Open Controller Enabled by an Advanced real-time Network

Informazioni relative al progetto

OCEAN

ID dell’accordo di sovvenzione: IST-2001-37394

Progetto chiuso

Data di avvio 1 Agosto 2002

Data di completamento 31 Ottobre 2005

Finanziato da

• Programme for research, technological development and demonstration on a "User-friendly information society, 1998-2002"

Costo totale € 4 108 388,00

Contributo UE € 2 599 983,00

Coordinato da

FIDIA S.P.A.
Italy

Risultati finali

Component-based open control reference architecture

The OCEAN project defined an extended component based open numerical control reference architecture for machine tools, not delivered as open source but whose standardised interfaces for motion control components are publicly available. In fact the results achieved within former research projects were not flexible and granular enough as the monolithic software blocks of the control system remained almost the same. In order to take advantage of open control systems it was necessary to extend the existing reference architecture and to decompose the monolithic blocks into components with clearly defined interfaces, described in standardised IDL format (e.g. for RT CORBA). These specifications were published for further input and implementation by users in the field of control techniques. Thus it is now possible to integrate additional functionality and third party software just using the standardised interface description, without any need of adapting interfaces. The key innovative features are: - Non-ambiguous interfaces have been specified in a reference architecture that will enable an interoperability of control systems and additional components that can be supplied by third parties. - Existing reference architecture have been extended with components which are not covered yet by any standard; the modelling of the single control components and the definition of their interfaces is for sure one of the most innovative aspects of the OCEAN project. - Conventional monolithic control systems can now be opened for the integration of additional functionality. - User interfaces for specific machining tasks or for an individual adaptation depending on the skills of the operator can be substituted in the control system on the basis of DCRF interfaces. - The integration of third party software with standardised interfaces will now allow the reuse of software in different control systems which will result in shorter time to market periods, reduced software development costs and more stable software applications. The project defined the reference architecture for the following motion control components: - HMI (Human Machine Interface) server component: it distributes data from all active components to the client applications; it represents the entry point to gain access to the whole run-time machine functionality. - PLC component: this is a general purpose tool, an application-independent logic sequencer which behaves in a different way according to the programmable logic code downloaded into it. - Kinematics component: this component handles coordinate transformations from the user reference system to the physical axes reference system and vice versa.. - The Motion Control kernel itself can be defined as a component including the command generation for a single channel of a NC. Such a complex functionality includes: interpretation of the part program, motion preparation (look-ahead, tool compensation, etc.), path interpolation and axis control, while it excludes the above mentioned kinematics functionalities. - Process Control component: This component provides different functionalities depending on the considered application and user needs, and offers standard interfaces for the control of specific processes. - Safety Component: This component has to guarantee a fail-safe behaviour of the control system and the machine tool. - Filter Component: The Filter Component includes a library of signal processing functions including standard frequency filters, path smoothing filters, control functions. - Hardware Access Component: it is specifically dedicated to act as an interface among the other components and the hardware, in order to make other components compliant with hardware modules of different vendors and consequently hardware independent. At the end of the project, the consortium published the standardised interfaces, in a public deliverable D4.9, a document giving all information about OCEAN compliant components and controls. As clearly stated in the Consortium Agreement, All specifications of component or other interfaces are distributed without any constraints that hinder their free and unconstrained distribution. All parties agree not to exercise any patent or other intellectual property rights they have about the subjects dealt with in the specifications. All partners agree that the specifications themselves are provided free of charge from the consortium webpage, and that they can be mirrored (without any changes) on other people's servers. Therefore this result is clearly not directly exploitable, but is the basis for the implementation of the OCEAN compliant Numerical Controls. Also its existence and features has been and will be broadly disseminated.

Distributed control system real-time framework (DCRF)

The "Distributed Control System Real-Time Framework" (DCRF) is a standardised communication platform real-time capable in distributed hardware environments, able to host numerical control components in distributed open platforms. It is based on standardised communication systems and delivered as open source. The DCRF provides a real-time communication API. This framework is to be coupled with standardized interfaces enabling the integration of external real-time critical and non real-time critical control components. The most innovative aspects of the DCRF are: -The DCRF is real-time capable in contrast to former approaches like OSACA, OPC, etc. This ability is crucial for distributed control systems. -The DCRF enables a flexible composition and reconfiguration of control systems with manufacturing task specific functionality on the basis of a common communication platform. In fact at the moment there is no software tool able to make a high-quality software engineering design for such a complex system as a numerical control. The result is a scalable, portable and flexible framework. -The DCRF is offered as open source, which helps to gain a wide propagation easily, especially for SMEs that do not have resources to develop a control platform by themselves. -New generations of control systems can be built upon the DCRF which will simplify the design and implementation of distributed real-time controls. -The DCRF is not limited to control systems for machine tools but it can rather be applied to various purposes in the field of control systems (e.g. chemical industry, packing machine industry, plastic machine industry, ...). The consortium found that TAO is the most suitable CORBA ORB, and RTAI/LXRT the preferable Linux real-time extension to form the implementation base of the DCRF. TAO has been developed at the Washington University of St. Louis with a special focus on its real-time capability. TAO offers a variety of CORBA services as well as the quality-of-service functionality that can be applied to the supervision of modular software systems. RTAI is preferred as the Linux real-time extension rather than RTLinux because RTAI offers the best real-time support, which is crucial for the realisation of the DCRF. RTAI offers the best C++ support and a more complete feature set. An additional criteria is the availability of LXRT for RTAI that enables the operation of hard real-time tasks in user space. This allows the use of various standard operating system functionalities in real-time applications. In conclusion, the TAO-RTAI solution has been selected as the best combination to fulfil the requirements of the DCRF. It allows a distribution and cross platform deployment of components in a real-time environment. The last release of DCRF 0.13-rc1 was tested on RTAI 3.0r4. A code release was published on the SourceForge site and mailed to the ACE/TAO and RTAI mailing lists. A full-time hardware platform was constructed for testing the port. But partners also analysed the item of the porting to different versions of RTAI, mainly focusing on the current Magma (3.2) and Fusion (4.0) branches, as these are the only two versions of RTAI which are getting any attention by the core developers. At the moment two source and build trees setup exist: one for TAO on top of RTAI Magma and one for RTAI on top of Fusion using the "native skin". The "native skin" is the API provided on top of RTAI Fusion which resembles RTAI 3.2 most. LibeRTOS was investigated as an alternative to RTAI/LXRT. The measurements based on the 2.4 Kernel were very promising and the tests were continued with the forecoming LibeRTOS based on a 2.6 kernel. At the end of the project, the consortium provided a working version of the DCRF, and Deliverable 4.10, a manual for the end users of the DCRF, a document describing how to compile, install and set up the DCRF. The DCRF is freely delivered as open source; therefore it is clearly a result that is not exploitable, but that sets the basis for all the outcome of the projects, and whose existence and features has been and will be broadly disseminated, in order to promote its use in different industrial environments. Several dissemination channels were and will be used to this goal: presentations in conferences and workshops, articles in specialised literature, status reports posted on specialized web sites, presentations in machine tool/controller fairs. The OCEAN consortium also made his best to make it visible to the Open Source community through the web site and a dedicated mailing list.