Portfolio

Contents

Research Work Projects

Soft Robotic Prosthetic Hand (Spring 2019)

Project Summary: This project is under the supervision of Dr. Turaj Ashuri of the Mechanical Engineering Dept. at Arkansas Tech University and funded by the National Institute of Health (NIH). The goal of the project is to use elastomer type polymer and create a prosthetic hand as a functional soft-robotic hand.

The first two iterations of the project were fabricated by a group of senior engineering students at Arkansas Tech University. The soft-robotic hand was programmed with the Arduino Uno board with the functionality of imitating the grabbing motion of a human hand.

At this level, the project was handed as the Graduate Research Assistant project to upgrade the functionality and create documentation of the project as the undergraduate team lacked proper documentation of the research.

Contribution to the project:

Soft-Robotic Hand Research Presentation:

SRHPpt.pdf
Functionality Portfolio Version.mp4

Controller Board Schematic (Prototype):

Adruino Coding:

Novel Additive Manufacturing Process Development (Current) x 2

Pending update

Developing new print module from design, fabrication, to control using custom circuitry and code that works with collaborative 6 axis robot system. This projects combined knowledge of additive manufacturing process such as FDM , LOM, along with robotic manipulation. 

Additional details will be updated upon publication of the research work

Additive Manufacturing Case with Swarm Robotics (Spring 2022)

Pending update

Demonstration of manufacturability of a miniature robotic vehicle using a swarm of robotics with manufacturing skills such as 3D printing, laser processing, and manipulation.

Additional details will be updated upon publication of the research work

Course Work Projects

Binary Image Classification with Neural Network (Fall 2020)

Project Summary: The objectives of this project were to collect readily available image data set and use deep neural network to train and test a binary image classifier with the data set. Experimenting with hyper-parameters to see the differences in test and train results from the classifier. In this project, image data set was collected from one of the sources where image datasets are readily available for research purposes. The next step was to analyze the data and process it for further use. A deep neural network was used as the binary classifier with different configurations of hyper-parameters to test and train with the dataset. The programming language was python for this project.

Dataset: The dataset contained the images of portraits of people wearing masks. There are two classes of images in this dataset. Class one was portraits of people wearing face masks properly covering their nose and mouth. The second class was images of people incorrectly worn face masks such as not covering the nose, not covering the mouth, leaving at the chin, etc. The entire original dataset source contained more than 134,093 images where 67,193 images with correctly masked face images and 66,900 images with incorrectly masked face datasets. Conveniently all the images were in the same resolution of 1024 x 1024. For simplicity of the project, datasets of 1911 training and 200 test images were used for the entire approach.

As images are three by three matrix each one representing red, blue, and green color to represent an image. With the dataset image resolution, 1024 x 1024 would provide three matrices of the size of the resolution. Thus, each image will have a size of 1024 x 1024 x 3. Each image matrix data need to be stored in a 3-dimensional array and later the training set array will consist of a 4-dimensional array with a size of 1911 x 1024 x 1024 x 3, for the test set the array size will be 200 x 1024 x 1024 x 3.

To reduce the computational cost, the images were reduced to 128 x 128 resolution. Which left the project with a training array with the size 1911 x 128 x 128 x 3 and a testing array of size 200 x 128 x 128 x 3.

Training result of the DNN model: The DNN on this project was trained with a learning rate of 0.01 and with 500 iterations. After training and testing the DNN the results received is given below:

Project Report:

Applied Machine Learning for I.pdf

Building Heating Load Calculation (Fall 2019)

Project Summary: This project was focused to determine the size of the heating unit(s) required for the BAZ-TECH building on the Arkansas Tech University campus. Building architectural, and structural blueprints were provided for necessary calculations for the process. In this project, the first objective was to determine all the transmission loads (heat transfers in walls, slabs, roofs, infiltration, and ventilation). The second objective was to find a suitable heating unit(s) to provide heating for the building.

The first step on this project was to study the CAD drawings and sketches to get all the dimensions, measurements, and construction of the structure. These values were used later in the project for load calculations.

Load calculations were done in multiple levels:

Calculating the total heating loads, a preferred heat pump was suggested for the Baz-Tech building project.

Project Report:

HVAC BAZ Tech Report.pdf

Industry Organizational Research: General Mills (Spring 2021)

Project Summary: This project's objective was to learn about business industry research and investigate industry-leading companies in specified industries. The assigned industry for this project was Breakfast Cereal Manufacturing (NAICS Code 311230). General Mills was selected for the industry research paper for this project. Several online business databases such as D&B Hoovers, MarketLine, Mergent, and Fiscal Annual Reports from General Mills.

Project Report:

Tushar_Nahid_GeneralMillsInc.pdf

Project Management: Prosthetic Hand Research (Fall 2020)

Project Summary: This project was to learn about different aspects of project management as well as learning to work with Microsoft Project.

Getting an idea and using it on a sample project to learn about:

As this class required a sample work project to use the information and idea learned from this class, I used my concurrent ongoing research on soft robotic hands as the sample work project for this class

Project Report: 

Tushar_N Final Project Report.pdf

Senior Year Design Project: "The Gen-X-Bike" (Spring 2018)

Project Summary:  The kinetic energy of a moving bike goes to waste as frictional heat when the braking system takes into effect. This project was focused to harvest the kinetic energy to storage and use that to power a motor for the bike. 

Excel Calculation Sheet to analyze power requirements based on variation on input parameters 

00.pdf

Gen-X Bike PowerPoint Presentation: Product Design Review

PDR MCEG4202.1.pdf

Figures, formulas, and calculations

SDP Project Tushar's Slides.pdf

Gas Turbine Engine: After-burning Turbojet engine cycle simulation (Spring 2018)

Simulation Program

Gas Turbine Engine: Modern Aircraft Exhaust System (Spring 2018)

Exhaust Nozzle Analysis

Aircraft Engine NozzlesLATEST.pdf

Finite Element Analysis of a Truss Structure (Spring 2018)

Finite Element Analysis

Truss Project: Design and analyze a truss structure to satisfy specific constraints and loading achieving minimal mass, while minimal deflection was a secondary objective

Modeling and Control Design of Aircraft Pitch Controller (Spring 2018)

Control System: 

FinalCSPRJ.pdf

Coffee Container Design (Fall 2017)

Heat Transfer

CoffeeMug.pdf

Misc.

The SAP S/4HANA Data Retrieval (Summer 2019)

Keyword:

Background: SAP S/4HANA was introduced during the internship opportunity at DBG Arkansas LLC, in Conway, Arkansas. The plant was going through system integration to SAP's cloud server from old digital and print files. Creating/Updating Routing and Work Stations, retrieving Bill of Material (BOM) data, or updating excel spreadsheets with detailed routing information of around 5000 parts were the typical daily job assigned to the interns. Value stream mapping and observation were assigned occasionally.

Maintaining an excel master file with over 20000 rows with more than 10 columns was challenging. Occasionally there were moments when a piece of very specific information about a specific part from SAP needed to be added to the master file. 

Updating the whole file on average took 1.5 weeks for 3 interns working on it. The process was more or less to log in to the SAP server, appropriate application (such as display routing, BOM or etc.) and input specific part numbers and wait for the server to retrieve the data table and find info in needed and update that in the corresponding excel file. Being assigned to this task was a great opportunity to get familiar with SAP S/4HANA cloud ERP.

Methodology/Approach: Occasional discussions between interns brought up the idea that if there was a program to get the information that would both save time and accuracy. While consulting with the manufacturing engineer, it was found that there is no API plug-in currently for the SAP S/4HANA cloud server that would help retract the data from the server. If it was an on-premises server, getting data with an already available API would have been inexpensive in terms of money and time.

Learning Python for scientific calculations and programming is another constant process since my Senior year of college. Basic computation, web scraping, and working with excel files, data frames, etc. were a good foundation for this project.

This brought an opportunity to try and work to automate the data-retrieving procedure. 

The first goal was to enable the code to open an internet browser and go to the cloud server and log in with the provided credentials. The next challenge was to navigate through the homepage and different applications. This is where more exposure to HTML coding was experienced.

Code Screenshots

Development: Upon developing an algorithm and decision tree/block diagram for the code and consulting with other interns and immediate supervisor for all possible outcomes in case of error like internet issue, server issue, SAP back end error issue, wrong material/part number issue, computer power issue, etc. All possible scenario was taken into consideration making the final algorithm and decision tree which perfected the block diagram. Then the first program was completed to handle "Display Routing" app in SAP and was demonstrated to the immediate supervisor. The supervisor set a time for the demonstration before the I.T. specialist and the Engineering Manager of the plant.

Upon demonstration, new tasks were assigned to create two more additional program that is used commonly in the day-to-day tasks in the plant. A specific workstation was requested by the immediate supervisor to run this program.

Following the same development footstep rest of the programs were completed and demonstrated to the immediate supervisor to test out the programs for code validation. Upon code validation, according to the request, one of the manufacturing engineer's workstations was set up with a related programming language to be able to run the scripts.

To avoid the complication of having programming language, IDE, etc installed to run the program:

The next step was making three programs into one GUI-based script. Upon the completion of the GUI-based code, the whole program was made into an executable to be able to run on all the computers with the Windows operating system. Later, for ease of distribution, a final development was to make the program an executable installation file.

Demo-video:

PortfolioDemo.mp4

End Results:

Having an updated master file with all routing data for group counter 1, routing data for the latest group counter, and BOM data for over 5000 parts is a time-consuming task if done manually as no API was developed at that time for SAP S/4Hana Cloud Servers. 

For the confidentiality and sensitivity of the SAP S/4HANA coding strategy and DBG Arkansas LLC's data, the detailed development process has been omitted from this page