The goal of my project was to apply mechanical dynamics to robotic designs – specifically those created through rapid-fabrication software. To do this, I read papers comparing a variety of physics engines in different situations. The papers discussed which situation each engine handled best. This information was used to decide what engines I should use for my project.
After deciding on engines, I apply the chosen engines to different designs depending on what the robot is supposed to do. This is done by breaking a design into its components and applying virtual sensors in key locations to take measurements. For example, a sensor that measures forces on a joint can help me decide what electronics I should be using to move that joint.
Looking forward, I would like to be able to automate this process. That way, there is no need to manually break a design into components and manually insert sensors, since this process can be very tedious for robots with many parts.
This project was my first major experience in research, so the entire time spent on it was a learning experience. The most surprising thing for me was how slow the process can sometimes be. Since there is no answer key when researching, much of any project is trying out ideas that don’t work until you come across what you really needed. Occasionally I stumbled upon successive leaps of progress, but often I had to attempt different methods until one of them turned out to be what was needed.
I very much enjoyed the research experience. Using what I have learned at Penn in a context besides problem sets and timed exams was very satisfying. It has given me a greater appreciation for the things I have learned thus far, and has made me more enthusiastic about what is left.