Authors: Sarah Aguasvivas Manzano, Patricia Xu, Khoi Ly, Robert Shepherd and Nikolaus Correll

Submission: International Symposium of Experimental Robotics, 2021

Abstract

We present a high-bandwidth, lightweight, and nonlinear output tracking technique for soft actuators that combines parsimonious recursive layers for forward output predictions and online optimization using Newton-Raphson. This technique allows for reduced model sizes and increased control loop frequencies when compared with conventional RNN models. Experimental results of this controller prototype on a single soft actuator with soft positional sensors indicate effective tracking of referenced spatial trajectories and rejection of mechanical and electromagnetic disturbances. These are evidenced by root mean squared path tracking errors (RMSE) of 1.8 mm using a fully connected (FC) substructure, 1.62 mm using a gated recurrent unit (GRU) and 2.11 mm using a long short term memory (LSTM) unit, all averaged over three tasks. Among these models, the highest flash memory requirement is 2.22 kB enabling co-location of controller and actuator.

Manuscript

The below paper is for the detailed description of the neural network architecture for both sensor modeling and controller implementation: