e-Kinematix is a DfG-funded project, in which the Institute for Mechanism Theory, Machine Dynamics and Robotics, the Fachinformationszentrum Karlsruhe and the FG Konstruktionstechnik of the Technical University Ilmenau are involved.
The e-Kinematix project aims to establish a virtual research environment to support the development of motion devices. To this end, technologies for project organization, implementation and documentation are to be provided. Tools for administration, research, synthesis and analysis are created in individual modules. Particularly important for the development is the documentation of the results of the work steps and the linking of the collected data. The basis for the e-Kinematix project is the e-Research environment eSciDoc, the Digital Mechanisms and Mechanism Library (DMG-Lib) and the application Gecko. The combination creates a comprehensive tool to support the development process.
In order to realize an optimal virtual working environment within the framework of the e-Kinematix project, this should be oriented to the already used methodical development processes, should have a supporting effect and should not entail any limitations or additional expenditure for the engineer. In practice, this requirement means an optional use of subaspects or modules of the working environment in which information can be automatically collected, linked and processed. The modules focus on both the organization and the execution of the process. A generally accepted procedure for the development of technical products is VDI 2221 and VDI 2222, Figure 1 describes a basic procedure for a general product development with a subdivision into logical steps that are run through iteratively.
In each process step, information is collected and sorted, subtasks are solved, solutions are analyzed and evaluated and the further procedure is decided. Results must be documented in all phases. The quantity of documentation generated can only be processed further if a clear structure or assignment of the individual documents/document areas to each other is created. In terms of process efficiency, this assignment should be carried out without great effort.
Virtual Workspace Modules
The basic module is the eSciDoc infrastructure, in which process organization is implemented in the form of project, user and rights management as well as document and data management (information organization). Particularly important aspects are the creation and configuration of projects, users and their rights as well as the setting, updating, retrieving and versioning of the documentation. In addition, metadata is to be automatically collected for long-term archiving, thus ensuring traceability of decisions. The basic module should also form the central entry point.
The search module offers access to many sources of information such as DMG-Lib, design catalogues as described in VDI 2727 (4), IGM mechanism model collection and mechanism lexicon, online databases for patents or technical literature, specialist vocabulary and general search engines. In addition to storing search results, a particularly important aspect is the collection and administration of metadata for searches, results and their sources, as well as further processing and linking to other documents.
The analysis module, the structural synthesis module and the dimensional synthesis module are combined within the Gecko program. The analysis module is used to evaluate the kinematic properties of a transmission such as position, speed, acceleration, ability to circulate, transmission behaviour etc. Results are stored in a neutral data format. Structural parameters must be known for analysis. These can either be entered directly in the structure synthesis module via a graphical interface or determined from the search module. In addition, analysis data from other programs can be stored and managed using the basic module. The structure synthesis module is used to collect and extend information from the search module. In particular, it allows the user to change parameters, enter classification criteria and evaluate the properties of the solution. As part of this information enhancement, the reference to the source of information is maintained in order to enable subsequent updates of results and improve traceability. In the custom synthesis module, different synthesis methods of the transmissions are described.As part of this information enhancement, the reference to the source of information is maintained in order to enable subsequent updates of results and improve traceability. In the made-to-measure synthesis module, different synthesis methods of transmission technology are implemented as interactive procedures. At this point, the supporting documentation should be linked to the synthesis methods by using the search module.
eSciDoc is an open-source e-Research environment that was developed from a joint project of the Max Planck Digital Library and FIZ Karlsruhe and is now being further developed by a community. It consists (5) of a group of core services ("eSciDoc Infrastructure") and an increasing number of applications that build on this infrastructure. In addition, eSciDoc offers complementary services that provide recurring tasks such as validating inputs, transformations or implementing protocol-specific interfaces in the form of independent services. Thus, eSciDoc can be regarded as a collection of loosely coupled services in the sense of a service-oriented architecture (SOA).
But what is an e-Research environment and what features should it have?
There are many different ideas and definitions. The term e-Science originates from data-intensive disciplines, mostly in the natural sciences. Data volumes have increased enormously in recent years due to the largely automated collection of data. The acquisition of knowledge through data analysis is even seen as a completely new form of scientific procedure. Grid infrastructures, such as D-Grid in Germany (8), have been created for the storage and analysis of data volumes. E-Research is now expanding this approach to include other disciplines such as the humanities and so-called "small sciences", i. e. disciplines in which relatively small amounts of data are generated. An important aspect is the collection and management of research data, i. e. the data that is relevant for the work of the scientists. In contrast to systems from the library area, which allows the support of all conceivable file formats and their description by specific and at best within a discipline standardized metadata profiles. Such systematic data storage meets the requirements of good scientific practice in handling research data, but on the other hand many scientists still face great challenges, as recent studies show.
But e-Research also allows new forms of collaboration. Increasingly, scientific work is carried out in teams. Here, a structured management of the data can also enable distributed work. For this purpose, eSciDoc offers functions such as granular access rights and the versioning of data objects, which allow changes to be tracked. So-called audit trails, which record changes with time and originator, also make it easier to understand the data retrospectively. On this basis, so-called virtual research environments such as e-Kinematix can be created.
Within the framework of the project, an application with several modules will be developed based on the eSciDoc Infrastructure. Existing software such as Gecko and DMG-Lib can be used for this purpose. The integration of these existing applications shows an important design approach of eSciDoc: integrating proven tools of the scientists instead of re-developing them for an e-Research environment. On the one hand, this reduces the time and effort required for implementation and ongoing maintenance of the software, and on the other hand, user acceptance is increased by maintaining the familiar and adequate tool. This integration is made possible by the open interfaces of eSciDoc, which are based on proven and widely used standards such as Hypertext Transfer Protocol (HTTP) and Representational State Transfer (ReST). The individual services of eSciDoc communicate with each other via these interfaces, and existing applications or services can also request or write data, for example.
eSciDoc is not based on a relational database, but saves the data as files in the file system. For this purpose, eSciDoc uses a proven open source repository system: Fedora Commons. All metadata is available in the form of a separate XML file. On the one hand, this approach restricts performance in complex queries or when working on large amounts of data. On the other hand, there are hardly any restrictions with regard to the size of data objects, the structure of metadata or possible field lengths. The data can also be read and understood without the actual repository system.
Repository systems are ideal for managing unstructured and semi-structured data. This is accompanied by a multitude of possible object forms that can make it difficult to create an application. For this reason, eSciDoc defines a few basic object types:"Item" as a content-bearing object,"Container" as an aggregation object,"Context" as a higher-level, structuring object (comparable to a volume in a file system to which access rights and configurations are attached) and "Organizational Unit" to map the organizational structures of an institution. Content models can be used to specialize items and containers, for example with regard to metadata profiles to be used. eSciDoc provides a separate service for each basic object type (see Figure 2). Above this is a security layer that checks all incoming requests against the rights management. Applications like e. g. the modules of e-Kinematix don't need to implement their own permission check. This is done centrally and thus consistently across all modules in the eSciDoc Infrastructure.
Dynamic geometry programs such as Gecko make it easy to create graphical elements and define their dependencies. In the context of product development, this enables the creation of models based on geometric coherences and, in addition, the modification of the model by simple manipulation of the elements with immediate updating of the entire construction. In particular, the ease of use via an appealing user interface and the interactive model update for analyzing the effects of model changes are particularly advantageous in early product development phases.
Using paper, pencil, compass and ruler, the graphic design techniques of the gearing technology are very static and often confusing and costly. The result often leaves a lot to be desired in terms of accuracy.
These disadvantages can be put into perspective by using dynamic geometry programs. This results in a particularly user-friendly alternative to analytical and numerical calculation methods. With Gecko, the graphical synthesis and analysis methods of gear technology can be used very precisely, easily, clearly and above all dynamically.
Gecko is developed as a rich client based on the Eclipse Rich Client Platform (RCP) with the programming language Java. The graphical user interface, the creation and representation of the geometry elements for the object-oriented structure of the program are realized with the help of individual blocks or plug-ins. Gecko can be used platform-independently. The only requirement is the installation of a Java Virtual Machine.
The implementation of further functions is focused on the gearing design tools for the creation of points, lines, circles, planes, intersection points and complete partial gears or auxiliary structures. Figure 3 shows an overview of the user interface (GUI).
Today's existing knowledge about motion systems is distributed worldwide and exists in the form of many different sources. These include books, magazines, photographs, physical models, technical drawings and many other media. In many cases, this information is only partially accessible. The form in which they are deposited usually does not meet the requirements for rapid information retrieval.
For these reasons, a comprehensive collection of gearing technology information was started in 2004. The Digital Mechanics and Transmission Library (DMG-Lib) was developed as an interdisciplinary project of the Technical Universities of Ilmenau and Dresden and the RWTH Aachen. Since 2010, the project has been continued within the European framework and since then it has also been developing gearing technology knowledge in the partner countries of France, Spain, Romania and Italy. The information collected can be accessed, inter alia, via the European digital library Europeana.
The aim of the project is to collect, preserve, systematize and present the worldwide knowledge of mechanisms and transmissions. The digital library is designed to meet the requirements of different user groups such as engineers, scientists, teachers, historians and librarians. The information sources of the DMG-Lib are very heterogeneous and range from monographs to journal articles, teaching materials, patents, CAD models, technical drawings, calculation and simulation software, descriptions of important authors and inventors to descriptions of mechanisms with descriptive videos and interactive animations of visualization models (Fig. 5).
The mechanism descriptions of the DMG-Lib are of particular interest for the e-Kinematix project. Together with the stored gearing metadata, they form a kind of solution memory for motion tasks in the design of machines and equipment. These descriptions of mechanisms are usually based on illustrations from literature, patents, teaching materials, technical drawings and, in particular, physically existing visualization and teaching models.
For each mechanism description of the DMG-Lib, more than 40 metadata describing structural and functional properties from a gearing point of view are recorded. This facilitates a targeted search for solutions that go beyond a simple full-text search (Figure 6).
Since no other digital library in the field of motion systems has a similar scope to DMG-Lib, it plays a central role for the research module of e-Kinematix. However, other sources of information should be included in order to broaden the research possibilities. Interfaces - to patent databases, - to full-text archives (e. g. GoogleBooks), - to search engines for Open Access databases (e. g. OAIster) - as well as to general web search engines are planned.
Concept and structure of e-Kinematix
The detailed concept and the software architecture are currently being developed. This section presents the current state of play and describes the considerations to date.
Organization of the data
Working with e-Kinematix is project-oriented, i. e. a separate project is created in the system for each task. This is done via the administration module, which also enables the administration of users and roles as well as the assignment of access rights. Below each project there are three subfolders:"Info" with the search results,"Synthesis" with the models from Gecko and "Documentation" with additional documents, such as the requirements list and the summary evaluation of the results. Different access rights can be assigned for each subfolder to meet the different roles in a team.
Basically, e-Kinematix should support the work in a team and enable several projects to be carried out in parallel. Project teams can be made up of employees from different institutions. Over time, e-Kinematix has to cope with a multitude of projects. For the basic structuring of the content of the underlying repository, e-Kinematix relies on workspaces, in which several projects, e. g. of an institute, a chair or a company, are combined. Figure 8 shows an example of the hierarchical structure with two workspaces and five projects.
Software modules and information flows
E-Kinematix is not developed as a monolithic block, but in the form of independent modules. This makes it easy to integrate existing tools such as Gecko. At the same time, the design of the work process remains flexible. The user decides on the basis of his project requirements which modules he wants to use. For example, he can omit Gecko if he does not have to create models.
Data is stored for all modules in eSciDoc. For example, Gecko can load, modify and restore an existing model from eSciDoc. eSciDoc then manages the change as a new version of the model. At the same time, metadata can be saved semi-automatically with the model. Research results are also managed in eSciDoc. On the one hand these can be documents and references, on the other hand (e. g. in the case of the DMG-Lib) models that can be used directly in Gecko. Interesting is the possibility to let models flow back, e. g. add a model from Gecko to an object in the DMG-Lib. The different data objects (requests, search requests, search results, models, ratings, etc.) are related to each other. This is explicitly expressed in eSciDoc by RDF triples, which form one of the foundations of the Semantic Web. In contrast to conventional hyperlinks, RDF triples also determine the significance of such a link through their predicate. Using this technology, a kind of "knowledge network" spans the data objects, which makes the implicit knowledge of individual project members explicitly accessible to all team members. By storing the data objects in eSciDoc in a uniform manner, users can retrieve them from the various modules. A search result from the search module is visible and readable in the information organization module, as well as the models from Gecko or the requirement list created with the module for managing the requirement list. This allows the user to always select the most appropriate module for his planned activity - always maintaining a consistent view of the project data.
E-Kinematix supports four basic steps in the development process: the specification of the task by recording the requirements, the search for necessary functions and possible solutions, the modelling of selected solutions and the interim analysis and evaluation. The system does not force a linear procedure, but allows any number of modules to be activated at any time for each process step. This accommodates the often iterative way of working in the development process and thus supports the procedure according to VDI 2221 (Fig. 1). The individual steps generate data objects that are captured in eSciDoc and provided with metadata. A framework for documenting the result can be generated at any time with easily automatically recorded information such as a timestamp, module used, registered user, project and object type. If time permits, experiments will be carried out with the enrichment of the framework with text modules in order to support the engineer in this work step and to achieve a standardization.