Abstract
Soft robots can be limited in applications when their pose in space is difficult to estimate, particularly when actuated by smart thermoelectric materials with difficult-to-model mechanics. This paper presents a comparative study of approximations and simplifications that could make pose estimation computationally practical in real-world settings. To do so, this article represents a planar soft robot arm as a discretized many-link rigid arm, mapping material stiffness and actuator states to torques at the robot’s joints. Four different sets of assumptions are proposed for these mappings, varying in how stiffness is distributed throughout the arm, as well the linearity and/or hystersis of the actuator torques. We demonstrate how to calibrate each model from experimental data in a soft arm powered by shape memory alloy (SMA) wires that contract via Joule heating. Then, we perform hardware tests to predict the robot’s pose in open-loop, using only actuator temperature, and compare model performance under each simplifying assumption. Results show that adding both nonlinearity and hystersis to the actuator model improves the pose predictions, and that open challenges remain in calibrating material parameters. This study provides a platform and an initial result toward real-time pose estimation of these robots.
