The project RoboMAP is a cooperation of different partners from science and industry, including the Institute for Photogrammetry. It aims at the development of an inspection system for quality assurance in the automotive industry. The objects to inspect are cylinder heads. In particular the valve seats and the valve stem guides are of special interest. The latter lie in the inner part of the cylinder head and are therefore difficult to access. The different kinds of geometry in combination with other goals, such as inline capability and flexibility regarding measurable object variants, impose the need for a new generation of optical measurement techniques. Tactile methods are not applicable, particularly because of their slow operating speed. A boost of flexibility is achieved by integrating a robot into the system in order to position the multi-sensor in an optimum configuration to the object for inspection. This additional capability is further exploited by integrating a single camera into the system which is used for detecting the cylinder head's pose. Thus there is no need for a predefined object position. Next to a general description of the project, we focus on the calibration of the system. Due to the various kinds of involved sub-systems, the overall calibration will be performed in separate steps in a network of coordinate systems.
|Figure 1: Diagram of the network of coordinate systems; double arrows depict transformations, dashed lines depict sensor measurements.|
|Figure 2: Left: Photograph of the device for multi-sensor calibration; Right: CAD drawing of the device for calibration of the Sensors to the robot hand.|
Each component produces or processes geometrical data in its own coordinate system. If the data to process is obtained through another system it has to be transformed accordingly. Figure 19 depicts the linkage of all important coordinate systems. The calibration of the overall system can be understood as the determination of a network of transformation parameters which, if concatenated, enable us to transform any geometrical data into the coordinate system of one of the components.
The sensor coordinate systems needs to be determined with respect to the flange coordinate system (or robot hand coordinate system). These two sets of parameters are referred to as Tool Center Points (TCP) and are directly used by the robot control unit for sensor positioning. Furthermore, the object recognition yields the cylinder head's pose in the camera coordinate system. In order to derive the correct measuring position of the fringe projection system in the robot coordinate system from this result, it has to be transformed accordingly. The necessary calibrations can be classified into the multi-sensor calibration, which includes both TCPs and the transformation between the two sensors and the calibration determining the transformation from the coordinate system of the object recognition camera to the robot coordinate system. The obtained results show that the concept works well. It is recently being integrated to the RoboMAP system for first performance tests.
|Figure 3: Plot of the cylindrical coordinates of a helix shaped measurement at a cylinder of a calibration object.|
|Figure 4: Plot of the cylindrical coordinates of a helix shaped measurement after transformation according to the calibration.|
Cefalu, A., Böhm,
Development of a Robot Guided Optical Multisensory Inspection System for Inline Inspection of Cylinder Heads. International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol. XXXVIII, Part 5 Commission V Symposium, Newcastle upon Tyne, UK. 21-24 June 2010. pp 137-142.