Part 2/10
To make production lines flexible, future-oriented and economically attractive today, automating work steps is essential. The demand for robots is rising continuously. It is assumed that the industrial use of robots will double in the next five years. This development brings new challenges in dealing with robots.
Programming robots requires a great deal of know-how and trained staff, because commissioning is a time-consuming process. Even after the robot has been set up, errors can occur in manufacturing if production processes are not sufficiently optimised. If the product being manufactured changes, further effort is required. In this context, learning robots make things much easier.
There are already AI-based industrial approaches for detecting errors in production, learning from them and avoiding them in future. However, they are not yet ready for everyday use and are still at the research stage. Complex tasks such as optimising a motion path or finding and grasping an object in a cluttered scene can require learning times of several months.
Looking at self-learning robots from industry reveals great potential in the social sector. There, robotic arms are increasingly being used to support people with physical impairments. Due to high safety requirements, there has so far been only a small amount of research and a small range of solutions on the market. The difficulty is that the tasks carried out in everyday life can vary greatly. While in production the ambient conditions, the objects to be manipulated and the interactions with humans are clearly defined and hardly vary, in everyday life they are constantly changing. As things stand today, classic systems and AI approaches can hardly be used. In future, we will need flexible systems that can be extended continuously.
This is where our current research project HIRAC comes in. Together with the Faculty of Business of Offenburg University of Applied Sciences, we want to develop hardware-independent software that can, for example, control different robotic arms from various manufacturers.
kununu