I currently work with a small team on research with robotic heavy construction equipment. We recently finished a robotic retrofit of a piece of heavy equipment, where I focused on proprioception and compute integration. This included setting up a new software stack in ROS2, writing device drivers for ROS2_control, setting up a new robot model, tuning controllers, integrating in-house autonomous blade controllers, reverse engineering some vehicle communications, and of course making tons of electrical harnesses. Really, it was a little bit (a lot) of everything, all the time.

We're currently focused on training / finetuning models to enable greater autonomy on our platform. Well, it would be nice to just finetune an existing generalist policy / foundation model, but...

Previously, I've worked on localization and computer vision for the same team (mostly on a different platform). This included working with loads of GIS data, both as inputs for localization and as outputs from mapping subsystems.

Additionally, I've worked on a historical preservation project recording the architectural, historical, and political significance of cold war era launch sites in Vandenberg Space Force Base. (This included site visits!)

I have also conducted some exploratory urban geography research, looking into the interplay between urban highway bridge infrastructure and the surrounding urban population (urban morphology stuff).