix also separates function from integration. A mechanically lifted part still needs the correct ID, destination and status. If these are entered manually, include that operation in the economics. If they are transferred automatically, require a demonstration of the interface version and recovery after an error.

What the system can actually remove

When suitability is confirmed, automation can:

• reduce manual sheet lifting and movement;
• release the pallet without waiting for a team;
• decouple the laser cycle from the manual-separation schedule;
• stack recurring parts in defined zones;
• reduce order mixing when data is correct;
• create predictable stacks for the next operation;
• support low-manning operation for a validated product range;
• reduce some damage risks associated with manual handling;
• transfer completion status into the software system.

AMADA states that TK II automatically picks, stacks and sorts, and that accurate stacking can support subsequent robotic bending or handling. This is a capability of a particular product, not a guarantee of end-to-end automation for every plant.

Operations that always remain

### Product-range selection and preparation

Someone must determine which parts are suitable for automatic gripping. Sharp, perforated, very small, long and narrow, heavy, unstable or surface-sensitive items may require a different gripping strategy or manual route.

### Input-data quality

The system relies on correct geometry, revision, material, cutting order, order identifiers and stacking plan. Incorrect master data is automated along with the process.

### Exception handling

A part may fail to separate, tilt, remain attached to the skeleton, shift or fail to grip. A safe state, notification, recovery procedure and authorized personnel are required. This article does not describe manual intervention inside guarding; it must follow manufacturer instructions and local rules.

### Quality control

A successful pick does not prove geometry or edge quality. Unless a separate inspection function has been validated, the quality gate remains.

### Containers and internal logistics

The machine can place a part in a zone, but the zone must be cleared. Pallets or containers, labels, routes, permitted stack height and access for the next operation are still required.

### Skeleton and scrap

Even when parts are removed, the skeleton and small scrap must be moved, recorded and disposed of. A reusable sheet remnant requires separate identification.

### Maintenance

Grippers, the vacuum circuit, sensors, axes, guarding and interfaces add maintenance objects. Automation does not eliminate personnel; it changes their work.

Geometry defines the limit

Before requesting a quotation, describe the portfolio:

• minimum and maximum dimensions;
• weight;
• thickness and material;
• available gripping surface;
• number of holes and cut-outs;
• long narrow shapes;
• similar mirrored parts;
• tendency to tilt;
• film-covered or scratch-sensitive surfaces;
• nest density;
• micro-joints or remaining attachment;
• number of orders per sheet;
• expected stacking arrangement.

AMADA publishes particular size, thickness and weight ranges for TK II and warns that they depend on configuration and options. This shows why a decision must be checked against documentation for the exact system rather than a general video.

Vacuum gripping is not magic

Successful lifting depends on available area, sealing, cup position, weight, centre of gravity, rigidity and separation from adjacent elements. Manufacturers use different grippers and strategies. Salvagnini mentions standard, multiple and double gripping, while AMADA describes individually controlled pick-up and optimized vacuum pads.

These descriptions must not be turned into improvised instructions. The manufacturer defines requirements, permitted load, vacuum monitoring and recovery. For a tender, require testing of real parts and an exception protocol.

Connection with CAM and cutting order

Part picking is not an isolated robot beside a sheet. It needs:

• an exact nest;
• part IDs;
• coordinates and orientation;
• completion confirmation;
• access rules;
• stacking destination;
• a permitted removal sequence;
• failed-operation status.

Cutting order can affect a part’s stability in the skeleton, while removal order affects access. A feasibility study must therefore include CAM, the postprocessor, controller, automation software and material flow. “There is an interface” is insufficient: demonstrate the version, supported objects and recovery scenario.

Buffers and stacking zones

An automatic system can become a new constraint if a completed stack is not removed. Calculate:

• number of simultaneous orders;
• number of stacking positions;
• time until each fills;
• maximum height and weight;
• collection frequency;
• access for a forklift or trolley;
• separate exception area;
• skeleton route;
• night-shift operation.

Do not design for an average day only. Check the peak combination of large and small parts. At the same time, do not use an extremely rare case as the sole reason to reject the system; it may retain a manual route.

What unattended operation means

Automatic unloading increases autonomy, but unattended operation requires a much wider system: material stock, jobs, gases, waste handling, a stable process, fault detection, fire and laser safety, remote monitoring, permitted responses and a shutdown plan. That is the separate subject of ART-149.

Within ART-148, it is more accurate to say “reduced manual operations in a validated scenario” than “operation without people.” Even when a system can continue a series technically, personnel are needed before and after the autonomous window.

How to perform FAT with your own parts

Build a representative set instead of selecting only convenient items:

1. a typical series; 2. small parts; 3. large and heavy parts; 4. long narrow parts; 5. a dense nest; 6. similar items from different orders; 7. a sensitive surface; 8. parts with holes; 9. a real skeleton with micro-joints where used; 10. a job with an exception.

Define in advance for each item:

• separation success;
• correct destination;
• stack quality;
• absence of damage;
• time;
• response to a failed pick;
• traceability;
• manual operations that remain.

Record not only pass or fail but the cause and required change. One demonstration cycle does not prove stability.

Calculating the effect

Compare the current and future process maps:

• labor time per sheet and order;
• pallet-blocking time;
• output of complete sets;
• work in progress after cutting;
• recutting caused by handling damage;
• hours outside the main shift;
• floor area required;
• energy and maintenance;
• downtime of the complete system;
• share of the product range that remains manual.

Do not multiply rated speed by every working hour. The effect depends on product mix, utilization, downstream capacity and process stability. It is better to express the production acceptance criterion as a demonstrated scenario across the part portfolio.

Exception architecture

Good automation does not hide an exception. For every type, define:

• how the system detects it;
• whether it can continue other parts safely;
• where the doubtful item is moved;
• who receives the notification;
• which context is retained;
• whether attendance is required;
• how the job is recovered;
• how the exception enters analytics.

If one failed pick stops the entire system until morning, autonomy must be calculated with that risk included.

Common purchasing mistakes

• confusing sheet unloading with part sorting;
• showing the supplier only easy parts;
• failing to require a list of unsupported geometries;
• ignoring the skeleton and small scrap;
• failing to account for clearing stacking positions;
• purchasing a robot without integration with CAM and master data;
• treating a picked part as acceptable without a quality gate;
• failing to plan a manual route;
• calculating return on investment from laser speed rather than complete sets;
• excluding service, spare parts, training and recovery;
• failing to check layout, guarding and logistics;
• promising unattended production without a separate safety review.

Technical-specification checklist

• [ ] Loading, sheet unloading and part sorting are separated.
• [ ] Actual operations before and after automation are described.
• [ ] Part-portfolio statistics are available.
• [ ] Manual exceptions are defined.
• [ ] The supplier provides documented limits.
• [ ] FAT includes typical and difficult parts.
• [ ] CAM integration and versions are verified.
• [ ] Stacking-destination rules exist.
• [ ] Skeleton, scrap and remnants have routes.
• [ ] The quality gate is retained.
• [ ] Layout accounts for buffers and access.
• [ ] Recovery and safe intervention are documented by the manufacturer.
• [ ] The KPI is a complete set, not only the robot cycle.
• [ ] Economics includes the share of the manual product range.

Conclusion

Automatic unloading primarily removes repetitive sheet movement and can release the pallet. Part picking adds separation, stacking and sorting of suitable parts. Planning, data, exceptions, inspection, logistics and service remain. The best solution is defined not by a module name but by the process map and FAT on the real product range.

Safe boundaries

• The article does not describe intervention in the automatic zone or recovery commands.
• It does not claim compatibility with a particular L-SEL machine.
• Published AMADA limits are not presented as universal.
• Unattended operation is not promised; that is the separate topic of ART-149.

Need help selecting equipment?