Abstract
Integrated motion information management for use in highly accurate sensor-based robot systems that increases the precision and quality of robotic tasks is described. In this method, the sources of the recognition errors are easily identified by introducing the error-model using the information obtained by the proximity sensor as well as the information about the robot motion acquired by commanded simple testing motions, and these identified error-sources are used to reduce the recogniton errors. Motion reliability is determined based on the robot-motion information, and this reliability is used to reduce the influence of the recognition errors to the whole sensor-based robot system. As a result of this tuning the sensor-based robot system is set up to be close to optimal as possible. Experimental results from a typical visual path-tracking task showed that the tracking error is significantly reduced by using this motion information management system.
