This is the final capstone project for ME449 Robot Manipulation class at Northwestern University. The problem was to control and simulate a mobile manipulator where the manipulator carries the box from the initial to the final configuration. The software used here is MATLAB.
- NextState - The next state function takes in the current configuration of the robot and returns an configuration after a timestamp
- TrajectoryGeneratory - The trajectory generator function gives a desired trajectory of the end effectory
- FeedbackControl - The feedback control function is a controller which takes in the current configuration and corrects it to give a better path
There are three runs in this repository. First one where the error is minimized, second where there is an overshoot and third where we change the location of the box.
The main script that runs all the functions together is final_project.m. Inside the script, there are comments for the configuration changes for all the possible values. The main scripts solve for a desired trajectory and then uses that as a reference to calculate an actual trajectory from the current configuration of the robot. The software uses feedforward + Proportional +Integral error correction for the current path to reach a zero steady state error, so that the robot can perform the task with no error.
The feedforward control plus PI parameters are as follows: Kp = 0.7 Ki = 0.0001 Tsc_initial = [1 0 0 1; 0 1 0 0; 0 0 1 0.025; 0 0 0 1] Tsc_final = [0 1 0 0; -1 0 0 -1; 0 0 1 0.025; 0 0 0 1]
The graph for the best run:
The video of the best run is below:
best.mp4
The feedforward control plus PI parameters are as follows: Kp = 3.5 Ki = 0.5 Tsc_initial = [1 0 0 1; 0 1 0 0; 0 0 1 0.025; 0 0 0 1] Tsc_final = [0 1 0 0; -1 0 0 -1; 0 0 1 0.025; 0 0 0 1]
The graph for the overshooot run:
The video of the overshoot run is below:
overshoot.mp4
The feedforward control plus PI parameters are as follows: Kp = 1.9 Ki = 0.001 Tsc_initial = [1 0 0 1; 0 1 0 0; 0 0 1 0.025; 0 0 0 1] Tsc_final = [0 1 0 1; -1 0 0 -1; 0 0 1 0.025; 0 0 0 1]
The graph for the new position run:
The video of the new position run is below: