Previous Projects

The latter half of the work I did for my thesis at the University of Louisiana at Lafayette was focused on morphology optimization. Having been inspired by recent work in the field of learning based methods for mechanical optimization, myself and the PI in the CRAWLAB set out to discover a unique method, utilizing reinforcement learning (RL), to learn optimal morphologies for flexible-legged locomotive robots. The work required a deep understanding of state of the art RL methods in addition to control methods for non-rigid robots. The method discovered, was evaluated on the robot shown, and proved to be a possible solution towards finding mechanical parameters to that increased performance metrics such as running speed, jumping height and power usage. The parts that were optimized on the robots shown, are those that are flexible within the links.

The latter half of the work I did for my thesis at the University of Louisiana at Lafayette was focused on power efficient control of flexible-legged locomotive robots. Given the many recent advancements using reinforcement learning for continuous control tasks, several focused on learning the control of locomotive robots, we wanted to push the boundaries and figure out how to better control flexible legged locomotive systems. It had been shown, in previous work in the CRAWLAB, that flexible components built into a locomotive system, could improve performance metrics such as running speed, jumping height and power consumption. Defining controllers for such systems is a tedious task however, so we set out to learn them using RL instead, with the goal in mind being that the final controllers would be more power efficient. The learned controllers were deployed and evaluated on the robot shown, and ultimatly showed that RL can be used to learn more efficient control strategies.  

Design and test a end-to-end system to process whole, cooked, Louisiana crawfish tails. This project was funded by the Louisiana Board of Crawfish, with the intent of bringing back, much of the market share that has been lost to a cheaper labor force in China. Hand processing crawfish in peeler houses is an extremely time consuming, laborious, and costly venture. To alleviate these challenges, this project was initiated with a multi-year timeline where my contributions where between the years 2020 to 2022. During my time in the CRAWLAB at the University of Louisiana at Lafayette, as a graduate research assistant, I was assigned as the lead graduate student on this project. In addition to myself, multiple undergraduate students worked on this project, who I was responsible for delegating tasks for in addition to reporting the progress of the tasks to the the PI of the lab. We were successful in designing a fully operational system, utilizing state of the art machine learning and computer vision techniques for many of the subsystem operations.

As a member of the design engineering team at Kearfott Corporation, I was tasked with defining a testing strategy in addition to the testing facilities, to evaluate the performance of the Control Stick Assembly designed by my team for the KFX Fighter jet, a craft built by Korean Aerospace Industries. Some of my responsibilities included the following. 

• Developed testing apparatus for the Control Stick Assembly (CSA) to ensure its readiness to be installed in the KF-X fighter jet

• Collaborated with Korean Airspace Industries to define and generate testing procedures and reports

• Defined and utilized sensor equipment such as multi-axis torqe and force load transducers needed for evaluating the CSA hardware and software

• Designed 3D models and 2D drawings employing geometric tolerancing and dimensioning

(Due to a disclosure clause, that is all I can share 😉) 

With an ever growing desire to increase renewable energy generation, the need for the skills and workforce to install/maintain those systems also grows. Solar power, being one of the leading sources of renewable energy, is particularly costly to setup on a large scale, where payback periods can easily exceed 10 years! This project seeks to reduce both installation cost and time, through the use of autonomous robots. 

My contribution to this project, as is shown in the model image, is the system needed to interact with the payload (wooded pallet loaded with solar equipment). I completed the design and mechanical analysis of the system and oversaw the fabrication of 1 of 4 fully operational units (sadly we were kicked out of the lab in spring of 2020 due to you know who).