E
My projects
RECYCLE RACER
RECYCLE RACER
During my freshmen year at Dublin High, I undertook a project in Intro to Engineering and Design called Recycle Racer. Designed for one of the editors in Makezine Magazine, our design statement was to design, build, test, document, and present a car made of purely recycled materials. We had constraints such as a dimensions requirements and trigger mechanisms. Using a design matrix, I was able to eliminate materials that would make our car ineffective. Our trigger mechanism consisted of a pipe connected to a mouse trap that rotated the axles. I glued DVDs to pencil straws to reduce friction between the ground and the model. Furthermore, we opted for a cardboard chassis to ensure that our car was as light as possible.
Initially, I designed prototype with wood, however, the block was too heavy to rotate on the wheels. I drew a rectangle on cardboard and used a utility knife to cut it out. I cut straws to act as bearings for the axles to rotate which were attached with duct tape and super glue. The implementation of the mouse trap into the card was personally my hardest task. The trigger for the mousetrap had to be centered exactly every time otherwise the string would catch in the cardboard. I tried multiple sizes and types of string to eliminate this predicament. Finally, I switched the orientation of the mousetrap to help solve this problem. Overall, our model worked pretty well and went 13 meters with one mousetrap.
Check out the project here .
PAPER TOY PROJECT
Another project that I undertook in Intro to Engineering and Design was the Paper Toy Project. In this project, we designed a paper toy model on Autodesk Inventor for a company called Kani Model. We were to design a new paper toy model to increase their sales. The parts were to print out as a net. We were then required to cut them out and assemble the toy ourselves. The final toy had to be completed in 8 days, had to contain at least a base and two additional parts that attach to it, tab attachments to secure the pieces into each other, and flat pattern files for reproduction. I modeled a 3-dimensional figure after the superhero The Flash. I modeled each separate limb, torso, and head using distinct sheet metal commands that allowed me to create a net of the figures. I printed everything out and cut the pieces out. After folding them and cutting the slits, I constructed the full figure. by putting the corresponding pieces together. The main obstacles that I had to overcome were during the rendering of the digital files and during construction. While creating the net of the paper toy on Inventor, it took spacial awareness to calculate the relative location and depths of the slits. During Construction, it took precision to ensure that the wrong lines and slits were not being cut. However, I enjoyed this project because it allowed me to use my skills that I had learned and apply them in a real-life project.
Check out the project here.
MECHCANICAL WINCH PROJECT
In the beginning of sophomore year, I was enrolled in the next engineering class called Principles of Engineering. In this class, we explore all mediums of engineering and analyze the different skills that engineers require in specific fields of study. For one of our projects, my class was assigned to build a mechanical winch that could allow to pick up an object from the ground with at least 30cm in height. We were supposed to design the winch with the highest achievable efficiency.
The assembly is fitted inside 3 five-hole C-channels that allow screwed to each other in 90-degree angles. The winch drum, motor, and sprocket/chain system all lie inside this open cage. The wires from the motor trail out from behind that then connect to switch and a power supply. Furthermore, the drum itself is placed at the other open end where the string can grasp onto to items below. I used a VEX motor as a weight. The motor is about 101 grams and is tied to onto one of the standoffs spanning the gears. To connect the motor to the main winch drum, I used a sprocket and chain system to extend the movement of the motor. Using a belt system or gear train seemed unreasonable because they were simply too complex and unstable. Therefore, I decided that using a chain and sprocket system would be the most reliable. To make our main drum, I used 2 large 84 teeth vex gear with standoffs in between. With such a large drum I was able to increase the speed as torque was already being outputted through the main motor gear. To make the Fischer Tech motor compatible with the rest of the VEX assembly, I used a clamping red gear to grasp onto the VEX axle. This way the Fischer Tech motor was able to power the Fischer Tech gear, which then turned the VEX chain and sprocket.
Check out the project here.
FACE UNLOCKER NOCKER
At the end of the year in Principles of Engineering, the entire class was required to create an entrepreneurship project for a current problem in society that hasn't been solved. The showcase for this end of the year project eventually becomes a competitive showcase. I had an existing project from the ACSEF that won first-place amongst 900 other projects. Basically, I built an automated door locking mechanism that utilizes facial recognition to unlock or lock the door for credentialed users. I used facial recognition APIs from Microsoft and artificial intelligence to prototype an efficient model. With advice from industry professionals and a new outlook on design, I set out to expand this project. My group was selected as 1st place winners in both POE classes and won the Best Presentation award at the showcase.
For this project, I was the lead developer. I was in charge of the back end code for our project. I researched different effective APIs for our system to ensure we could create an efficient facial recognition system. My main research focus centered around implementing deep-machine learning algorithms within the code to allow the system to learn by itself without being explicitly programmed to. I researched current solutions and used a mixture of previous research for the project. Another important area of research for this project was a study of darker skin complexion vs lighter skin complexions. After throughout testing, I realized that on average users with darker skin complexions had a tendency to yield a lower positive call back ratio, or a lower match with the software. I studied this repeatedly occurring phenomenon and implemented a solution in the code to handle this issue. I was also in charge of making all the 3D models which were sent to the lead designer for final renders.
The camera system is mounted onto a protruding plate that holds a door knocker. With the entire module disguised as a door knocker, your guests won't even know it’s there. By the time someone is detected outside the door, F.U.N. automatically starts recording. With the advanced IoT processor built in, F.U.N. is able to recognize the people in the entryway in day and night with extreme precision. Whether it's one person or multiple, the program alters itself to be more secure, allowing multiple credentialed users. Furthermore, with deep-machine learning, the system adapts to your needs. Regardless of whether you put it on your front door or yard door, the software automatically adjusts to ensure maximum efficiency.
BLACK JACK PROJECT
In the beginning of junior year, I was enrolled in the next engineering class called AP Computer Science A. In this class, we explore Java programming with emphasis on basic data structures and recursive algorithms. For one of our projects, my class was assigned to create a card game in which the user could play against the computer in a simple game. We chose to create the game Blackjack, or commonly known as 21.
As the name 21 implies, the object of the game is to come as close to 21 points as possible with your cards -- without going over -- and having a higher total than the dealer. The game begins with the player placing a bet, then receiving two cards face up. The dealer also receives two cards. The player wins the amount that was bet if the total value of his cards is closer to 21 than the dealer's cards. Another way to win is if the dealer goes over 21, while the player does not. The player loses the amount bet if his total goes over 21. The game of blackjack is dealt with six decks of cards (to prevent card counting), and after each hand (or a set number of hands), the cards are re-shuffled. The dealer must continue drawing cards if the value of his two cards is 17 points or less, while the player can stop at any time. The Player can split his cards when dealt cards of the same value (for instance, 6-6 or J-Q), doubling his opportunity to play that particular hand. Split aces receive only one additional card.
In the game, we coded for the game to automatically decide between picking for the value of the Ace card. The game always picks in favor of the user against the computer. The overall display of the game is supposed to be user friendly while encompassing the major tasks needed in order to play the game.
Check out the project here.
HEART PROTECH
At the end of the year in AP Computer Science A, the entire class was required to create an entrepreneurship project for a current problem in society that hasn't been solved. The showcase for this end of the year project eventually becomes a competitive showcase. I had an existing research project that evolved to be a customer oriented project.
Various cardiovascular (CV) illnesses have been recurring as the leading cause of death for the past few decades. Different models that predict the progression of CV illnesses and/or the possibility of their occurrence exist, however they either too holistic or inaccurate to benefit an individual. The goal of this project is to create a portable all-in-one heart health sensors that uses different sensors, classification algorithms, data mining techniques, and neural networks to create an efficient and accurate prediction model that can be used by a patient to determine the deterioration of the quality of their heart. The application allows the patient to receive a risk assessment report based on the quality of their heart and defined parameter. Various different health sensors provide patient data over time which is inputted along with pre-trained modules into the supervised algorithm. The sensors are all connected into the portable all-in-one unit that parses data through every use. The model sends the data to the model in the AWS instance. Since the entire prediction model continuously runs within an AWS lambda instance, the heart health prediction is constantly being updated. A data archive of 14 different attributes from 200 people was used from the Stanford archive to training purposes. The different testing instances gave accuracy values that ranged from 85% to 92%. With further testing and more datasets, patients will eventually be able to use this device to be better educated about their ailments, possibly identifying general occurrences of certain CV illnesses before they occur.
Check out the project here.
