Page - 49 - in Proceedings of the OAGM&ARW Joint Workshop - Vision, Automation and Robotics

were interpreted and conclusions drawn. Visual feedback during teach in was requested. If possible, information should be projected to the work piece surface. This would require additional projection technology as proposed by AssistMe. VI. PROPOSAL FOR FINAL EVALUATION An additional projection technology would enable spatial augmented reality methods. A. Robot C Spatial augmented reality interfaces are proposed and implemented as tangible user interface. Physical interaction with the product to process only might further minimize programming effort and be an easy to perceive means of interaction. A tangible marble is used for teach in of process points and the sequence of their processing. Therefore a 3D camera is integrated with a projector to detect marbles [21] positioned on top of screws to acquire spatial process points as Fig. 11 - Tangible User Interface well as taps onto projected buttons to confirm their order or other interactions with the programming system. B. Robot D Robot D is controlled via a 2D interface as depicted in Fig. 13. Process points are entered by tapping onto a 2D representation of the processed object. A machine vision algorithm determines the spatial region of the tapped point and therefore determines both 3D process points and the sequence of the process points from the tapping order. Fig. 14 shows the technology applied to a bin-picking process where one of several objects in the 3D sensors field of view can be selected in a 2D representation of that data. The same technology is applied to the selection of regions and process- points on a single object in the sensor’s field of view. Implicitly also the order of the process points can be entered. Fig. 12 - Tangible User interface system setup Fig. 13 – Define process points in 2D C. Robot E Robot E is programmed by positioning an externally tracked device (Fig. 15) or an extension like a stick to the process point. Once calibrated a precise position of a stick’s tip mounted on an externally position tracked device can be calculated in real- time. Process points and their order are programmed by ordered tipping onto screws in question. Fig. 14 - 2D tap based process point selection (https://www.youtube.com/watch?v=nrhXEqG014o) VII. CONCLUSION The presented study demonstrated that the not- intermediated (direct manual) interaction with the robot can increase the experience of the robot's capabilities (usability, Select positionby click in 2D view Position 2Position 1 Live View of2D Camera Camera Position 49