ndustrial automation was largely built around repetition. When a factory produces the same part thousands of times, engineers can justify spending considerable time developing fixtures, robot programs and dedicated production cells. Once optimised, the system performs the same operation repeatedly with very little variation.
High-mix manufacturing works differently. A machine shop may run several part families in one shift, while production priorities change according to incoming orders. In this environment, the cost of changing the automation can become just as important as the speed of the robot itself.
The automation problem starts at changeover
Consider a CNC machine that produces ten different parts during a week. Automating the loading process sounds straightforward because every job follows the same basic pattern: a raw workpiece is placed in the machine, machining takes place, and the finished component is removed.
The difficulty appears when dimensions, gripping points and loading positions change. If each new job requires a robotics specialist to modify the program, the automated cell may sit idle during every changeover.
The challenge is therefore not simply automating one part. It is making the automation economical across many different parts.
Repeatable skills can replace some one-off programming
A more flexible approach is to define reusable robotic behaviours and change only the parameters that are specific to each job. Picking, placing, loading and unloading are not completely new problems every time the workpiece changes.
This is the type of production environment addressed by Trener. Its Acteris platform currently focuses strongly on machine tending, using reusable robot skills, vision and a standardised application layer intended to reduce the amount of reprogramming required between jobs.
The approach is particularly relevant to high-mix environments because the economic value comes from reducing the friction between production runs rather than maximising the speed of one fixed cycle.
Vision can reduce dependence on perfect part placement
Traditional robotic cells often use fixtures to ensure that every part appears in exactly the expected position. This remains one of the most reliable ways to automate handling, but fixtures themselves can become inconvenient when products change frequently.
Vision provides another option. If the robotic system can locate parts within an allowed area, it may tolerate more variation in presentation. Trener currently describes Acteris as using vision for part identification and handling, including machine-tending configurations where parts are detected in drawer-based systems.
Vision is not a universal replacement for fixturing. Poor lighting, reflective surfaces or overlapping workpieces can create their own challenges. The most effective solution often combines enough physical organisation to keep the process predictable with enough sensing to handle normal variation.
Operators need to be part of the automation model
A high-mix factory cannot rely on an external integrator every time a new production order appears. If routine changeovers require specialist programming, the system becomes difficult to use in day-to-day operations.
This is why operator interfaces are increasingly important. Instead of presenting every change as a programming task, the system can ask for production information such as part type, location or job parameters and translate that into robot behaviour.
Trener’s current materials describe conversational job creation as part of this approach. The objective is to make routine setup more accessible while leaving the underlying robotic execution to the platform.
Automation still needs predictable boundaries
Adaptive software does not mean a robot can handle unlimited variation. Parts still have to remain within the gripping, sensing and motion capabilities of the system. Machine interfaces must be reliable, and safety requirements remain exactly as important as in any other automated cell.
Manufacturers should therefore define the expected variation rather than simply demand “flexibility.” A system designed for ten related components is a much more realistic engineering target than one expected to cope with every product that might appear in the factory.
For high-mix production, successful automation is less about creating a robot that can do anything and more about reducing the cost of moving efficiently from one known task to another.
