Create Fully Tuned Simulink Controller

[JuanR-Git]

The goal of this task is to apply our system's state space representation into Simulink. This model will take our mathematical expression and tune the input value base on the feedback.

The first objective for this task is to create a simple PID controller design using the best tuning method determined from #3 and #4. The requirements will be the same as our previous controller, this being:

• Overshoot <= 10%
• Settling Time <= 50ms

The second objective for this task is determining and applying our discretization method in Simulink. This is necessary to accurately model the hardware delay in our system (Sampling Time). By the end of this task we should have the following:

• Calculated sampling time based on our hardware
• Documented decision matrix with best discretization strategy (ZOH, discrete state space, etc.)
• Simulink Implementation

As a Bonus Task I would like you guys to explore more robust control strategies. This will account for discrepancies in our state space calculation cause by estimates/assumptions/approximations.

Keep up the good work!

[deshpr3]

Created a preliminary simulink model for the knee using the torque control strategy in this document. Section 4 gives a brief explanation of the simulink model.

Next steps: If we decide to use this model, we need to move the info from the word document to the Milestone 2 google doc.

[JuanR-Git]

Working Simulink model can be found on the google drive here: https://drive.google.com/file/d/1DBGKNOR8Nrl1lvRG4kl5fVLGbzIcxcvo/view?usp=drive_link

[deshpr3]

I've updated the Simulink model to reflect the new dynamic equation for the torque controller. I've implemented the torque controller as a MATLAB function so its easier for C++ converstion.
Currently, the AI-generated knee angle and hip angles are represented as sigmoid functions.

Running tourqe_output_analysis.m will generate graphs for the hip and knee torque responses.

Updated model is uploaded here: https://drive.google.com/drive/u/0/folders/1FuHS1YMNKVGHjlBGtPs4jIKYaXwsJBka @JuanR-Git Feel free to move it around.

Next steps:

• Replace the sigmoid functions with AI-generated knee and hip angles
• Consider whether to use MU-measured angles and velocities for PID control