There are two mainstream methods for motion capture: visual capture can capture complex 3D geometric deformations, but relies on expensive optical equipment and suffers from line-of-sight occlusion; IMU-based methods are simple but difficult to capture subtle 3D deformations. In order to solve this problem, researchers from Zhejiang University proposed a configurable self-aware IMU sensor network, which solves the data sparsity problem and the deployment problem of sensor nodes in IMU sensor networks.
In the proposed method, a kinematic chain model based on a four-bar structure is used to describe the minimum deformation process for 3D deformation; three geometric priors obtained from initial shape, material properties and motion characteristics are also introduced to help the kinematic chain model reconstruct the deformation and overcome the data sparsity problem. In addition, in order to further improve the accuracy of deformation capture, a fabrication method to customize the 3D sensor network for different objects is proposed. Experimental results show that this method performs very well.