Analysis of skin flow field of human face

Although facial expression is known to the one of important communication channel, deformation control of robotic facial soft skin is difficult and therefore facial expressions by android-type robots are poor, unreadable, or unnatural. In my project, measurements and analyses of humans’ rich facial skin deformation are conducted to find the design methodology to make robot facial expressions more rich, expressive, and natural.

Principal component analysis

3D flow vectors of 200 measurement points on humans face skin around the mouth were measured for 16 patterns of facial deformation. Principle component analysis on superposed vectors in each measurement point that each facial point moves along the almost only one direction and complex human expressions are realized by combinations of different flow lengths at each skin point. This finding is beneficial in facial robot design because its design can be more effective and efficient based on the finding. This finding will be submitted to an internal journal paper after further advanced analyses.