I understand that most of us seek productivity, aiming to produce acceptable results (without the need for absolute accuracy) within the shortest time frame. AI-powered models and algorithms, such as Reinforcement Learning or Deep Learning (the technology that revolutionized image processing), can effectively achieve this goal. Unfortunately, MATLAB lacks a dedicated function to process images and return the transfer function of the system.
You might want to try using Simscape, a MATLAB-based graphical programming environment, to model the system based on your robot finger image. Then, use Model Linearizer to perform linear analysis of nonlinear robot models.
Else, derive the transder function from first principles.
