Simulation and execution of autonomous robot systems

Sammanfattning: Traditional off-line programming is not enough in order to use the full potential of virtual models and simulation system in industrial robot applications. The interface between off-line programming system and the robot controller is today restricted to program transfer. In this way a large amount of information is lost since typical robot programming languages is rather limited. If the virtual model could be accessed during the execution of the robot task, intelligent decisions could be made to avoid collisions, singularities, exceeding joint limits and poor process quality despite changes in the state of the robot workcell which not was anticipated in advance when the robot task was planned. Instead of a simple feedback loop to the robot movement, the virtual model is continuously updated so the new information together with previous knowledge is accumulated in a common format. High level re-planning of the robot task can then be automatically performed. An experimental system has been created that take advantage of the suggested approach. A commercial robot simulation system has been extended with new functionality for e.g. real time control of a physical robot, sensor feedback with automatic on-line calibration of the virtual model, robot tasks described in a high level language that can adapt to changes in the robot workcell and continuously supervision of the robot system to handle limitations without failing the task process. A modified industrial robot system have been used that allows low level joint control. Gas Metal Arc-welding (GMA) have been the target process during experiments. It is a complex process that needs continuous corrections in a delicate cooperation between robot and process equipment. A seam-tracker have mainly been used as sensor to illustrate the possibility to close the control loop in the virtual system.