Unlike many other production steps, assembly is still a largely manual process. There are several reasons for this. Often, the available automation solutions are not adaptable enough. Setting them up or retooling is costly, which is a disadvantage in times of increasingly personalized production. Finally, the joining processes themselves are sometimes very complex. This is due to flexible components, large supply or part tolerances, extremely high demands on process accuracy or complex joining strategies.
After all, the spectrum of assembly-related tasks is large. Screwing, riveting, clipping, pinning, gluing, stapling and nailing are just some of the tasks that spring to mind. For all of them, the workpieces require special handling as well as a combination of specific movements and the application of a defined force - things that humans often get right intuitively or through experience are very difficult for robots.
Despite these challenges, the demand for more automation in assembly is high. This is mainly driven by the shortage of skilled workers, the monotony of some tasks, the need for consistent quality and economic aspects in a high-wage country like Germany. Fraunhofer IPA addresses these market needs with its consulting and development services for robot-assisted assembly. The range literally goes full circle, covering the entire product life cycle.
Automation-friendly product design
Accordingly, this starts with the design of the product. This is because the assembly team often finds that components are designed in a way that is unfavorable for automated assembly. This makes the use of robots technically difficult or uneconomical. Such an unfavorable design is often recognized by the experts during an inspection of the production facility, where they perform an Automation Potential Analysis. By the way, more than 500 companies have already used this analysis service! If the APA concludes that the process can only be automated with great effort or makes no sense under the given criteria, this can sometimes be changed by using Design for Automation (DfA).
A catalog of questions is available for this purpose, which is used to determine how “automation-friendly” a product is and what needs to be changed. The questions deal, for example, with the modularity of a product structure, the use of standardized components or the number of parts involved.
Pitasc software supports high variant numbers
If, on the other hand, the APA concludes that robot-based assembly would make sense, a series of software and hardware developments from the Fraunhofer IPA come into play that can significantly simplify the use of robots. The procedure for this ranges from identifying process and product requirements, through designing tools and fixtures, to programming robots and producing a small series on a laboratory scale.
The robots are programmed with the aid of the “pitasc” software, which particularly addresses applications with high variant numbers. This is because pitasc eliminates the need to program an assembly task step by step. Instead, programming is carried out in a modular and structured manner in relation to the workpiece. This is based on data from a sensor attached to the robot. Another advantage of this programming method is that predefined and reusable program modules already exist which can also be easily adapted to new variants. If, for example, the position of the robot, the fixtures or even the end effector are changed, the same program can still be used.
Artificial intelligence opens up new possibilities
Artificial intelligence (AI) technologies, and in particular the subfield of machine learning, are currently a real game changer when it comes to assembly. They enable robots to be programmed faster and with less effort, even without in-depth robotics knowledge. One example of such an AI-based implementation is “imitation learning”. Here, the robot learns by observing humans performing a task and adopting these sequences. Sensors record the movements of the components and the resulting contact forces. In this way, even demanding programs can be implemented, such as sliding a robot over a surface or executing complex snap-fit connections.
The assembly team implemented a further AI-based application in the “rob-aKademI” research project. Here, the method of reinforcement learning, i.e. learning by trial and error similar to the way children learn, was used. Tasks are first learned in a simulation environment to make sure that no hardware gets jammed or worn out. In this, the robot autonomously explores its environment and plans its behavior, thereby continuously optimizing it independently.
Robot-based disassembly for more sustainability
Incidentally, the low level of automation mentioned at the beginning of this story applies not only to assembly, but also to disassembly, i.e. the end of the product life cycle. Here, too, there are many good reasons for increasing automation: The tasks are monotonous and sometimes dangerous and costly. Very few suitable automation solutions have been available up to now. The objects to be dismantled have highly varying dimensions and different types of joins, so that in many cases - if the process is actually automated at all - they are simply shredded. Consequently, valuable raw materials are lost.
To remedy this situation, the “DeMoBat” research project has developed numerous tools and technologies that make it possible for robots to dismantle e-car batteries. Solutions range from a disassembly-friendly battery design, through methods for testing capacities and for handling the batteries, to eight fully-implemented robotic tools for non-destructive disassembly right down to cell level.
Other projects also demonstrate how disassembly technologies can increase sustainability by automating recycling processes. The “Desire4Electronics” project focuses on ways to reprocess electronic waste from private households. The “ProDiREC” project aims at achieving a more sustainable use of rare raw materials in lithium-ion battery production. “ReNaRe” explores how robots can be used to dismantle future electrolyzers. This shows that robot-based disassembly is really taking off. In this context, the institute will also be providing a new consulting service called Design for Disassembly, starting in the fall. Similar to the afore-mentioned Design for Automation, this is about designing products in a way that makes automated disassembly easier. New products can then be planned and assembled from the dismantled components, thus closing the circle around Fraunhofer IPA's range of assembly services.
Fraunhofer Institute for Manufacturing Engineering and Automation IPA
