虚拟柔性体实时形变仿真模型研究

摘要

To improve the stability, continuity and accuracy of flexible deformation and computational efficiency in haptic rendering. A novel approach based on spherical harmonic (SH) representations to simulate the virtual deformation of soft objects is proposed. The object surface is represented by SH, which features of orthonormality, rotational invariance, and multi- resolution. Given the density, Young's modulus, and Poisson ratio of the object and the applied force, the deformed SH models are estimated by a radial basis function (RBF) based neural network model, which is trained with Least mean-square error. The simulation results indicate that the deformation could be accurately calculated using the proposed approach and the computational time is acceptable for real-time applications in medical education and training.%为了在虚拟现实柔性体力触觉交互研究中得到稳定、连续、真实的力触感,提出一种基于球面调和函数表达的虚拟柔性体实时形变仿真模型,利用球面调和函数的正交归一、旋转不变、多尺度等特性实现物体的快速准确表达.在变形体的密度、杨氏模量、泊松比等参数已知的情况下,基于径向基函数神经网络模型预测柔性体受力形变后的SH模型.仿真结果表明,该方法不仅可以准确表达柔性体的实时形变,而且使得基于SH表达的柔性体形变的视觉刷新描述与柔性体反馈力的触觉刷新描述同步,从而满足虚拟手术仿真训练等虚拟柔性体力触觉交互研究要求.