A glance inside a car’s chassis will reveal a disorganized collection of long cables.
As Markus Wnuk, a scientist at the University of Stuttgart’s Institute for Control Engineering of Machine Tools and Manufacturing Units (ISW), puts it: “What we see here is a very unstructured scenario, which is a bad starting point for automation.” That is why, even now in the Industry 4.0 era, cable looms are still laid out and fitted within the vehicle entirely by hand from unpacking the bag in which they are delivered to the final fitting. “What makes automation difficult,” the expert explains, “is the huge number of variants, as no cable loom is manufactured twice.” This is due to the enormous number of components and assembly operations, the very small contact systems, but above all the material.
It looks easier than it is: the robot is supposed to learn to accurately locate a deformable object, such as a cable, and then to grasp it with the aid of control software. Image credit: Universität Stuttgart/ISW
Cables are flexible and easily deformed; they get twisted up, form loops and change shape whenever they are touched. To be able to automatically assemble components such as these in spite of all this, researchers at the ISW are collaborating with scientists from the SimTech Cluster of Excellence to develop new approaches, which will make it possible to combine existing sensor data and new simulation technologies, the aim being to integrate the real-time simulation of a machine and its control system thereby optimizing production processes as they happen.
Their research is being received with open arms within the industry. Programming a mobile robot to perform complex manufacturing processes autonomously is a challenge, particularly as production environments are becoming more dynamica and products increasingly customized. “The psychological pressure is high,” says Prof. Alexander Verl Director of the ISW. As head of the “IC SimTech” industrial consortium, a spin off from the cluster, Verl is committed to the transfer of new know-how to industry and to close collaboration with potential users. Data-integrated simulation science, he says, is booming. The evaluation, automation, optimization and visualization of manufacturing processes based on data produced by the machines themselves are all high on the agenda of research funding bodies and industry stakeholders, which makes it all the more important to ensure that developments meet demand.
As a bridge between basic research and practical application, the “IC SimTech” not only wants to provide information about the new methods and models, but also to actively promote their application in industrial practice. Among other things, the consortium organizes bilateral projects with regional, national and international industrial partners, and provides support for PhD students and postdocs in collaborating with industrial partners and encourages them to use their ideas for start-ups.
Discussing current research topics
“We want to keep companies updated on our research, but we also need their feedback to ensure that our research assumptions are realistic,” Wnuk explains. As a member of the “DataCon [de]” project, which was launched in 2020 and is funded by the German Research Foundation (DFG), the engineer is currently working on another important building block for the transfer of knowledge, a prototype, whose first use case, specifically addresses the assembly of cable harnesses, to demonstrate the potential of data-integrated simulations and to boost SimTech’s profile among the manufacturing industries. This will cater to the needs of interested companies, especially in the automotive industry.
The objective for the production model is to demonstrate how data from the simulation, which is continuously updated in the manufacturing process, can be combined with the recorded data from the machine to control an industrial cable assembly robot as precisely as possible. During the development of the relevant control software, the experts at SimTech are moving from one subproblem to the next in collaboration with their partners in industry. The goal for the initial stage is for the robot to use the merged data to learn how to correctly locate a deformable object such as a cable. This primarily involves recognizing the respective grip point before positioning the object in the correct end location. The completion date for the prototype is 2023 and it will be used at trade fairs and other events.
“What we want is for our inventions to play a useful role in industry,” says Verl. There is still a long way to go before it will be possible to manufacture a production model of the new assembly system. But for the head of the ISW, the value of transfer-oriented research projects is not only about the saleable end product; he is also interested in what his students, the vast majority of whom go on to work in industry, learn along the way. “I also want something to stick in people’s minds.”