During my most recent summer internship at OffWorld, a robotics company developing autonomous mining solutions for terrestrial and extraterrestrial applications, I worked most closely with their flagship surveying robot shown. During the course of the internship I closely collaborated with software engineers working in the perception and localization teams to improve sensing capabilities and autonomous navigational accuracy. Through the development of new mounting hardware and the selection of new sensors, localizational error was reduced by 198%.

The image pictured to the right is of an early prototype “mast” made by me from machined aluminum extrusion and 3D-printed hardware to improve localization by relocating the lidar placed atop the mast. This mast is also capable of housing several other sensors such as depth cameras for even greater perception capabilities . Dampening material between the mast and the chassis reduce vibrations harmful to the operation of both the lidar and the depth cameras. The retrofitting of a retired robotic test platform was another task taken on by me over the summer. This test platform was meant to replicate the company’s flagship mining robot to aid in the development of multi-agent simultaneous localization and mapping (SLAM). This overhaul consisted of newly machined mounting platforms for electronics and sensors, restoration of the electrical system providing power to the drive motors and electronics, and modeling of this test platform in SOLIDWORKS which was later translated into a usable URDF.

I took ownership over a third robotic platform intended to act as a mobile testbench for rapid perception and SLAM software iteration using a Clearpath Robotics Husky as a base. The Husky served as a platform onto which I designed mounting and structural elements to facilitate sensor integration on a mobile testbench.