How can artificial intelligence be used to create a robot gripping system that is as versatile and energy-efficient as possible and that can trigger a leap forward in the use of intelligent, self-regulating and flexibly scalable handling processes for SMEs and large companies alike? The Fraunhofer Institute for Non-Destructive Testing IZFP, which is based at Saarland University, and the companies Innocise - a spin-off from the INM - Leibniz Institute for New Materials - and Next. robotics are working on this research and development project.

The aim of the project is to further develop artificial intelligence methods in order to design a new type of sensor system that enables the auto-adaptive, robot-assisted manipulation of objects with its own object recognition: the robot arm is given the ability to sense which product it is and adapt accordingly.

Cost-effective, resource-saving robotic gripping systems are required, particularly in highly flexible and frequently changing production lines, but also in demanding environments such as clean rooms and in a vacuum. So-called 'adhesive pads', which are attached to the robot arm, are to be equipped with intelligent sensors for this purpose and then used in production independently of the product. The range of applications extends from sensitive objects or surfaces to heavy loads such as fiberglass or car doors.

The model is gecko feet, which, among other desirable properties, also have the ability to hold heavy loads. They use physical adhesion based on intermolecular van der Waals forces - in other words, they walk and climb without leaving any traces.
In the 'GecKI' research project - funded by the Federal Ministry of Education and Research (BMBF) with around 1.5 million euros - the project consortium aims to develop intelligent and energy-efficient adhesion systems based on nature's example. The research project will make Innocise's innovative adhesion grippers intelligent. "In addition, the information and data obtained from this sensor intelligence will be converted into auto-adaptive motion sequences by a control intelligence in the robot," explains Simon Herter, scientist at Fraunhofer IZFP. Another advantage in a time of drastic energy savings: the adhesion grippers do not require an external energy supply such as vacuum or magnetic grippers, which extends the application possibilities to modular production lines. The integration of the pads into industrial systems could follow the 'plug-and-play' principle.

First milestone for sensing

A first important milestone has been reached with the design of a test stand. "With the help of this demonstrator, we are able to visualize the robot-assisted gripping contact and its structures. We have developed the first sensor prototypes that can teach the bio-inspired grippers to feel," explains Dr. Sarah Fischer, the responsible project manager at Fraunhofer IZFP.