Motion Editing And Its Application In Virtual Environment

Virtual human is an important component of virtual environments and it plays an important role in enhancing the reality and immersion of the scene. The research on the motion control technique is one of the important tasks in the study of virtual human. The development of the motion control technique includes four phases: kinematics control, dynamic control, motion controller based control and motion capture technique. Among these, motion capture technique is a new technique that becomes popular in recent years. It can capture real human motion, so it has the characteristic of realism. But as to the technique itself, motion capture is the reconstruction of real motion, and the data it captured can be used only in specific environment. In order to make motion capture data re-usable so that it can be adapted to new environment and meet new requirements, we need to manipulate motion capture data. This paper mainly focused on the manipulation of motion capture data and it also discusses the path-planning algorithm in virtual environments.This thesis presents a footprint-editing algorithm based on analytical inverse kinematics(IK). We divide foot constraints into three categories: heel fixed, sole fixed and toe fixed. Analytical inverse kinematics is used to solve foot constraint. When solving IK, we use position constraint of the knee joint to decide the swivel angle. The result obtained in this way can preserve the continuousness and the detail signal of the original data. Footprint-editing algorithm eliminates foot glide phenomenon and improves the quality of motion capture data.The author implements a path synthesis algorithm that makes the virtual humanwalk along any path. First we pre-treat the three motion capture clips-walk-start,walk-loop and walk-end, so that they can concatenate with each other smoothly. In the pretreatment, we also adjust the root track of the virtual human to make each motion clip advance along the same line. The pretreatment also needs to guarantee the quality of the pretreated data. We use footprint-editing algorithm to solve the foot constraints for each motion clip. Concatenating the three pretreated motion clips together in a certain order generates walking animation along a specific path. Since most work is done in pretreatment and there is only a little edit work to do during walking time, the algorithm can be achieved in real time.Finally, we present a path-planning algorithm based on synthetic vision. The virtual human in the environment observes the scene through synthetic vision andupdates the scene octree that represents the memory of the virtual human. To accelerate the update of the octree, we present a new method: to divide the scene into the static part and the dynamic part. For the static part, we partition the scene in advance and for the dynamic part we need to update the memory dynamically. Then A algorithm is applied for shortest path search. In the final step, synthetic vision is employed again, together with parabolic fitting, to optimize the path. The algorithm can plan the walking path of the avatar quickly and the path generated by this algorithm is naturally looking.