Team Maestro (UCSB Capstone)
https://capstone.engineering.ucsb.edu/projects/maestro
Left to right:
Geoff Tsai (UC Santa Barbara)
Shadmehr Gitiforooz (Mechanical Engineering)
Gerson Cruz (Mechanical Engineering)
Ryan Tong (Mechanical Engineering)
Colin Yang (Mechanical Engineering)
Luca Zovich (Mechanical Engineering)
Vincent Shyu (SONOS)
Nick Jones (SONOS)
Nigel Pemberton-Pigott (SONOS)
Celine Liu (Art)
Elizabeth Lee (Art)
For my senior project, I was part of the Maestro team consisting of four other mechanical engineering students and two art students. Working with SONOS, the team was given $5000 to design personal audio systems. The process started with understanding the company's design language, as seen by the image on the right of the team dissecting current SONOS products. This transitioned into competitor analysis, where the team analyzed the dimensions, material choice, and features of other products that were currently available. After conducting user interviews and surveys, the team gained an understanding of what customers prioritized and was able to use this information to influence initial designs.
As the rapid prototyping stage began, this required a constant collaboration between the mechanical and industrial design teams: the product had to be feasible to manufacture while meeting the visual expectations set by SONOS. Following numerous iterations on both SolidWorks and 3D printed models, the team narrowed down the designs to two ideas, ranging from wild to mild. The former pushed the limits of the competitor products in hopes of reaching a new audience, symbolizing the collaboration of creativity and performance. The latter followed a more familiar look with the added element of class expected from the company, targeting the existing customer base. These concepts were then brought to life in the machine shop as seen in the images on the left. My machining expertise allowed me to tackle the easier cuts with manual mills and shears, and I was able to write G-code to handle some of the more complex geometries. Once this was complete, the parts were rolled to a specific curvature and heat treated based on data in the Machinery's Handbook to achieve a desirable stiffness and yield strength. The same part was manufactured using 1050 spring steel and 1095 spring steel for their elastic properties and 6061 aluminum for its low mass.
All of the parts were brought to UCSB's testing facility and secured onto a MTS 810 machine. As the parts experienced up to 170N of force, their load-displacement data were recorded to reveal material properties. This included the identification of when warping would occur and whether this number was much greater than expected to avoid failure with typical usage. The experimental data was then plotted against FEA simulations done on the CAD models to ensure that the test results were sound, helping the team move on with 1050 spring steel when manufacturing the final product.
Unlike some of the other capstone teams that were exclusively engineering oriented, our team included art students and an industrial design lead from SONOS. This allowed me to learn more about the artistic side of product generation: I was able to conduct trend research and participate in ID meetings to understand how they considered factors like ergonomics and appearance into their design reasoning. Through this project I got to work with software like Miro and Figma, as well as design tools like 2x2 matrices and Pugh charts. In addition, I got to contribute to the project's branding, which included logo creation and a promotional video that highlighted the products' features, as seen in the renderings on the left.
While the exact details of this project cannot be shared, the work done by our team paved the path for SONOS's first headphone product. More details about the SONOS Ace can be found here: https://www.sonos.com/en-us/shop/sonos-ace-black
Natural Fibers (Biomaterials)
Jefferson Cam (Mechanical Engineering)
Ryan Tong (Mechanical Engineering)
Colin Yang (Mechanical Engineering)
ME 125RS at UC Santa Barbara is a research topics course on material structure-property relationships, processing, and characterization techniques for metals, polymers, and ceramics. Throughout the course, students learn about various medical devices and the considerations taken to manufacture these products. Once a steady foundation is established on strategies to enhance biocompatibility, students wrote a report on how to improve an existing class II or III medical device by changing its material properties.
Particularly fascinated with hip replacements, my team proposed the use of natural fibers like hemp or flax as an alternative to other fibers woven into composites used for the femoral stem. The inspiration comes from Porsche's GT4 class car, which uses a flax-based fiber in the body panels to reduce cost and environmental impact.
The report starts off with a research on current hip prosthetics, including a breakdown of the components and a brief history. It then identifies the benefits of using natural fibers as a biomaterial, such as its vibration dampening properties, as well as its drawbacks like its hydrophilic nature. To address the concerns that are brought up, the team ends by proposing a series of both in-vitro and in-vivo tests like direct contact of the material with mouse fibroblasts or intramuscular implant.
Learn more about Natural Fibers:
MNIST and Mushrooms (Deep Learning)
ME 125ML at UC Santa Barbara is a research topics course on recent developments in deep learning. It covers the following topics, presented through lectures and relevant papers: feedforward neural networks, backpropagation, optimization approaches, regularization methods, convolutional neural networks, sequence modeling with recurrent neural networks, and deep residual learning.
The reports below are a few examples of the neural networks that I created throughout the quarter using Python (Anaconda). On the left is a program that is able to classify a hand-drawn digit from the MNIST data set. It creates a sequential class with pooling and dense layers and uses an Adam optimizer to learn from the data available. At the time it was run, the model had 96.170% accuracy after nine epochs and was correctly able to identify a hand-drawn number four. On the right is a program that analyzes 22 attributes and determines whether a given mushroom is edible. It simplifies the data set initially by using a one hot encoding layer, then creates a sequential class similar to the one used for the MNIST data set. The model utilizes a SGD optimizer and had an accuracy of 98.706% after 100 epochs at the time it was run.
Learn more about MNIST:
Learn more about Mushrooms:
Additionally, as part of an interview process for Sandia National Lab, I prepared a 45 minute technical seminar on deep learning. An explanation of some basic machine learning concepts are included, such as how computers learn, how to make predictions more accurate, and what kind of models are available. The seminar then dives into two of the three main deep neural networks: convolutional neural networks and multi-layer perceptrons. Each explanation is supported by a relevant project, as seen in the reports above.
Learn more about Deep Learning:
CAD/CAM (Manufacturing Tools)
ME 158 at UC Santa Barbara is a class on programming, operation, and design of automated manufacturing tools. Students learned to manually program using G&M code, as well as using HSM to generate tool-paths. The machines used throughout the quarter include Sherline mill, X-Carve machine, and Full Spectrum laser cutter, and four projects were completed.
HSM Practice: As an introduction to HSM, students learned to use various 2D milling operations on a pre-made SolidWorks part. The job was post processed using Flashcut and machined using a Sherline mill.
X-Carve Oval Tag: Using Inventables, students created a design for an oval tag. This was then etched onto a piece of acrylic using an X-Carve machine, incorporating stylistic features like valleys by changing the tool used.
Laser Cut Tower: Students learned to use the Full Spectrum Laser RetinaEngrave software by creating the tallest tower possible out of a 5.5" x 5.5" sheet of acrylic. This required SolidWorks modeling, making the file into a drawing with different colors, and completing the job with a laser cutter.
Replica Lathe Tool Holder: My final project was creating a replica lathe tool holder out of a 1" x 1" x 2" prototype foam block, as shown in the images on the left. It involved designing the part, stock simulation using specific tools, writing out job descriptions for repeatability, and using the Sherline mill to drill and face different surfaces.
Learn more about CAD/CAM:
Truss Design (Structures)
ME 154 at UC Santa Barbara dives into matrix methods of structural analysis and the mechanics of linear elastic bodies. Students applied the Rayleigh-Ritz method to calculate linear elasticity by hand, and conducted FEA through the Abaqus software. FEA simulations done during the quarter included: determining how many lines of support blocks are needed to prevent yielding of steel plates, limit testing blocks of grey cast iron with various holes, and truss design to withstand a specific weight while minimizing the material used.
Learn more about Truss Design:
Fluids and Vibrations (Mechanical Engineering Lab)
Jefferson Cam (Mechanical Engineering)
Mason Triplett (Mechanical Engineering)
Colin Yang (Mechanical Engineering)
ME 105 at UC Santa Barbara focuses on four different experiments drawn from thermal sciences, fluid and solid mechanics, dynamics, and materials science. The primary goals of this course was for students to learn how to calibrate and use appropriate instrumentation to gather data, quantify measurement uncertainty, and present the results through professional communication methods. The fluid mechanics report is written in a traditional format, while the vibrations report describes the results in a presentation form.
Learn more about Fluid Mechanics:
Learn more about Vibrations:
ME Design.mp4Bike Signal (Engineering Design)
Jameson Johnston (Mechanical Engineering)
Colin Yang (Mechanical Engineering)
ME 153 is an introductory class to the design process geared towards mechanical engineering majors at UC Santa Barbara. The primary goal of the class is to prototype a product that addresses a prevalent issue, and improve upon it through several iterations of testing (durability, connection range, etc.). The problem my team tackled was the inability to properly signal when biking, whether that be for nighttime bikers or children that are just learning how to ride. Our product is made of two components: a button pad that goes on the handlebar and a LED display that goes on the back of a helmet. The two communicate via Bluetooth, and the user is able to signal turns (shown via a gliding motion of blue LEDs) and stops (shown via solid red LEDs) with the press of a button. The report and video go into more detail of how our device works and the step taken to reach this final point.
Learn more about Bike Signal:
IMG_2901.MOVMotors, Electronics, and Controls (Mechatronics)
ME 104 is an introductory class to mechanical and electrical systems at UC Santa Barbara. During the year, students worked with sensors, actuators, and computer interfacing and control primarily though the projects found in the Arduino Engineering Kit Rev. 2, as well as gaining familiarity with MATLAB and Simulink as modeling tools.
Drawing Robot: Maneuvers a robot around a whiteboard via closed-loop control by mathematically determining the coordinates needed to draw out an image.
Self-Driving Rover: Used a closed-loop system to drive a rover along a predetermined path, tuning the proportional and integral gain values to provide better results.
Self-Balancing Motorcycle: A motorcycle is designed to balance on its two wheels despite being pushed around, thanks to a system that can control a reaction wheel's speed and direction to use its mass moment of inertia to counter the external forces.
Learn more about Drawing Robot:
Learn more about Self-Driving Rover:
Learn more about Self-Balancing Motorcycle:
COVID, Advection, and Cooling (MATLAB and LaTeX)
ME 17 is an introductory class to MATLAB implementation at UC Santa Barbara. During the year, students used MATLAB to run various simulations, as well as to verify solutions for mathematical problems involving linear algebra, matrices, eigenvalues, and determinants. In addition, students learned to write professional technical reports using LaTeX and analyze how the results were applicable to the real world.
COVID: Analyzes the number of cases over the number of days, then plots these values against the health system's maximum capacity. It utilizes Euler's method and a varying social distancing factor (percentage of population practicing social distancing) to emphasize its effects.
Advection Equation: A numerical simulation of a moving sphere that incorporates the upwind scheme to model how the sphere's density changes under a velocity field. It demonstrates how proper boundary conditions and dissipation modeling allow the sphere to realistically move through a 3D grid.
Cooling System: Visualizes the advection-diffusion equation with Dirichlet and Robin boundary conditions by simulating a hot room's temperature over time as cool air is introduced by a fan. The model shows the temperature distribution of the room, as well as how it reaches a steady-state within a couple seconds.
Learn more about COVID:
Learn more about Advection Equation:
Learn more about Cooling System:
Dancing Robot (CAD & Design)
Alexander Escalante (Mechanical Engineering)
Jason Fong (Mechanical Engineering)
Nathan Green (Mechanical Engineering)
Colin Yang (Mechanical Engineering)
Our Isla Vista Elementary School junior engineer asked to resemble his character, Rift, who had a red, blue, and black color scheme with triangle goggles and stripes. He also asked that his robot be capable of some fighting moves, including a double punch, a front kick, and a 360 kick. To address this, our team designed a mechatronic skeleton composed of an Arduino, a battery pack, and servo motors moving 3D printed arm and leg components. Additionally, we added wheels to accomplish the 360 motion and blue LEDs to make the eyes glow.
Learn more about Dancing Robot:
