Int J Softw Tools Technol Transfer (2008) 10:205–206 DOI 10.1007/s10009-008-0065-2

INTRODUCTION

Introduction to the special section on self-optimizing mechatronic systems Wilhelm Schäfer · Matthias Tichy

Published online: 31 January 2008 © Springer-Verlag 2008

Mechatronics is the engineering discipline concerned with the construction of systems incorporating mechanical, electronical and information technology components. The word mechatronics as a blend of mechanics and electronics has already been invented 40 years ago by a Japanese company. Today, mechatronics is an area combining a large number of advanced techniques from engineering, in particular sensor and actuator technology, with computer science methods. Typical examples of mechatronic systems are automotive applications, e.g., advanced braking systems, steer/flyby-wire, or active suspension techniques, but also DVDplayers or washing machines. Mechatronic systems are characterized by a combination of basic mechanical devices with a processing unit monitoring and controlling it via a number of actuators and sensors. This leads to massive improvements in product performance and flexibility. The introduction of mechatronics as a tight integration of mechanical, electronical and informationdriven units allowed for turning conventionally designed mechanical components into smart devices. Today, we see the first steps in the emergence of the next generation of mechatronic systems. While “intelligence” in the behaviour has so far always been achieved by gathering information (and reacting to it) from one single machine, the usage and retrieval of information in the future will be characterised by an exchange of information between machines.

W. Schäfer · M. Tichy (B) Software Engineering Group, Department of Computer Science, University of Paderborn, Paderborn, Germany e-mail: [email protected] W. Schäfer e-mail: [email protected]

This can for instance already be seen in the automotive and rail domain: Intelligent lighting systems combine information about their environment obtained from their own sensors with those collected by other cars. In the Paderborn rail system (railcab1 ) shuttles autonomously form convoys as to reduce air resistance and optimize energy consumption. In general, the smart devices of today’s mechatronic systems will turn into “populations” of smart devices, exchanging information for optimizing their global behaviour as well as possibly competing for limited resources. Advanced mechatronic systems are expected to behave more intelligently than today’s systems by building communities of autonomous agents which exploit local and global networking to enhance their functionality. The mechatronic systems of the future will be characterised by the following properties (cf. [2]):

– High degree of concurrency: Systems will consist of a large number of autonomous components, exchanging information while running in parallel. Components may form cluster to collaborate on a common goal but may also compete as to optimize their own aims. – Decentralisation: Due to the high degree of concurrency and distribution systems cannot be centrally observed and as a consequence not centrally controlled. – Self-optimization: These intelligent systems are especially able to adapt their behaviour during runtime. This poses new challenges on their design, in particular concerning the usually complex software and the integration of discrete software for the communication between agents and continuous control software.

1

http://www.railcab.de/en.

123

206

Several disciplines in computer science are affected by this change. For achieving reliable and secure transmission of information the areas of network technology and cryptography are challenged. Current self-* developments in software engineering are already making small steps in this direction. For the design of complex mechatronic systems of the future these have to be combined and complemented with other advanced techniques as to form an engineering method for self-optimizing systems. Such a method in particular has to involve – new modelling formalisms integrating model transformations (describing adaptation, reconfiguration etc.) themselves into the model, – new code generation techniques operating at run-time and taking platform specific parameters into account, – elaborate formal analysis techniques being able to cope with the high volatility of systems (and properties emerging by a continuous dynamic change), – new optimization techniques supporting multiple and possibly conflicting goals, – appropriate development processes which are explicitly tailored towards mechatronic systems of the future. In addition to the challenges that future mechatronic systems will bring to software engineering, there are also a lot of unsolved issues in the design of current systems. While mechatronic systems indeed incorporate parts constructed by different engineering disciplines and computer science, the actual cooperation during the construction is less developed. There is no joint development process, no joint tool usage, no joint modelling formalism and no joint analysis. Every discipline has its own approaches; an integrated framework for the construction of mechatronic systems is missing. The papers in this special section focus on those selfoptimizing mechatronic systems. They address particular aspects of the above mentioned challenges, namely an integrated modelling formalism, particular optimization techniques, and the design process as a whole.

123

W. Schäfer, M. Tichy

In a little bit more detail, the first paper by Burmester et al. [1] presents a model driven development approach towards the design of those systems including a proposal how to integrate the traditionally separated approaches for designing discrete and continuous control systems resp. It also describes an approach to verify crucial system properties by model checking which addresses the safety critical nature of most mechatronic systems. The second paper by Witting et al. [4] focuses on optimization techniques for multiple and possibly conflicting goals so-called multiobjective optimization problems. This supports the dynamic adaption of behaviour. Here, numerical algorithms for the solution of time-dependent problems are developed. The approach is specifically tailored for the restricted amount of resources typically found in mechatronic systems. The third paper by Stein [3] addresses another issue of the design process of complex self-optimizing systems. Here the target of the approach is the design process itself. As this process as well as the resulting designs are usually fairly complex, the paper proposes a semi-automatic support of this process based on methods taken from the artificial intelligence domain.

References 1. Burmester, S., Giese, H., Münch, E., Oberschelp, O., Klein, F., Scheideler, P.: Tool support for the design of self-optimizing mechatronic multi-agent systems (this volume) 2. Schäfer, W., Wehrheim, H.: The challenges of building advanced mechatronic systems. In: Proceedings of the Future of Software Engineering (FOSE) 2007, pp. 72–84. IEEE Computer Society, Washington (2007) 3. Stein, B.: Coping with large design spaces (this volume) 4. Witting, K., Schulz, B., Dellnitz, M., Böcker, J., Fröhleke, N.: A new approach for online multiobjective optimization of mechatronic systems (this volume)