Structural Self-Similarity Framework for Virtual Human’s Whole Posture Generation

There are many virtual human joints and degrees of freedom involved in man–machine simulation; however, the commonly used man–machine simulation software cannot provide enough constraints for the whole posture solution of virtual human. This study proposes a novel framework for whole posture calculation and generation. First, a new typical posture database is created. Based on these typical postures, virtual human approximates whole posture that is generated according to the new posture generation conditions and selection rules. Second, a novel method for accurate calculation of partial postures is presented. This method can calculate and adjust the partial posture of the virtual human. It can also achieve the whole posture generation to design the calculation and generation process. Then, this study proposes a meaningful calculation method of anthropomorphism which can evaluate the generated posture accurately. The experimental process is verified by abundant simulation examples compared with three state-of-the-art methods. It turns out that the novel framework proposed can reduce interaction effort and improve the effectiveness of posture generation. Meanwhile, the evaluation of anthropomorphism can achieve 0.91 to get Level 1 which indicates that the generated posture is beneficial to simulation requirements.