Advanced sensing and control techniques to facilitate semi-autonomous decommissioning. 1998 annual progress report

'This research is intended to advance the technology of semi-autonomous teleoperated robotics as applied to Decontamination and Decommissioning (D and D) tasks. Specifically, research leading to a prototype dual-manipulator mobile work cell is underway. This cell is supported and enhanced by computer vision, virtual reality and advanced robotics technology. This report summarizes work after approximately 1.5 years of a 3-year project. The autonomous, non-contact creation of a virtual environment from an existing, real environment (virtualization) is an integral part of the workcell functionality. This requires that the virtual world be geometrically correct. To this end, the authors have encountered severe sensitivity in quadric estimation. As a result, alternative procedures for geometric rendering, iterative correction approaches, new calibration methods and associated hardware, and calibration quality examination software have been developed. Following geometric rendering, the authors have focused on improving the color and texture recognition components of the system. In particular, the authors have moved beyond first-order illumination modeling to include higher order diffuse effects. This allows us to combine the surface geometric information, obtained from the laser projection and surface recognition components of the system, with a stereo camera image. Low-level controllers for Puma 560 robotic arms were designed and implemented using QNX. The resulting QNX/PC based low-level robot control system is called QRobot. A high-level trajectory generator and application programming interface (API) as well as a new, flexible robot control API was required. Force/torque sensors and interface hardware have been identified and ordered. A simple 3-D OpenGL-based graphical Puma 560 robot simulator was developed and interfaced with ARCL and RCCL to assist in the development of robot motion programs.