go back

System Architecture

A System Architecture describes the interplay and relations of a number of elements that together create a defined output. It allows, e.g., for easy analysis, control of dependencies and computational timing requirements. In particular, for systems consisting of modules of different types or origins.

The theoretical analysis of System Architectures is important for constructing increasingly complex systems. It becomes mandatory for the design of general architectures, as required, e.g., for building autonomous systems.

The most prominent questions when researching System Architectures are:

  • Can general beneficial decompositions be found that facilitate incremental composition with respect to functions, stability, testability, maintainability, re-usability and even certifiability?
  • Can we find a series of abstractions for perceptions and behaviors that allow researchers to focus on their specific challenge without the necessity to consider all technological dependencies in detail?
  • Can we provide scientific models that guarantee specific requirements by means of the design process rather than by tedious testing and validation?

Cooperative Multi-Robot Systems Architecture Modeling and Simulation

A Multi-Robot System (MRS) is an Intelligent Cyber-Physical System, in which robots cooperate to fulfill a human’s needs.

In modelling the MRS, a proposed system architecture can be simulated and its performance measured prior to implementation. The challenge is to consider many dependent and dynamically changing behaviors due to the current status of the overall system.

To demonstrate the modelling approach, a general purpose MRS case study was created. A system architecture with several components is proposed to enable the robots interaction and planning according to their current, individual capabilities when triggered by a human’s task. Each component is modeled as an activity diagram to precisely represent the behaviors of the component. The components’ behaviors and interaction are then simulated in a software agent environment called Java Agent DEvelopment (JADE). By defining a number of quantitative measurements, obtained from the simulation in JADE, the performance of the proposed architecture is analyzed. This allows a comparison to other architectures and an online observation of the overall system’s performance.

For more information

A. Sadik, C. Goerick, M. Mühlig, “Modeling and Simulation of a Multi-Robot System Architecture”, International Conference on Mechatronics, Robotics & System Engineering, 2019.