Research On Collision Detection Of Moule Motion Simulation In Augmented Reality Enviroment

The virtual objects are simply overlaid on the real video in most AR systems. This results in penetration between the virtual and the real. Because of the mistake, the user who performs the task for a long time may get eye fatigue and operational difficulties caused by the visual disorder.. To improve the sense of reality in computer simulation and the immersion in AR, the problem of collision detection between the real and the virtual is inevitable. Only when the problem was solved, can the penetration between the real and the virtual be prevented.First, the paper took the motion simulation of robotics for example, and set up the AR environment for the simulation. The experiment took the robotics as the virtual object, and took the objects acquired in the video as the real objects of the scene. The paper compared the advantages and disadvantages of VR and AR, then, set up system control framework based MFC, and carried out the robotics modeling & motion simulation on the basis of OSG/osgART.Next, the paper studied the image-based three-dimensional reconstruction, and got the depth of the objects obtained in the video, then, construct its bounding box with the depth obtained. This provided an effective way for the collision detection between the real and the virtual. There are several key technologies including camera calibration, feature extraction, stereo matching and depth determination.Last, the paper studied the current collision detection algorithms, and raised the way of its own. The paper proposed to apply the image-based modeling technology to get the three-dimensional information of the scene, then, computed the bounding box of the actual objects with the depth information acquired previously. In this way, the collision detection between the real and the virtual was changed into the collision detection between the virtual objects. The paper introduced the algorithm of "AABB bounding box—k-D tree" into the simulation system. The method combined bounding box technology and space subdivision technology, and took advantages of both. Through this method, the efficiency of collision detection was greatly improved, so was the performance of the overall system. The results of the paper can be used the research areas such as manipulator path planning, teleoperation, robot motion simulation etc. The method presented in the paper—applying image-based three-dimensional reconstruction technique into the collision detection—is effective proved by the experiment.