Model-/simulation-based engineering puts digital models and computer simulations at the core of an engineering methodology (consisting of workflows, methods and tools) for robotic technologies (systems and services).
The methodology is rooted in model-based systems engineering (MBSE). “MBSE is the formalized application of modelling to support system requirements, design, analysis, verification, and validation activities beginning in the conceptual design phase and continuing throughout the development and later lifecycle phases” (Int. Council on Sys. Eng.)
The basic idea is to implement comprehensive, simulatable virtual replicas of robotic technologies for the incremental development and optimization of system hardware, control algorithms and lifecycle services (i.e. to support deployment, operation and maintenance). The transition from the virtual to the physical world is then carried out using Digital Twins, which represent and connect to the physical assets – 1) Digital Twins represent the parameters and functionalities of physical assets for purposes of model-/simulation-based workflows, methods and tools, – 2) Digital Twins connect model-/simulation-based workflows, methods and tools to physical assets, e.g. for controlling and/or monitoring them directly.
Digital Twins. Although some prototypical and/or commercial solutions exist to implement Digital Twins in automation (e.g. Siemens PLM and Siemens Mindsphere), there is still a lack of a general, standardized methodology in the I4.0 community which can be applied beyond a limited set of components, manufacturers and applications. Here, our research activities aim at analyzing the requirements and developing a systematic Digital Twins approach for robotic technologies, and to design and develop Digital Twins solutions for and with our industrial partners.
Database technologies in robotics. The robotics community by and large still relies on file-based data storage and exchange which is insufficient for robotic technologies in Industry 4.0 applications. We work on establishing database technologies to enable large-scale/long-term data handling for novel robot technologies such as data-driven optimization using sampling-based techniques for Digital Twins scenarios in new application fields.
Sampling-based optimization. In particular, for motion planning, sampling-based techniques have been established to explore and exploit complex, high dimensional state spaces for feasible trajectories. Beyond motion planning, we research and design novel sampling-based optimization strategies to navigate the complexities of robot technologies in Industry 4.0 applications.
Visual programming. New efforts must be made to support users and other end user roles with managing and maintaining the complexities of robot technologies in Industry 4.0 applications. Here, visual programming allows for breaking complex operations down into so-called skills which can be arranged and executed also by non-roboticists. We deploy VP prototypes to program Digital Twins scenarios and conduct research on systematic skill hierarchies and exception handling.
Virtual Human. As Digital Twins are mostly interpreted to only encompass technical systems today, the concept of Digital Twins still lacks being applicable in collaborative robotics, where human operators are engaged in direct interactions with robotic technologies. Our research is addressing this missing link by investigating Virtual Humans in collaborative scenarios, e.g. to adjust robot motions according to operator actions and/or ergonomic conditions.
Service-oriented architecture. We aim at systematically organizing the methods developed in our various research areas in a modular service-oriented architecture that allows for easy exchange of approaches across the sections, thereby fostering synergies and innovations. Its core concept is to provide our methods as services which can be set up and called in a uniform way from our planning, programming and simulation tools, thus forming an abstraction layer between high level user interfaces and low-level hardware controllers.
Digital Twins to unlock expert knowledge in applications. With the services, we aim to form an extendible library of reusable skills, which encapsulate basic and advanced methods from robot control and computer vision. A skill has inputs to set the parameters of the method (e.g. target poses for a robot motion) and outcomes of the method are communicated as outputs of the skill (e.g. progress with the robot motion). Skills can call other skills to yield new, more complex functionalities (e.g. a complete pick-and-place operation). Based on the abstraction layer, we aim for providing an easy-to-use platform to analyze, simulate, program and command robotic assembly systems. More specifically, we are aiming at developing a platform for assembly automation based on Digital Twins and services for advanced robot control and computer vision that can easily be used by typical operators.
Towards a large structure production. Based on SDU Robotics’ service-oriented architecture, we are entering novel application fields that can highly benefit from Industry 4.0 concepts and technologies to cope with high mix/low volume production. Beyond targeting factory automation as the primary field of application, we aim for transferring our results to novel application fields, in particular for the industrial sectors dealing with large structures such as the construction and the maritime sector. In addition to the fact that large structures are often one-of-a-kind designs and thus extreme examples of high mix/ low volume products, their construction is taking place in highly dynamic production environments.
For more information contact Professor Christian Schlette