| Manufacturing process has still been a traditionally applied field for robot. The sustaining demands of batch production provides a stage for robot. In the future industrial application, the robot not only is required to go on a simple repetitive work but also integrated to manufacturing system being a programmable and reconstructed cell which is high-flexible, opening and good man-machine interaction performance. That situation asks for the whole advance of robot technique at present, in which, the robot teaching technique is an important component.Differ from the task being done by teleoperation, the typical tasks being finished by industrial robots are usual repetitive, and appear in known environment. They need to be finished by the robot teaching system. There are three types of robot teaching system now: teaching playback, off-line programming, and robot teaching based on virtual reality (VR). In which, robot teaching using VR technique is a new research direction in robotics. Although robot teaching study based on VR has gained some significant results, its practice application is limited. There exist some problems in current systems: â‘ effective supervisory control has not been seen; â‘¡ human operator's action skills have a direct influence on the teaching system. Considering these problems, we go on to study and develop the robot teaching system based on VR and supervisory control theory. A novel idea is proposed, that is, under the supervisory control, the merits of teaching playback and off-line programming are combined. Sensor information feedback at worksite is introduced to virtual environment. Geometry motion data are gained in on-line procession, and are planned in off-line stage so as to improve the flexibility, autonomy and reliability of the robot teaching system. At the same time, there are some good performance on reducing teaching load on human operator, simplifying complexity of robot teaching process and shortening teaching time. The main research of the paper is in the following:The modeling study of robot teaching system based on VRA modeling and development method of robot virtual teaching workcell based on Envision TR is studied. In the process of constructing VR environment, for MOTOHAN-K10s robot, we study how to build 3D geometry model, kinematics, relevant physical constrains of virtual object and collision detection between virtual objects in Envision TR environment. In order to make virtual environment reflect really and exactly the pose relationship in real world between robot motion and work object, we build mathematic computing modeling of robot teaching programming. If the relation got from the modeling is sent to virtual environment, a motion relationship the same as real environment can be formed in virtual environment between robot and work object.On the base of the kinematics error modeling, in order to get merits between teaching playback type and off-line programming, the model-based supervisory control method is introduced in virtual operation environment. As a result, robot teaching mode of man-machine sharing task world is formed. From these, some faults which are generated by robot teaching system that used master-slave operation and supervisory control with local autonomy are overcome. From software engineering requirement on the data and its encapsulation, task-oriented robot teaching programming is achieved by modular design. By these, event-driven robot teaching simulation and execution structure are built.2) Study on reflective motion planning method of robot teaching system For avoiding collision free complex path planning in 3D space, a desired robot teaching path is generated by man-machine interaction mode. According to the characteristics of robot teaching system with model-based supervisor control, we optimize the desired teaching path using genetic algorithm, so that make robot have the largest manipulability within the tolerance of the desired teaching path.We provide reflective motion planning method based on sensor feedback...
|
PDF Full Text Request