Seite - 27 - in Proceedings of the OAGM&ARW Joint Workshop - Vision, Automation and Robotics

abstraction levels and the (mostly) automated translation of models to machine readable codes increases efficiency for creation and maintenance of applications. Combined with the improved quality of implementation and reduced fault susceptibility, flexibility in production can be increased significantly. Another approach is the use of domain specific languages (DSL) [4]. These have been drawing attention especially in the area of service robotics during the last few years. C. Complex Wood Manufacturing Processes for Robots Typically a user of such a robot based environment has to performdifferent subtasksalongagivenmanufacturingwood processing chain [6]. A technician may look at technical features, the customer indeed is mainly interested in how satisfactory theyare solved.Automatically theuser is scoring his satisfaction within the process execution, either in terms of time to perform, the process stability or comparing out- comequantity, efficiencyand limitations.Complexitymaybe defined as the necessity to involve more than pure kinematic robotic control to perform a task, therefore going beyond the well-known operations. III. PROBLEM STATEMENT Themain focusof innovativeandnewtypeofmodelbased robot programming for wood manufacturing industries lies on the ease of use for non-experts in robotics coming from this specific domain. No knowledge about traditional off- line-programming or specifics of robot programming should be required. This requires the selection and implementation of an applicable method for creating the necessary data about manufacturing steps for the machining of solid wood elements with an articulated robot using different kinds of tools. As human labour is an integral part of manufacturing however, to make such a system available on the market today, interactive methods for the collaboration between human workers and a robotic system have to be examined and established. Production steps where human interaction with the work piece is necessary should be kept at a minimum, where the machine operator can decide between using human labour where it might save time or material, while keeping the robot as fully engaged as possible. To ensure this, the commands for the robots need to be generated directly from the CAD plans, which have been enhanced with semantic information. Interfaces between CAD systems and robot installations need to use standardized formats as often as possible. In CAD there are various specific formats available, which need to be analysed, evaluated and may be adapted or enhanced for the intended use. The evaluation of existing open standards is also an important task in the RobWood project. However, future industrial use and uptake of the proposed technologies are subject to support by the CAD system providers. There are standardized interfaces in existence for use in CNC (computerized numerical control) production environments (DIN 66025/ISO 6983). These are mainly used for portal systems and toolsets and might need adaptation for usage with buckling arm robots. Starting from a desired pose of an robots attached tool, one has to solve the so called inverse kinematics prob- lem for a robot to obtain the corresponding robot axes configuration [5], [10]. This problem is difficult to solve mathematically, and typically has several solutions as shown in Figure 1 for an elbow up and elbow down configura- tion. A model-based approach, however, would offer this functionality for a broader spectrum of robot types and in particular, as accessible component earlier within the model- based software tool chain. For the intended use case the number of produced unique pieces is one, although the reuse of parts of the models has to be considered. This also affects the specifications for creating enhanced user- and programming interfaces. Apart from specific variables such as technical interfaces, drivers, catalogues and configurations, the resulting user interface should strive to be as independent from the robot system as possible. System configuration should be kept to a minimum and is done with simple configuration procedures and minor manual settings. The process should not require an expert in robot programming or setup. Special care is to be taken to ensure that the person resetting the system is not able to bypass security measures built into the system and that access to the robot for configuration is only available during idle times. Especially the manual steps required during a reset are to be built with simple visually enhanced instructions so that the wood manufacturing personnel can safely perform the necessary procedures without the help of experts in robotics. The particular requirements of the wood manufacturing industry imply the necessity of a tool catalogue, which holds all required and possible tools as well as their corresponding procedureparameters suchasspeedofoperationand logistics of operation. These catalogues can differ between system configurations, for example for the same procedure but requiring different tools. Bringing the various configurations into a form that is both readily comprehensible as well as comprehensive will be one of the challenges of the proposed project. To provide this system also for timber frame construction the various steps related to treatment, positioning and assembly need to be considered as part of the overall procedure, even though these tasks are not be fulfilled by the same robot system but with an assisting system, yet at the same time keeping the transparency for the user. Special focus is granted to the usage of the envisioned systems in time sharing and collaborative environments, where different companies share one robot system or a specialised provideroffers the robot system as aservice. This increases the importance of using standardized interfaces to offer the simple exchange of treatment models and object data while at the same time ensuring IPR (intellectual property). The vision for the final system is to integrate the whole class of production control systems and production planning systems. As various robot systems could be part of the same production not only the machine-to-human but also 27