Rob-aKademI – AI simplifies robot programming for assembly tasks

In short

The project partners of "Rob-aKademI" are developing technologies to simplify robot programming and make assembly tasks easier. A digital image, a so-called digital twin of the manufacturing environment, combined with a special programming framework enables robots to learn skills for flexible assembly tasks.

In detail

Artificial intelligence (AI) serves as the basis for this. More precisely, a sub-field of machine learning, the so-called »Reinforcement Learning« (RL), is used. This technology uses an algorithm that learns according to the principle of trial and error, similar to a human being. The robot thus explores its environment autonomously and plans its behaviour while optimizing it independently and continuously. The combination of RL with methods of »Deep Learning« enables the robot to learn even the most complex processes and abstraction to unforeseen tasks is made possible.

Three application-related learning modules will be built in the project. They combine expert robot programming knowledge and expertise about the assembly operation to be performed: 

• "Force-controlled Joining" module for learning robust joining strategies
• "Snap-on Joints" learning module with a detailed physical joining model
• "Perception Module" for object recognition

These modules are used to create robust robot programs to transfer simulation results into reality. The practice-based use cases – namely the assembly of a control cabinet, switch, and THT PCB – serve to validate the results. Some modules are based on the already-available IPA software »pitasc« for force-controlled assembly tasks and extend its capabilities.

»Rob-aKademI« focuses on the needs of increasingly personalized production. Robot programming for assembly applications is still very demanding. Diverse and complex, often force-controlled, processes, high product variants, and short cycle times are some of the challenges. Especially for small and medium-sized enterprises (SMEs) with their customer-specific products, it is therefore often not yet worthwhile to use robots for assembly tasks. However, robots can certainly be a decisive competitive factor for SMEs as well. Thanks to their big experience based on automation potential analyses carried out worldwide, the researchers at Fraunhofer IPA know exactly what companies need and incorporate these findings in the project.