The Construction And Deformation Of3D Modeling For The Flexible Humanoid Multi-airbag Robot Based On Human Characteristics

The study of flexible humanoid multi-airbag robot is a new research topic. The content of this research is to construct a deformable multi-airbag robot model based on the data collected by the3D scanner. Techniques included in this research are:3D point cloud data pretreatment, construction and deformation of multi-airbag robot. Existing1human modeling and deformation techniques can not be directly used for multi-airbag robot modeling. The airbags distribution must be decided before robot modeling, which is different from human modeling. The content and innovation of this research are listed as follows:(1) Apply the adaptive arc segmentation method to airbags distribution and the construction of robot modeling.(2) Deform the robot model by shaping the cross-section rings through changing the radius of filling ellipses.First of all, the3D laser scanner is used to obtain the discrete3D point data of human model. The3D point cloud are pretreated by smoothing, de-noising, feature detection and simplification to get more accurate and suitable data.The model of flexible humanoid multi-airbag robot is constructed based on the3D point data collected by the3D scanner. The adaptive arc segmentation method is suitable for airbags distribution and the parametric surfaces are applied for the3D robot reconstruction. By setting the piecewise quadratic field function and energy equation, the adaptive method is firstly proposed to approximate a silhouette curve with a set of circle or ellipse.Then, the adaptive method is used to decide arc segmentation for the feature curves of human body, which is related to the distribution of airbags. The single-line parametric surface is made by mapping sub-arcs between adjacent feature curves. The resulting robot model is finally constructed by splicing those monolithic surfaces.Lastly, the deformable robot model can simulate airbag inflator process by adjusting shape parameters of fitting ellipses which approximate the feature curves. Since the stretching of the cross-section ring is associated with the shape parameters of those fitting ellipses. Therefore, the back propagation neural network can be used to learn the relationship between the shape of cross-sectional rings and the geometric parameters of filling ellipses. Finally, the resulting robot model can simulate airbag inflator process and show the deformation of flexible humanoid multi-airbag robot.