Development of intelligent robot systems based on sensor control

University dissertation from Mikael Fridenfalk, Lund University, Division of Robotics, P.O. Box 118, SE-221 00 Lund, Sweden

Abstract: Sensor based control systems increase flexibility and autonomy in robotics. One of the most common applications in robotics is welding, which is often limited to objects with fine tolerances. To be able to manage objects with relatively large tolerances, such as ship-hulls, sensor guided control systems have to be incorporated that are able to handle large deviations from the drawings. To be able to handle the complexity at the development of such intelligent autonomous systems, a new development environment for robot systems was designed in this work, called FUSE, that integrates software prototyping with virtual prototyping. By the application of FUSE and the methodology developed in this thesis, a seam tracking system for robotic arc welding based on arc sensing was developed and physically validated on-site at a shipyard in the European ROWER-2 project. The aim of the ROWER-2 project was to develop a robot system for welding of hull-sections of large structures such as oil tankers and cruisers. In addition, a true 6D seam tracking system was designed and validated in FUSE that allows seam tracking of seams following 3D spline curves by the real-time correction of both the position and the direction of the torch. Previous systems based on arc sensing only allow seam tracking in 2D or 3D. Similar 6D seam tracking system was also developed and validated based on laser scanning. Since kinematics singularities may disturb the sensor guided control process in robotics, a new method was also suggested that among others eliminates position errors near and on inner singularity points, based on knowledge of the kinematics and dynamics constraints of the robot system and its surrounding environment.

  This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.