search

nozaripo / Simscape_DMP_Biped

This repository wraps up all the codes and a biped model in Simscape alongside the libraries used in it for the purpose of a DMP-based optimization of bipedal locomotion.
0 stars 0 forks source link

Questions related to startupDemo.m and robotParameters.m #3

Open jmfinley2 opened 5 years ago

jmfinley2 commented 5 years ago

Hi Pouria,

I started reviewing some of the code for the project and thought I'd use this feature of Github to share my comments/questions with you. Can you update the files noted below to address each of the bullet points?

In startupDemo.m

  1. Is line 22 necessary? This folder didn't exist when I tried to run the code on my computer.

In robotParameters.m

  1. Change the foot density to 1000 kg/m^3 and add units to all parameters
  2. Are world_damping and world_rot_damping used in the model?
  3. Are the contact parameters consistent with values used previously in the literature?
  4. Please add a brief description of each variable and corresponding units as comments adjacent to each of the parameters in robotParameters.m
  5. Are all of the parameters defined within this file used within the current model?

Simulink model

  1. Remove all components from the Simulink model that are not attached to anything. It may not make a huge difference in performance, but it should reduce the amount of memory needed to run the simulation.

In OptimizeRobotMotion

  1. Why do you "clear all" at the beginning of this file? Do you no longer need the parameters that were created in robotParameters?
nozaripo commented 5 years ago

Hi Dr. Finley,

Here are my responses to your comments:

startupDemo.m

  1. Line 22 is not necessary. I forgot to comment this line out alongside many others.

robotParameters.m

  1. These parameters are not used in my recent model. They actually are the ones used in the model developed by Sebastian Castro. I added units to the parameters.

  2. The damping coefficients are all set to zero in the model. But I hadn't modified that in the code.

  3. I remember to have checked it before with a study but yet made some changes to the parameters. I think I need to double-check these parameters. (Or alternatively, do a ground interaction optimization to obtain them. The way I think I can do it is to damp the torso's rotation and just actuate it with the demonstration so we don't need to worry about the stability. And by also using the actual GRF data we can obtain the values of the parameters through a simple optimization.)

  4. I did add descriptions and units for each of the parameters to the file.

  5. Not all the parameters defined in the file are used in the model. But they could be depending on what type of actuation (position/torque control) you want to have. So I rathered not commenting those last parameters out.

Simulink model

  1. I removed all the blocks that weren't connected to anything in the model.

OptimizeRobotMotion_openLoop.m

  1. Once you call a cost or a constraint function in the optimization the workspace becomes inaccessible to the content of those files, so I have to run the "robotParameters" in each of those files anyway. But still, a better way to do that I guess is to define the parameters as a structure in the main file and pass them to each of the cost and constraint files.