Research article Sensor integration in task-level programming and industrial robotic task execution control Rolf Johansson Anders Robertsson Klas Nilsson Torgny Broga ˚rdh Per Cederberg Magnus Olsson Tomas Olsson and Gunnar Bolmsjo ¨ The authors Rolf Johansson, Anders Robertsson, Klas Nilsson, Per Cederberg, Magnus Olsson, Tomas Olsson, and Gunnar Bolmsjo ¨ are all based at the Lund Institute of Technology, Lund University, Sweden. Torgny Broga ˚ rdh is based at ABB Automation Technologies – Robotics and Manufacturing, Va ¨ stera ˚ s, Sweden. Keywords Robotics, Sensors, Force, Material-removal processes Abstract Presents an approach to improved performance and flexibility in industrial robotics by means of sensor integration and feedback control in task-level programming and task execution. Also presents feasibility studies in support of the ideas. Discusses some solutions to the problem using six degrees of freedom force control together with the ABB S4CPlus system as an illustrative example. Consider various problems in the design of an open sensor interface for industrial robotics and discusses possible solutions. Finally, presents experimental results from industrial force controlled grinding. Electronic access The Emerald Research Register for this journal is available at www.emeraldinsight.com/researchregister The current issue and full text archive of this journal is available at www.emeraldinsight.com/0143-991X.htm Introduction Modern robot control systems used in industry at present provide highly optimized motion control that works well in a variety of standard applications. To this end, computationally intensive model-based robot motion control techniques have become standard during the last decade. The principles used have been known for much longer, but deployment in products required affordable computing power, efficient engineering tools, customer needs for productivity/ performance, improved end-user competence in utilization of performance features, etc. The main objective in robot programming is to complete a desired set of tasks. This is accomplished by describing a sequence of robot motions and activities and teaching the robot these motions and activities. The teaching of the motion paths and activities can be divided into two groups: online and off-line programming. In online programming, the robot is involved at several stages during the programming process; in off-line programming, the robot does not participate in the program design, but uses programs that are ready- made or predesigned (Nilsson, 1996; Nilsson and Johansson, 1999). Online programming of robots – also known as teach-in or show-and-teach – involves manual moves of the robot to each desired position. The resulting “program” is then a sequence of vectors (joint coordinates) plus activation signals from external equipment used to command the robot to perform the same motion at a later stage. Alternatively, the off-line programming is a language-oriented programming technique, similar to conventional computer programming, which enables the programs to be written in high- level or low-level language – sometimes with simulation and 3D graphical support (Nilsson, 1996; Nilsson and Johansson, 1999). Task-level programming In task-level programming, the desired goals of the task are commanded directly instead of specifying Industrial Robot: An International Journal Volume 31 · Number 3 · 2004 · pp. 284–296 q Emerald Group Publishing Limited · ISSN 0143-991X DOI 10.1108/01439910410532369 The Lund Research Program in Autonomous Robotics was sponsored by the Swedish National Board for Industrial and Technical Development (NUTEK) and the Swedish Agency for Innovation Systems (VINNOVA). Our research was partly funded by the EC 5th Framework Growth Project GRDI-2000-25135 Autofett. The authors are grateful to the friendly system-integration atmosphere and support at Kranendonk Production Systems BV in Holland, and to the support within our organizations, including Ha ˚kan Brantmark, Peter Eriksson, Hui Zhang, JianJun Wang and ZhongXue Ganat ABB, and many others. 284