Projects Portfolio

Background

I was tasked with improving on a previous design from the Traverso Lab known as L-SOMA which uses self orienting mechanics to inject 80uL of liquid drug subcutaneously in the stomach. The goal for the project was to increase the capacity of the device to 250uL and reduce the footprint to the size of a previous device approved for daily dosing (to prevent blockages). Three separate devices were produced: one using a hypodermic needle, one using drug infused microneedles, and one using a high powered jet of mRNA formulation.

My Contributions

I worked in a group of 4 people and led the design of the devices. The parts were created using Fusion 360, and early prototypes were made using SLA printers, a metal laser cutter, and cast materials. The designs were made smaller than L-SOMA while retaining self-orientation capabilities and increasing drug loading capacity. Once the initial concepts proved successful, I refined the design to use off-the-shelf parts, and modified designs so they were suitable for CNC machining out of acetal resin and stainless steel. The jetting device required literature review and mathematical modelling of a free jet to better understand the power requirements of needleless injection. The mathematical model is being used in conjunction with ex-vivo testing to inform future designs.

These efforts were part of a government contract, and along with engineering responsibilities, I contributed to weekly presentations outlining the work done towards deadlines.

Outcomes

I designed three separate devices which are able to self-orient, deliver mRNA formulation, and that are small enough to pass reliably through the GI tract. Drug loading capacity was improved by 300%, and volumetric efficiency (volume of drug / total volume of device) was increased from 9% to 30%. All three devices are currently being tested in-vivo, and I am further refining their designs based on experimental feedback.

***The concept I came up with for the increased drug loading capacity is currently going through MIT's Technology Licensing Office, and is in the process of being patented, so I can't talk fully about the design. All images come from previously published papers.***

Background

For the second semester of my freshman engineering course, we were tasked with creating a solution to a futuristic problem. I decided to research proposals for Martian habitat construction, and the most interesting option to me was 3D printed structures using Martian regolith. This made me consider how the regolith would be harvested, leading to our project of a small autonomous rover capable of collecting rocks from the Martian surface.

My Contributions

I was chosen by my peers to be project lead, and wrote the code for our rover, designed several parts using SOLIDWORKS, and oversaw the assembly of our project.

The main feature of our rover was its autonomy. To achieve this, we used a Raspberry Pi Cam and OpenCV to detect rocks and drive the rover towards them. I had never done a computer vision project before, so I used this project as an opportunity to further my Python skills and learn more about the capabilities of the OpenCV library.

I also used SOLIDWORKS to create the servo mount and claw attachments. 

Outcomes

Our rover was capable of detecting objects using its camera and autonomously driving to them and picking them up with a success rate of 60%. The device was designed using only equipment that came in our freshman engineering toolkits, and with more time and money, the device could be further improved.

This project was undertaken during the first semester of my freshman engineering course. We were tasked with creating a product that could help a marginalized community, and I decided to research navigation solutions for blind people. In our assigned groups, my idea for a walking cane with sensor integration was deemed the best, and we created a prototype product. In my research, I found that other "smart canes" are prohibitively expensive and unnecessarily complex. Our goal was to create a simple, inexpensive device that would provide the user with GPS based directions and warnings about potential obstacles.

 A key aspect of my original idea was making the device attachable to any existing cane. This would lower the cost of the device, and also allow the user to keep their current cane. I came up with the idea of a two-part system. The first part would attach to the cane using two plates. Four bolts would sandwich the cane between these plates, similar to how a scope attaches to a gun barrel. The second part uses a dovetail joint secured with a pin. This allows the user to quickly slide the device on and off of their cane, and allows them to use the cane while the device is charging separately.

I did the bulk of the code. This involved learning how to use the Arduino IDE and incorporate GPS data to drive navigation. We used small motors attached to the inside of the device to give haptic feedback whenever the user was going the wrong direction.

At the beginning of the project, I was chosen by my peers to be the team lead. I was responsible for delegating tasks, holding my teammates accountable, organizing our meetings, and overseeing the project's overall progress.

One of the biggest challenges NU Rover faced last year was accurately controlling the end effector of their robot arm. This was mainly due to excess backlash in the primary axis, which caused the arm's claw to deviate around 2 inches left to right when fully extended.

Another student and I researched ways of reducing the backlash, and reanalyzed the force required for our applications. When doing calculations of the acceleration needed, we realized the torque in the primary axis could be reduced by nearly 60%. We came up with several solutions to reduce backlash, and devised 3 possible solutions, each using different configurations of motors and gearboxes.

The final parts were made out of aluminum, and succesfully reduced backlash to 1.4" at full extension.

NUAV's research drone, LISARD, is designed to accept attachments on the underside of its frame. I created a LIDAR attachment that allows the drone to create a point cloud of its environment, aiding in object detection and avoidance.

All parts were designed in SOLIDWORKS and 3D printed on Prusa Mk3s. Three one dimensional LIDAR sensors are mounted to a servo "head" using heat set inserts. We are currently using a servo motor to rotate the LIDAR sensors, but may switch to a stepper motor in the future. By having three LIDAR sensors, we give ourselves a 360 degree field of vision and increased scanning rate.

We are currently working on integrating our LIDAR sensors with the flight controller software and outputting the data to a remote computer.

This drone was built using a variety of tools to introduce us to the parts that make up a drone. In a group of 4, we used a laser cutter to fabricate the frame out of wood, and 3D printed motor mounts and arm clamps. We soldered the electronic speed controllers, motors, and radio receiver, and made efficient use of the drone's small frame.

We then used Mission Planner, a flight control software, to program autonomous flight.

Since it was an introductory drone, half the challenge was working out bugs and optimizing performance.