A Quick Interview with Parth Parekh
by: Isabella Richardson
It’s Fall 2021, Parth Parekh and his roommates Jake Imyak and fellow UWRT member Cedric McGuire are in the midst of a difficult, yet interesting class on Natural Language Processing (NLP) at OSU. NLP is the ability of a computer program to understand human language, as it is colloquially spoken and written. Their goal was to see just how far they could stretch Recurrent Neural Networks (RNNs).
Being apart of UWRT, Parth was able to think outside of the box and apply their project to another level. The group saw the benefit of being able to quickly translate sentences into commands our robot can easily understand. Parekh states, “While our classmates were only doing basic applications of RNNs to translation, we were able to synthesize something new and very useful.” That useful thing Parth is talking about, is enabling the robot to build task code directly from natural language prompts. His connection with UWRT was imperative for this project to succeed, as the trio needed to have a base level of understanding on what maintainable and deployable code looks like for our robots.
Parekh believes that these findings can make an impact on unique ways to interact with robots. “NLP has gone a long way in determining intents, but this research helps connect that intent to an action that an agent (our robot in this case) can take.”