Automatic Generation of Control Software for Mechatronics Systems

IOP-IPCR Project 602

The Automatic Generation of Control Software for Mechatronics Systems Project is a joint project between University of Twente (Faculty CTW) and Delft University of Technology (Faculties 3ME and LR) funded by the Dutch government within the IOP-IPCR (Innovative Research Program – Integrated Product Creation and Realization) program.

The development process of mechatronics products is multi-disciplinary and typically involves mechanical design, electronics design, control design, and software development. Such mechatronics products include home appliances, office equipment, industrial machines, medical equipment, and even automobiles and aircrafts. Although each of these different design activities is supported by self-contained tools (such as SolidWorks and Pro-Engineer for mechanical design and Matlab for control design), these tools are in principle independent and not integrated. Sometimes data transfer between these tools has to be manually done and a small change in mechanical design can be forgotten and is not reflected in control design. These situations can lead to longer product development time and software quality problems.

To attack this problem, this project aims to a set of prototype tools and a framework to facilitate seamless integration among these tools, with which an interdisciplinary product development team can (almost) automatically generate control software for mechatronics machines. The prototype will demonstrate generation of control software beginning with functional information defined at a very early stage of the development (not from block diagrams, quantified ladder diagrams, or state transition diagrams). It will correctly reflect physics of mechanical systems in the CASE tools. It should also allow verification of generated control software through virtual reality (VR) based simulation techniques, besides software simulation and hardware experiments for test cases. To do so, the project has to build the following subsystems.

(1) Model-based representation systems for functional and behavioral modeling

(2) Mechatronics feature based product definition

(3) Model-based qualitative behavior generation system from functional information

(4) Model-based quantitative behavior generation system

(5) Control code generation system

(6) Control model generation system

(7) Model-based verification system to verify generated control software

(8) Hardware-based verification system to verify generated control software

(9) Multiple model management system on which these subsystems operate

The project employs four PhD students. They are supposed to carry out research as a member of a closely integrated multidisciplinary research team in a mutually collaborative manner. The industrial partners in the project are Océ, VanderLande Industries, Philips, and Science & Technology. The four research groups and their specific research tasks are:

·

TU Delft, 3mE, BioMechanical Engineering: Prof. Tetsuo Tomiyama
Functional modeling and integration

·

University of Twente, CTW, Design, Production and Management: Prof. Fred van Houten
Mechatronics feature based function modeling and VR-based verification

·

TU Delft, LR, Aerospace Design, Integration and Operations: Prof. Michel van Tooren
Control model generation and software based verification

·

TU Delft, 3mE, DCSC : Prof. Robert Babuska
Quantitative behavior generation, control code generation, and hardware based verification

PhD 1: Functional modeling and integration

Intelligent Mechanical Systems Group

Department of Biomechanical Engineering

Faculty of Mechanical, Maritime and Materials Engineering (3mE)

Delft University of Technology

This project aims at generation of control software for mechatronics machines from functional and behavioral descriptions. The AIO first looks at modeling of functions and behaviors based on qualitative modeling techniques. Then, he/she will develop an algorithm to generate qualitative behaviors from functional information and then implement a system to do so. The second task of the AIO is to develop a framework for multiple model management which integrates other subsystems (mechanical CAD, electronics CAD, CASE tools, and any other tools developed by the team). Such a framework requires ontology-based integration, for which this AIO becomes responsible. The AIO is therefore primarily responsible for developing the following subsystems.

·

Model-based representation systems for functional and behavioral modeling

·

Model-based qualitative behavior generation system from functional information

·

Multiple model management system

PhD 2: Mechatronics feature based function modeling and VR-based verification

Design Engineering

Department of Design, Production and Management (OPM)

Faculty of Constructing Technical Sciences (CTW)

University of Twente

Functional information defined by PhD 1 must be embodied with so-called mechatronics features. This embodiment information will be used in the mechanical design of a mechatronics product and such information as dimensions, material, motion, etc., will be determined. These data will become crucial in subsequent control systems design, because for example data about mass, stiffness, and motion timing of mechanical components need to be incorporated in the controller as design parameters. The AIO will first develop a method to define mechanical aspects of the designed machine using mechatronics features on a mechanical CAD, so that control design parameters will be transported to other subsystem through the integration framework. This task also includes the connection of the mechanical CAD to the integration framework. The second task of this AIO is to develop a virtual reality (VR) based verification method of the generated control software. The VR system is available at the OMP lab. This will demonstrate the system’s capabilities of generating control software in an integrated manner. The AIO is therefore primarily responsible in developing the following subsystems.

·

Mechatronics feature based product definition which generates product definition

·

Virtual reality based verification system of the generated control software

-----------------------

PhD position 3: Control model generation and software based verification

Design of Aircraft and Rotorcraft

Department of Aerospace Design, Integration and Operations

Faculty of Aerospace Engineering

Delft University of Technology

To automatically generate control software, it is crucial to automatically generate a control model. A control model includes, for example, equations and design parameters for servo control and state transition diagrams for sequence control. These data can be partially obtained from a functional and behavioral model, considering physics behind it. Generated control models can be represented as Simulink models. The first task of this AIO will develop a method to automatically generate control models (for both servo control and sequence control) from functional and behavioral models, obtaining for example mechanical design data through the integrated framework. The generated control model is used for two purposes; one to further generate control software and the other to verify generated control software. The second task of this AIO is to develop a software based verification system for generated control software. In other words, this second task is to develop a virtual functional model of the designed machine that can be operated with the generated control software. This virtual functional model should be further transported to the VR simulation environment (see PhD 2) for verification. The AIO is therefore primarily responsible for developing the following subsystems.

·

Control model generation system

·

Model-based verification system to verify generated control software

PhD 4: Quantitative behavior generation, control code generation, and hardware based verification

Delft Centre for Systems and Control (DCSC)

Faculty of Mechanical, Maritime and Materials Engineering (3mE)

Delft University of Technology

The AIO is responsible first for developing a method and subsystem to generate quantitative behavioral information from qualitative behavioral information generated by PhD 1 and embodiment information created by PhD 2. The generated qualitative behavioral information will be used by PhD 3 to generate a control model and by this AIO to generate control codes including both for servo control and sequence control. The control model itself will be described in Simulink and once complete control algorithm is ready, control codes in Matlab codes should be generated. This task plays the central role within the project. The second task of this AIO is to demonstrate a hardware based verification of generated control software. A simple experimental set-up should be built on which generated control model as well as control software are tested. The AIO is therefore primarily responsible for developing the following subsystems.

·

Model-based quantitative behavior generation system

·

Control code generation system

·

Hardware-based verification system to verify generated control software

Contact

At the University of Twente you can contact: Ir. H. Tragter (h.tragter-at-utwente.nl)

Overall coordination is commissioned by Prof. Tetsuo Tomiyama (@ TU Delft)