A modification of the Instant NeRF algorithm in order to scan environments and allow for the easy replacement, placement, and /or removal of objects from a scan of the environment.
This can allow for easy use of augmented and virtual reality for applications ranging from enabling the integration of real world elements into a virtual world to allowing for virtual rearrangement of furniture.

A PID based control algorithm for a Autonomous Underwater Vehicle.
This algorithm allows for point to point navigation in a virtual environment even with unexpected conditions.
To achieve this we are used a pre-existing dynamic representation of an Autonomous Underwater Vehicle.

This project details the development of a modular robot for testing the dynamics of robotic self-assemblages
To achieve this we created a custom dynamic model of the robot for each individual robots
To test the dynamic model we created a Gazebo Simulation.
However, while we were successful in creating the dynamic model, we could not test the self-assemblages.

This project details the development of an Ensemble network using the GR-CNND, GG-CNN-RGB, and GR-CNN grasping experts.
The network was trained on the Cornell image dataset.
The accuracy of the experts was raised from 70% to 85% by utilizing this ensemble.

This project details the development of a ROS program with a team that used SLAM, AMCL, and A star to enable a Turtlebot 3 Burger to autonomously navigate, map, and localize itself inside a maze.
C-space, frontier detection, navigation, and frontier detection algorithms were made from scratch.
The nodes were programmed in Python, and the launch files were custom made.

This project details the creation of a computer vision controlled robotic arm system.
An event driven state machine was coded using MATLAB on a Linux workstation.
Test trials resulted in the robot being able to detect and localize balls before sorting them by color.
The system was then able to perform dynamic object tracking and unique object manipulation.

I worked with a team to create a final system implemented on three robots that covered: Romi, Tybalt/Juliet, and Mercutio/Friar Lawrence (two robots played two characters).
The team’s system heavily utilized the MQTT protocol for WiFi communication between the robots to coordinate cues and timing for the scenes in Romeo and Juliet.
Using WiFi helped replace other communication methods like using an IR emitter and receiver, which had multiple disadvantages, namely range and visibility.
The team was also able to utilize the AprilTag tracking with theOpenMV camera, which allowed for the robots to consistently follow each other.
The code of the system was organized in a way so each character had a “script” for the scene with the necessary tasks needed to complete.
All robots also shared a common program that covered general functions for all due to having the same drive base.

On a team I helped develop a robot mounted 4-bar mechanism to place panels on tilted surfaces at differing heights and locations by utilizing the Solidworks design suite.
Additionally we designed 2 different grippers and mounted them on 2 robots to complete the pick and place tasks.

On a team I helped program a prebuilt robot to retrieve and deploy cups to and from platforms of varying heights and locations using C++ and multiple sensors.
This was done completely autonomously, however, there was the option of remote control.