Nonlinear MIMO Controller for Generalized Soft Robots

by Sarah Aguasvivas Manzano

• Step 1: Learn the forward dynamics
• Step 2: Train the feedforward MLP
• Step 3: Configure Controller
• Step 4: Run!

To cite this paper:

We encourage our users to cite the publication that we prepared for this work. Please find it at this direction:

@misc{manzano2021highbandwidth,
      title={High-bandwidth nonlinear control for soft actuators with recursive network models}, 
      author={Sarah Aguasvivas Manzano and Patricia Xu and Khoi Ly and Robert Shepherd and Nikolaus Correll},
      year={2021},
      eprint={2101.01139},
      archivePrefix={arXiv},
      primaryClass={cs.RO}
}

The model: The model is a hierarchical, recursive, neural network model that is capable of doing forward predictions by feeding back to itself the predictions from previous timesteps. Here is the output of the system.