Footnote:
Diese Datenquelle enthält auch Bestandsnachweise, die nicht zu einem Volltext führen.
Description:
Industrial robots are very flexible machines that can perform almost any task – depending on the tools attached and the program they run. Nowadays industrial robots are mostly programmed using proprietary programming languages provided by the robots' manufacturers. These languages are mostly based on very old programming languages and lack support for modern concepts such as object-oriented design; and programs can rarely be reused. To reduce the cost of robot programming, improving reusability is a key instrument. This can be achieved e.g. by using an object-oriented design process. However, standard off-the-shelf programming languages cannot fulfill the hard real-time requirements of robotics applications and robot control. This thesis introduces a data-flow graph based interface that allows the specification of real-time critical tasks, the Real-time Primitives Interface (RPI). Larger robot applications can be split up into independent parts that inherently require real-time safety (such as single motions, or synchronized tool actions). Each such part can be expressed using the RPI and executed with all timing guarantees. The tasks themselves can be specified and joined using an object-oriented interface. To achieve guaranteed transitions from one or more tasks to another set of tasks, synchronization rules are introduced. A reference implementation, the SoftRobot RCC has been created to execute robot programs specified using RPI. To convert programs specified using the object-oriented Robotics API framework, an automatic mapping algorithm from Java-based applications to data-flow-based real-time tasks is presented.