Intuitive programming of robot manipulators (INTROP)

机器人机械手有几个优点，如力量、速度、耐力、准确性和重复性。到目前为止，这些优势还不能被用于各种潜在的应用。这些应用除了经典的生产领域外，特别是在服务、娱乐和家庭领域。它们的特点是任务执行的重复次数相对较少，工作环境不断变化。因此，只有当当前设置的工作量显著减少，并且执行得到环境感知的支持时，机器人才是合理可用的。设置工作主要受机器人编程的时间和成本的影响，对于许多潜在的应用程序来说，这仍然太昂贵(即使对专家来说也是如此)。此外，基于传感器的任务执行所需的编程特别高。INTROP项目的目标是研究新的、直观的编程概念，使没有事先编程知识的用户能够快速、轻松地为基于传感器的机器人操作器的应用编程。用户应该直观地给机器人“手”，并引导它完成所需的任务。他还可以激活机器人的工具(例如抓取器)或激活环境传感器(例如相机)。这里不仅考虑由一系列动作组成的简单任务，而且考虑具有情况变化和重复执行的复杂任务。与以前的方法相比，这里的重点是直观地编程命令式控制结构(例如，命令、条件和重复)和直观地编程声明性行为(例如移动、躲避、搜索和排序)。编程只需要一个任务演示，随后可以由用户纠正，并且不需要额外的硬件(例如VR眼镜或手套)。

Robot manipulators have several advantages, such as strength, speed, endurance, accuracy, and reproducibility. Up to date, these advantages could not be harnessed for a variety of potential applications. These applications are - in addition to the classical field of production - particularly in the areas of service, entertainment and household. They are characterized by relatively few repetitions of the task execution and by a changing work environment. Therefore, the robots will only be rational usable if the current effort for setting-up is significantly reduced and if the execution is supported by environment perception. The setup efforts are mainly influenced by time- and cost-consuming programming of the robot, which for many potential applications is still too expensive (even for experts). In addition, the programming required for sensor-based task executions is particularly high.The aim of the project INTROP is the investigation of new, intuitive programming concepts that enable users without prior programming knowledge to program sensor-based robot manipulators for their applications quickly and easily. The user shall give the robot intuitively "the hand" and lead him through the desired task. He may also activate the robot's tool (e.g. a gripper) or activate environmental sensors (e.g. a camera). Here not only simple tasks are regarded, consisting of a sequence of actions, but also complex tasks with situational variations and repetitions of executions. Compared to previous approaches, the focus here is on intuitively programming imperative control structures (e.g. commands, conditionals, and repetitions) and on intuitively programming declarative behaviours (such as moving, dodging, searching, and sorting). The programming requires only one task demonstration, may subsequently be corrected by the user, and does not need additional hardware (e.g. VR glasses or gloves).