Robot Dynamic Control Lab at USC

This research focuses on developing novel control algorithms for achieving extremely agile and robust locomotion on dynamic robotic systems. My approach lies at the intersection of nonlinear control, optimal control, and trajectory optimization.

key skills:

Matlab

Simulink

Controller Design (LQR, QP, MPC, PD, etc.)

Close-loop Diagram

This is a simple diagram that represents the control logic behind a quadruped robot.

Equation of Motion

Before we design controllers for a robot, we need to find the dynamic of the robot first. In the figure below, we can see that there are 7 parameters for a quadruped robot in 2D.

We will use the Lagrangian Equation to construct the dynamic. Once we found all the parameters of the dynamic, we can apply these parameters to Matlab, Python, or any other programing language you are familiar with to find the equation of motion. It is always better to use symbolic parameters at first. Then, plug the initial condition into the equation.

The figure below is an example code of the EOM of the quadruped robot in 2D.

Now, we can construct a balance controller based on this EOM for the quadruped robot. The figure below is an example code of the MPC balance controller of the quadruped robot in 2D.

Here are the simulation result, and data analysis.