Research On Operating Environment Visual Reconstruction Of Virtual Force Guidance For Telerobotic

The presence-based teleoperation robot played an increasingly role on theoperating under extreme conditions such as high temperature, high pressure, strongradiation and suffocation with the reasonable development and utilization of undersea,underground resources and outer space. The pre-modeling form of the3-Dreconstruction for the objects in the remote environment, which is usually used in theexisting virtual reality assisted teleoperation system, can not guarantee the consistencyof the virtual scene with the remote site. There is also a large deviation between thereconstructed and true position and attitude of the object when the environment ischanged. It is a huge difficulty for the teleoperation task that the collision detectionmodel in the virtual environment can not reflect the real situation of the remoteworking site. In addition, based on the existing tele-operation bilateral servo controlmethod, the reaction force perceived by the operator is past force. If it was applied inhigh-speed or heavy loaded situation, the irreparable damage which is caused by theinstant impact to the operating environment from the robot may be occurred beforethe reaction force has been perceived by the operator.In this paper, combined with the JiLin province science and technologydevelopment plan item-Research on Intelligent Control of Virtual Force Guidancefor Industrial Robots, and the Ph.D. programs foundation of ministry of education ofChina-Research on Master-slave Robot Bilateral Servo Control with Virtual ForceGuidance Based Stereo Vision. Based on the image feature extraction and the stereomatching calculation, the3-D virtual simulation environment of the tele-operationrobot which is guided and controlled by virtual force has been reconstructed by the binocular vision technology in real time. Further more, to improve the operatingefficiency and local autonomy of the robot and reduce the proficiency requirement tothe operator, the manipulation of the operator is guided through the virtual force.The main tasks completed by this paper are as follows:1. The inner and external parameters are solved by Zhang-plane calibrationmethod after the binocular camera is calibrated, then the relative posturerelationship between the cameras is determined.2. The two images obtained by the camera are pretreated separately todetermine their edge information, then the feature points of the object in theimage are extracted by Harris algorithm.3. After the window-based gray matching and the homography matrix-basedguide matching had been combined, the mismatching points in the stereomatching process of the image are eliminated by the second matching.4. After the3-D dense points cloud of the object in tele-operation environmenthad been calculated, the3-D geometric model of the object is reconstructedby the feature constraints-based Power Crust algorithm in real time. And thevirtual simulation environment of robot which is guided and controlled byvirtual force is built on the basis of the works above.