Automated Modular Synthesis for Reliable Cyber Physical System Design

• Applicant: Dr.-Ing. Anne-Kathrin Schmuck
• Subject Area: Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics ; Theoretical Computer Science
• Term: since 2020
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 441512781

Technological progress is currently revolutionizing our society through machines taking over truly complex actions - the first self-driving cars are being deployed, smart grids are being integrated into the global power network and production plants are starting to realize individual product designs in the spirit of the new "Industry 4.0" paradigm.Such automated complex behavior is realized by a large software stack orchestrating the interaction of physical and digital components resulting in a Cyber Physical System (CPS). While such systems operate fully automatically, they are typically built in an ad-hoc manner. This manual design approach is currently reaching its limit, as the resulting code base is becoming so complex that it cannot be correctly handled by humans anymore. Additionally, exhaustive testing of CPS is too costly and time consuming to reach suitable confidence margins for their safe and reliable operation. At the same time, CPS need to be affordable and performant if they are to replace existing technology in industry and our daily life, and thereby unfold their full potential.To change today's design methodology, a promising research direction is the use of formal methods: automated methodologies that ensure system requirements during design-time. The main challenge in their application to CPS is the large amount of interacting heterogeneous components for which synthesis tools must automatically and locally generate code implementing a desired joint behavior. When considering for example a two-joint robot arm mounted on a mobile cart, its physical components are coupled (e.g., the arms' joint angle influences the wheels' friction and the cart moves the arms' mounting point). When actuating these components, synthesized feedback controllers must handle such couplings properly.In addition, synthesized coordinating software must ensure that different tasks performed by different components are scheduled and executed correctly, both sequentially and concurrently, despite component interactions. E.g., when the cart-and-arm assembly is fetching a distant object, the cart must move to the objects location first, before the arm can reach for it. Further, the mass of the arm differs before and after object pick-up. This changes the cart's dynamics due to the changed center of mass of the assembly. Hence, the cart's motion controller implementing the task 'move' depends on its position within the schedule w.r.t.\ the arm's 'pick-up' task.It is the overarching goal of this project to tackle the outlined challenges in automatic CPS design by significantly expanding the scope of formal automated synthesis techniques for CPS.In particular, we will provide a framework that automatically synthesizes and negotiates contracts to resolve component interactions. This allows for automated modular synthesis of controllers, abstractions and coordinating software throughout the layers of a CPS resulting in its reliable behavior.

• DFG Programme: Independent Junior Research Groups