The aim of the project is the development of dexterous, robust and cost-efficient multi-fingered robotic hands in combination with control based on modern artificial intelligence methods (especially machine learning) for use in industrial automation. Compared to the two-jaw grippers that have dominated production to date, this is intended to open up fundamentally new fields of application for robotic automation, particularly in the flexible gripping of known objects (CAD models available) with complex and varied geometric shapes and in skilled (in-hand) manipulation. The basis for this is the combination of a highly innovative mechatronic design with antagonistic drives and adjustable stiffness with state-of-the-art deep reinforcement learning methods.
Project details
The aim of the project is the development of dexterous, robust and cost-efficient multi-fingered robotic hands in combination with control based on modern AI methods (especially machine learning) for use in industrial automation. Compared to the two-jaw grippers that have dominated production to date, this is intended to open up fundamentally new fields of application for robotic automation, particularly in the flexible gripping of known objects (CAD models available) with complex and diverse geometric shapes and in skilled (in-hand) manipulation. Especially for the individualized production of the future, as also propagated by Industry 4.0, there is a great need for flexible and dexterous manipulation capabilities. But only through a high degree of automation can production continue to be competitive, even in the high-wage location of Bavaria. In addition, the Corona crisis also shows that a higher degree of automation is generally necessary to keep production running even during a pandemic. However, an expansion of automation means above all the opening up of new areas of application, especially where skilled and flexible manipulation is required. Example applications would be automation in sorting plugs or filling parcels for shipping.