Negative activation for Thelen2003 muscle model
Posted: Tue Mar 26, 2024 2:10 am
Dear Moderators,
I have already seen similar questions in the forum but I do not think that there is an answer yet.
I am using Forward Dynamics for the Arm26 model with the Thelen2003 muscle model.
I realized that when using the "Simulate" button (no excitation, initial activation at the default value of 0.05 and default pose), the activation of all the muscles get negative. It appears that they first have (as expected) an exponential decay and then a linear decay (see 1st image).
I was wondering what these negative values of activation mean. It looks like they introduce some damping in the system? The integration is very fast and it looks like the energy of the system is reduced.
I tried replacing the Thelen2003 muscle model by the DeGrooteFregly2016 muscle model with no damping. For this muscle model, the activation of all the muscles exponentially decays to 0 as expected (see 2nd image) and the integration of the equation of motion do not find a solution when running the "Simulate" button for the default simulation time of 5 seconds (probably because of the stiff differential equations due to tendons and that there is no damping).
Could you give some explanations of how the Thelen2003 muscle deals with negative activation?
Thank you for your reply,
Benjamin Moreno
I have already seen similar questions in the forum but I do not think that there is an answer yet.
I am using Forward Dynamics for the Arm26 model with the Thelen2003 muscle model.
I realized that when using the "Simulate" button (no excitation, initial activation at the default value of 0.05 and default pose), the activation of all the muscles get negative. It appears that they first have (as expected) an exponential decay and then a linear decay (see 1st image).
I was wondering what these negative values of activation mean. It looks like they introduce some damping in the system? The integration is very fast and it looks like the energy of the system is reduced.
I tried replacing the Thelen2003 muscle model by the DeGrooteFregly2016 muscle model with no damping. For this muscle model, the activation of all the muscles exponentially decays to 0 as expected (see 2nd image) and the integration of the equation of motion do not find a solution when running the "Simulate" button for the default simulation time of 5 seconds (probably because of the stiff differential equations due to tendons and that there is no damping).
Could you give some explanations of how the Thelen2003 muscle deals with negative activation?
Thank you for your reply,
Benjamin Moreno
- Activation when hitting the "Simulate" button with default parameters for the Arm26 model and Thelen2003(the activation is superposed for all the muscles)
- Activation_simulate_Arm26.png (25.88 KiB) Viewed 192 times
- Activation when hitting the "Simulate" button with default parameters for the Arm26 model and DeGrooteFregly2016 muscle (the activation is superposed for all the muscles)
- Activation_simulate_Arm26_deGroote.png (23.01 KiB) Viewed 192 times