Ever-shorter development cycles, cost savings and higher demands on service and product quality challenge companies to use their resources more efficiently with the help of digitization. For example, Robotic Process Automation (RPA) mimics human interaction with user interfaces (e.g. on screen or keyboard) and thereby provides a way to achieve this goal. Compared to human input, software robots offer a more mature, scalable and reliable approach to handle repetitive processes. However, current process automation solutions require a considerable manual effort. The reason for this is that in order to manually configure a software robot the business processes should be previously either captured and clearly documented or preprocessed and prepared by a domain expert. An intelligent software system that would automatically capture the process information, learn the whole underlying process and execute it on new, previously unseen data is unfortunately not yet available.
AIM OF KI.RPA
As a funding project of the German Federal Ministry of Education and Research (BMBF) within the framework of the funding measure „KMU-innovativ: Informations- und Kommunikationstechnologie (IKT)“ KI.RPA aims to develop a self-learning system that automatically captures and analyzes process information and ultimately creates an intelligent software robot that can independently mimic the learned process. This should considerably reduce the process costs as the routine task of process recording no longer has to be performed by employees. Instead, the employees could focus on more demanding tasks. As a result, companies, and SMEs in particular, should be strengthened by the above-mentioned intelligent solutions, thus obtaining economic benefits.
AWSI as a part of KI.RPA
The project consortium of KI.RPA consists of partners from different domains in order to provide a meaningful representation of the different competences necessary for the implementation. In addition to the AWS Institute, ServiceTrace GmbH, Process Analytics Factory GmbH, Technical University of Darmstadt and Deutsche Telekom Service GmbH are participating in the project.
The main tasks of the AWS Institute as a research partner are reconstruction of executable process models and development of a process configuration component that would make it possible to integrate expert knowledge into the configuration of software robots.
In order to reconstruct a process model from execution data one has to automatically identify and capture relevant process information using artificial intelligence. On the other hand, development of the process configuration component includes implementation of a data model suitable for representation of the acquired process information, implementation of necessary algorithms as well as user interface for dynamic interaction with the process model.