Spinal controller

The current git repository contains the necessary files to perform optimizations with the spinal controller. In order to run these files, it is necessary to the SCONE version found in https://gitlab.com/simgait/SCONE.git. The solutions contained in this repository have been obtained with Ubuntu 18.04.

Repository content

SCONE scenario scripts containing experiments carried using the spinal controller.

spinal_controller_main.scone: optimied gait using the spinal controller

spinal_controller_imitate.scone: optimized from a reference solution using imitation.

ong_main.scone: optimized gait with the controller proposed by Ong et al. (2019).

delay: colder containing the neural delay for the muscles modeled.

import:

• initializations: folder containing par files used to initializze the optimization and files containing initial states of the model.
• measures: folder containing scone files with the implementation of the cost function to optimize.
• models: folder containing osim models.
• CMA_settings.scone: scone file defining optimizer's parameters.
• ModelAndIntegrationSettings.scone: scone file defining teh integration method and accuracy.

ong_controller:

• ong_2019.scone: implementation of Ong's controller (2019).
• ong_2019.scone: implementation of the balance controller from Ong et al. (2019).

scripts: folder containing python scripts to generate graphs and perform the correlation analysis

• spinal_modulation: folder containing the experiments and scripts used for the correlation analysis.
• plot_neural_input.py: generate a graph checking the contribution of CPGs, reflexes, and balance to the motoneuron activity.
• plot_sim.py: plot kinematics, GRFs, and muscle activation.
• utils.py: python file where we implemented the functions used in the other scripts.

spinal_controller:

• spinal_cntroller.scone: controller implementation.
• spinal_cntroller_no_cpgs.scone: controller implementation without CPGs parameters.
• spinal_disable_controller: spinal controller where all parameters are set to 0 (necessary to obtain the imitation solution).

stastes: folder containing initial states to imitate.

Instruction to perform the three optmization steps

If you want to try obtaining a new solution from scratch with the proposed controller. However, these step are not necessary if you use the solution already provided (import/initializations/spinal.par).

1. Optimize the scenario ong_main.scone to obtain a solution to imitate from Ong's controller (2019).
2. Save the evaluated storage solution (.sto) into the state folder.
3. In spinal_controller_imitate.scone, replace the .sto file with the solution you obtained at line 9 and 41 (file = states/your_file_name.sto)
4. Optimize the scenario spinal_controller_imitate.scone.
5. Save the parameter file (.par) representing the solution where the optimization converged into the folder import/initializations.
6. In spinal_controller.scone, use the parameter file obtained from the previous step to initialize the optimiztion (init_file = import/initializations/your_file_imitation_name.par), comment "<< import/measures/measure.scone >>" at line 102 and uncomment "<< import/measures/measure_mimic_unstable.scone >>" at line 103.
7. Optimize the scenario spinal_controller.scone to obtain a first stable solution.
8. Save the parameter file representing your stable solution into the folder import/initializations.
9. In spinal_controller.scone, use the parameter file obtained from the previous step to initialize the optimiztion (init_file = import/initializations/your_file_stable_name.par), comment "<< import/measures/measure_mimic_unstable.scone >>" at line 103 and uncomment "<< import/measures/measure.scone >>" at line 102.
10. Optimize again the scenario spinal_controller.scone to obtain your final solution.

Once a good solution is found, it can be used as initialization to perform other experimets (e.g. speed modulation).