In close cooperation with the fundamentals-oriented project area A, the design-oriented project area B and the application scenarios of the project area D, project area C deals with the implementation of self-optimizing mechatronic systems on the basis of three central aspects. Concentrating on the aspects “reconfigurable hardware”, “self-optimizing real-time operating and real-time communication systems”, and “agent-based control/regulation” results in very promising and innovative implementation approaches.

Based on the self-optimizing mechanisms that have been analyzed in the first two periods of funding an integrated approach for realization of self-optimization is being worked on. This approach involves the reconfigurable information-processing platform (subproject C1), the real-time operating system and real-time communication system that support self-optimization (subproject C2) up to the integration platform and the agent based control structures implemented upon that (subproject C3). The components RAPTOR-X64 (hardware platform), DREAMS (RCOS/RTOS) and OCM (self-optimizing controllers) make up the hardware/software infrastructure.

Hardware Reconfiguration

In the subproject C1, the opportunities of hardware reconfiguration are still considered a very promising technique for the realization of self-optimizing mechatronic systems. Reconfiguration, e.g., based on “Field Programmable Gate Arrays” (FPGA) and other structures inherited from them, opens new potentials for a cost/value ratio and for the flexibility in respect to self-optimization through replacing the traditional static partitioning in hardware and software (classic HW/SW-co design) with a dynamic partitioning at run-time. Furthermore, applying reconfigurable hardware enables scalable system concepts, which can be implemented efficiently also in future microelectronic technologies. A reconfigurable system environment that efficiently utilizes state of the art technologies is being developed. It consists of a toolchain for the design and analysis of self-optimizing hardware systems that is being adopted in several areas of this collaborative research center. Using abstract models, new methods for placement and defragmentation of hardware modules in heterogeneous systems have been developed and evaluated.

RTOS for Self-Optimizing Systems

Subproject C2: Upon choosing an implementation of self-optimizing mechatronic systems on distributed microcontrollers, an administration layer in terms of a distributed real-time operating system (RTOS) in connection with a real-time communication system (RCOS) is essentially required. When considering conventional systems, they already constitute highly-complex objects. Therefore, in case of self-optimizing systems the challenge on the problem is much higher. In contrast to the conventional proceeding in RTOS environments the system parameters in self-optimizing systems are not known a priori. This is the challenge that subproject C2 “RTOS for self-optimizing systems” confronts. The innovative approach treats the RTOS/RCOS itself as a multi-agent system, which is subject to the principle of self-optimization. Self-optimization in this case is exploited to enable the flexible adaptation to the dynamically changing requirements of the application software in a resource-minimizing and efficiency-optimizing manner. In a close cooperation with subproject C1, it will be examined to what extent it is useful to source out specific services into the reconfigurable hardware.

A focal point of the work are questions of distributed communication-centric self-optimizing systems. Offering a special type of formal verification as an RTOS-service using an online-acceptance test is an essential contribution to ensure high dependability.

Agent-Based Controllers

Considering self-optimizing mechatronic systems as a consequent further development of adaptive controls, the development of innovative concepts for controller design is essential. This is where subproject C3 takes a new way by symbiotically combining behaviour-based and planning techniques with the model-centric and optimization techniques. This way the undisturbed operation of a technical system is to be guaranteed, especially in safety critical areas on the basis of control theoretic methods. At the same time behaviour based programming together with model-based optimization allows for flexible adaptation to environment situations that are hard to predict.

Following the conception and the aim of the project area C, it is closely connected and interlocked with the other areas of the project. The fundamentals of self-optimization that have been explored in the subprojects A1 and A2 are applied in all subprojects of the project area C, especially in subproject C3. A parallel can be recognized between the basic approaches that are followed in the subproject A1 and subproject A2 and the symbiotic approaches including model-based and behaviour-based approaches of the subproject C3.

The design methods from the project area B will be specified in the project area C towards implementation. Hence, the subprojects C1 and C2 provide the required target platforms, by which the realization of the concepts from the project part B will be enabled, while subproject C3 covers the micro architecture agent-based controller implementation that have been consciously excluded from the subproject B1.

The project area covers HW/SW platform that is, in turn, required for the implementation of the application projects of the project part D. This platform will be supplied in form of services on the basis of which application systems will be constructed, and respectively in form of a design technique which will support dealing with the central problem of the design of specific self-optimizing components (Operator-Controller-Module/OCM).