Non-linear system identification and state estimation in a pneumatic based soft continuum robot

Sensory data play a significant role in the control of robots. While soft robots are promising for operation in unstructured environments, it may be difficult to sensorize them due to their inherent softness. One way to overcome this challenge is to use an observer/filter to estimate the variables (states) that would have been measured by those sensors. Nevertheless, applying an observer/filter to a soft robot introduces the challenge of requiring an analytical model of these highly nonlinear systems. In this paper, we develop a framework based on nonlinear system identification and state estimation to estimate the curvature angle of a pneumatic-based tentacle soft robot. We model the tentacle using the wavelet/sigmoid network, and use an Extended Kalman Filter (EKF) to estimate the curvature and verify the estimate using camera vision. The results show that EKF can estimate the curvature angle at a low error, even when the identified system model is not accurate and the sensor measurement is noisy.