It is not the robot that decides
Many automation projects start by asking which robot to buy. That is understandable but often misleading. A robust gripper is of little use if parts vary widely in the feeder, and the best camera system only works reliably if lighting, part position and tolerances are matched to it.
The better starting question is: what performance must the system reliably deliver in series production? That includes:
- Cycle time including reserves for ramp-up and variants
- Part variety, batch sizes and frequency of changeovers
- Required availability and unattended running time (autonomy)
- Quality requirements and documentation obligations
- Space, environment and interfaces to existing production
Only on this basis can you decide whether a simple handling cell is enough or a turnkey production system is more economical.
The building blocks of a robot cell
| Building block | Task | Consequence of poor design |
|---|---|---|
| Robot | Motion, cycle, repeatability | Too little cycle reserve or unnecessarily complex kinematics |
| Gripping | Secure pick-up and transfer | Dropped or damaged parts, long changeovers |
| Feeding | Presenting parts in a defined position | Stoppages, mis-picks, manual intervention |
| Vision | Position detection and feature inspection | Wrong orientation, scrap |
| Quality assurance | Inspect, release, document | Rework, undetected defects, missing traceability |
| System integration | Control, safety, interfaces | Long commissioning, communication problems |
The best individual component does not automatically make the best overall system. Weak points usually lie at the interfaces.
Using vision and quality assurance properly
A vision system detects the position, orientation or features of a part and passes this information to the robot so it can pick and align reliably. Quality assurance has a different job: it checks whether the result meets requirements and releases parts for the next step. Both may use the same camera, but they need different criteria.
Inspect as early in the process as possible rather than only at the end. Detecting defects on entry prevents faulty parts from being processed further. With many variants or changing surfaces, even small changes in lighting or contrast can affect detection.
Feasibility with real parts
Have gripping and image-processing concepts tested with real production parts before ordering, including borderline samples at the edge of tolerance. Pre-series samples are often too good.
From idea to production system
- Understand the process: which steps are to be automated, and where are today’s bottlenecks and quality problems?
- Record requirements: part geometry, tolerances, volumes, cycle time, autonomy and environment in a specification.
- Check feasibility: test critical points such as gripping, separating and recognition with real parts.
- Define the concept: design robot, gripper, feeding, vision, inspection and material flow together.
- Plan integration: consider control, data connectivity, safety, usability and maintenance access early.
- Accept and ramp up: factory acceptance at the supplier, site acceptance and structured optimisation up to series performance.
Machine safety belongs in the project from the start. In Switzerland, machine safety is governed by the Machinery Ordinance, which is aligned with EU law. The complete system requires a risk assessment and a declaration of conformity. Agree contractually who takes on this responsibility, and check the current state of the rules, as the new EU Machinery Regulation applies from 2027 and Switzerland is adapting its law accordingly.
Economics and choosing an integrator
Robot automation pays off above all for repetitive, time-consuming or error-prone tasks and where positions are hard to fill because of the skills shortage. The robot itself often accounts for only part of the investment; engineering, peripherals, safety technology and commissioning come on top.
Questions for an integrator
- Which comparable systems have you built, and can we visit one?
- How do you test feasibility before we commit?
- Who is responsible for the conformity of the complete system?
- What do training, documentation, spare parts and remote maintenance look like?
- Which criteria govern acceptance, for example cycle time and availability over a defined run?
Involve future operators and maintenance staff early as well. They often know the quirks of the parts and today’s process better than any specification. A system that is easy to restart after a fault usually achieves higher availability in daily operation than a technically more ambitious solution that only specialists can handle.