Innovation

Recent developments in robotic hardware for manipulation promise vast new application domains and a huge increase of market penetration by robotics in a diverse set of industries. However, complex motion requirements, enviroment variance, small lot sizes down to lot size one as well as the necessity to manage installations without classical robot programming experts place high demands on robot programming technologies. Since more than two decades, intuitve robot programming is seen as a paradigm to solve this challenge, yet technical obstacles were underestimated for a long time and versatile practical solutions for robotic manipulation had not been found.

The ArtiMinds founders were trained in a leading research group in the field of intuitive robot programming and their domain knowledge is the culmination of over 25 years of research in that group. Our experienced team of engineers, each having many years of experience in research and development of intuitive generation of robot task descriptions, believes to have finally found a suitable set of methodologies. It enables intuitive, fast, robust, well-defined and straightforward generation and incremental improvement of complex robot manipulation tasks.

At the heart lies our proprietary technology to represent robot manipulation tasks by a graph which contains all temporal, geometric, force and tactile constraints as well as conditional branches describing the properties of a task with variability. Such a graph is then executed by robust motion controllers and also state-of-the-art motion planning if necessary.

On top, sophisticated machine learning techniques enable the automatic generation of a graph with all respective parameters from few, intuitive input sources, as e.g. in a CAD-like graphical user interface and/or guiding a robot arm in zero-force mode. These machine learning methods not only perfom generation of the graph and all its parameters, but also care about adjusting imperfect and slightly off data from the intuitive input, leading to robust programs for complex robot tasks in few interaction iterations.

Together with industry, we continually adapt our motion execution back-end for different requirements, from automatic generation of classical robot code over online controllers up to live motion planning. In collaborative research projects we continually transfer cutting-edge intuitive input methods and workflows into our product.