基于云模型的虚拟人摇头动作不确定性控制*

以虚拟人摇头表达否定性信息为例，设计了最大摇头角否定云、摇头次数否定云和摇头速度否定云。分别运用正向云生成算法产生云滴，获得摇头动作序列的最大摇头角度、摇头次数和摇头速度。三个控制参数均以各自的期望值为中心正态分布。仿真实验表明，所提出的基于云模型的虚拟人摇头控制算法能够产生呈规律性差异的摇头动作序列控制曲线。给出了三类云模型数字特征的建议值。

Uncertainty control of head shaking for virtual human based on cloud model

Currently, certainty methods are adopted to synthesize body actions of virtual human or humanoid robot. However, the process people expressing deep information such as emotion, motivation, and mental state, etc. through body languages, has uncertainty characteristic of fuzziness and randomness. Cloud theory can integrate the two uncertainties. Taking the example of expressing negative information with head shaking of virtual human, three cloud models are designed, which are the maximal head shaking angle negative cloud, the head shaking times negative cloud, and the head shaking speed negative cloud. Their drops are generated by respective positive cloud generation algorithm, to obtain maximal head shaking angle, head shaking times, and head shaking speed. The three control parameters distribute normally around their respective expected value. Results of simulation experiment show that the proposed cloud-based head shaking algorithm can generate control curves of head shaking action serials with disciplinary differences. Feature values of the three kinds of cloud models are proposed.