This paper was the culmination of my final project for MEAM 517: Control & Optimization with Applications in Robotics, and I worked on it with Tommy Mulroy. The prompt was to spend half of a semester on an open-ended research project of our choosing.
Since most of what we had learned about in class had dealt with ideal situations that directly followed the laws of Newtonian physics, we decided to investigate robust controls approaches and techniques. We ended up discovering DIRTREL - Direct Transcription with Ellipsoidal Disturbances, a 2017 paper from MIT, that claims to produce robust trajectories for multi-DoF systems. As detailed in the paper below, we replicated the results of the paper and performed further analysis of the DIRTREL algorithm with respect to speed, robustness, and trackability. The code for this project was done in MATLAB with the SNOPT nonlinear optimization solver.
The code is available in a repository here. The code works in Linux and Windows environments, but is currently not compatible with MacOS.
Click here to open in a new tab.