Stairclimbing Tumbller
In a team of five we integrated a variety of control algorithms in order to enable the Tumbller to climb a step while balancing a load.
I worked on developing and integrating the LQR algorithm on the Tumbller to balance and climb stairs. We used an inverted pendulum to model the Tumbller’s dynamics and then linearized the non-linear model. I computed the gains using Matlab and then adjusted them based on the Simulink simulation results and then the experimental results.
In order to enhance performance when the Tumbller was nearing unstable positions, I implemented a gain scheduler. With the LQR algorithm the Tumbller was able to climb a step that was three-fourths of an inch with a weight attached at the end with ease.
The simulation model and LQR controller were developed using Matlab and Simulink, but was integrated into the Tumbller using an Ardiuno and .ino files written in C++.