CSE 2023 Spring Departmental Demo Day

(EAS 561, MS in Robotics, Masters Project)

Autonomous Plant Watering Robot

• Charmin Desai

This project is about developing a plant watering robot. The goal is for the robot to water household plants autonomously. It should explore the unknown environment and build a map, and scan for possible candidate targets (plants) at the same time. It should validate the data with the user, through a friendly interface. Once the map is finalized, the robot should be able to navigate in the map and reach the plants avoiding obstacles, then water them according to the user's preferences.

(CSE 611: Master's Capstone Project)

The Food Court App

• Neel Arvind Patel
• Tejas Gajanan Bhitle
• Rajeshwari Barman
• Riddham Rajeshbhai Suvagiya
• Abhishek Paharia

The Food Court app aims to help people discover food socially and make food eating/preparing a social and fun experience. By providing a Tinder-style interface, filtering options, and the ability to add/share food images, recipes and restaurant info, the app will help users in decision making easier and discovering new recipes to prepare together from their homes.

IBM Digital Health Diary App

• arshiyac
• sowmiyam
• banerje3
• sagnikmo
• sk394

We created the IBM digital health diary, a smartphone app for tracking health, to assist users in performing routine health assessments and monitoring their progress over time. Flutter was implemented for the front-end, and Nodejs ,MongoDB and APIs were used for the back-development. It  has additional features including the ability to update tasks on a calendar, provide analytics in the form of charts and plots, retrieve results according to date or task, and provide context for where a given result falls on the scale. Overall, a Digital Health Diary can provide patients with a convenient and accessible way to manage their health and help healthcare providers deliver more effective and personalized care.

What Should I Wear ?

• Meghana Peram
• Venkat Subrahmanyam Upputuri
• Sai Kiran Angalakudhiti
• Esther Mona Kodavalli
• Vamsi Abhiram Guru

What Should I Wear is a website that provides details of weather in the current location or the location that user has selected and website provides user with appropriate clothing suggestions, Radar Information, Live Traffic Camera Images and alerts. User has an opportunity to select his/her preferences such as the workout (Run, Walk, Casual, Bike) and how do they feel (Normal, Hot, Cold) and the website will suggest the list of clothing suggestions based on the preferences.

Consentize Event Platform

• Shubhi Srivastava
• Anagha Vivekanand Joshi
• Manpreet Ajmani
• Smriti Rai
• Mohit Sunil Gandhi

Developed an Event Management portal for Consentize for handling multiple events

UB Attendance Tracker

• Sai Kiran Janga
• Hrishikesh Kakkad
• Ijaj Ahmed
• Subhojeet Roy
• Mehul Mittal

The objective is to create a website or dashboard that provides visualization of the statistics of attendance and demographics of various visitors attending spaces and events in a clear and straightforward format, along with options to generate and download reports. The site will also have features like limited access depending on the users, the capability to create QR codes for each space and events as well as the ability to track live attendance by any hosts.

NYCE

• Kushan Devarajegowda
• Naman Agrawal
• Pranaya Satwika Reddy Maddi
• Harshavardhan Reddy Bommireddy
• Hemanth Kumar Mutta
• Venkata Bala Vamsi Atukuri

NYCE (New York Cannabis Exchange) is a cutting-edge online platform engineered to foster robust connections between cultivators, processors, and dispensaries in the New York cannabis market. Utilizing a centralized auction marketplace, our team has crafted a solution that emphasizes transparency, efficiency, and a streamlined trading experience. Leveraging advanced features such as live bidding, real-time notifications, secure transactions, and a seamless onboarding process, NYCE enables sellers to maximize their revenue while providing buyers with access to a wide array of premium cannabis products. The platform's sophisticated technology stack includes React, Django, AWS, Redis, Docker, Git, and WeasyPrint, ensuring a scalable and robust solution that revolutionizes cannabis trading.

Kaleida Health App

• Kevin Truong
• Franky Naidu
• Chaoping Lin
• Hon Ching Li
• Runkai Jiang

An App that allows patients to view more details about their assigned diagnosis and give permission to family or friends who wish to view it too.

Photo Editor

• Mallikharjuna Rao Annam
• Manikanta kalyan Gokavarapu
• Rakesh Kumar Gavara
• Ruthika Juluri
• khyathi Reddy

Welcome to our photo editor web application, where you can edit your photos with ease! Our web application offers a variety of features to help you make your photos look great. Here are some of the features we offer: 1. Passport photo maker - Our web application allows you to create passport photos that meet the standards of different countries. You can choose the country you want to create the passport photo for, and our application will automatically adjust the photo to meet the size, format, and other requirements of that country. 2. Resize - You can resize your photos to any size you want. Whether you want to make your photos smaller or larger, our web application makes it easy for you to do so. 3. Crop - You can crop your photos to remove unwanted parts or focus on specific areas. Our web application offers crop options to help you achieve the desired effect. 4. Format change - You can change the format of your photos to JPEG, PNG, or other popular formats. This can be useful if you need to share your photos on different platforms or devices. 5. PDF maker - Our web application allows you to convert your photos to PDF format. This can be useful if you need to create a document that includes multiple photos. 6. Photo collage - You can create a photo collage by combining multiple photos into one. Our web application offers a variety of collage templates to help you get started. 7. Mosaic - You can create a mosaic effect on your photos by pixelating them. striking variation of a photo collage where you place multiple photos next to each other to create a bigger, often more impactful design. Just like the colorful pieces in a jigsaw puzzle, 8. Background change - You can change the background of your photos to any color you want. Our web application offers a variety of background options to help you achieve the desired effect. These are just some of the features we offer in our photo editor web application. We hope you enjoy using our application to edit your photos!

Self Describing Asset Catalog

• Raja Ramachandra Murthy Nukala
• Vishnu Kiran Nemani
• Tony Jason Anga
• Solomon Raj Panduga Ebenezer Vinod
• Harichandana Vejendla

In this digital world every enterprise will have a lot of digital assets which might include datasets, machine learning models, software code etc. Oftentimes these digital assets need to be shared for collaboration purposes. This project is a CLI tool - an extension built on top of the GitHub CLI which makes the collaboration easier by providing the ability to perform CRUDS(Create, read, update, delete, search) and other equivalent operations on the digital assets right from the command line.

Exam Study Tool

• Subhasri Rallabandi
• Aakash Baruva
• Sri Sannihitha Gangina
• Saishankar Vanam
• Sai Rochan Mantri

The goal of this project is to create a comprehensive, user-friendly study platform for students being administered the New York State Math Regent Examinations. Designed for all students this website aims to equip students with the necessary tools and resources to excel in Algebra I, Algebra II and Geometry. By providing problems from past Regent Exams, students have an authentic opportunity to practice for their Exams. English Language Students have the opportunity to practice questions in their home language whilst viewing the English version of the question. Performance tracking will help students and teachers monitor progress and identify areas of weakness. The Exam Study Tool aims to empower students to reach their full potential and achieve proficiency in these important exams.

Data Volume Reduction

• Suryakant Joshi
• Madhura Ashok Satao
• Vineet Kalghatgi
• Pranav Bellam Prasad

The Data volume reduction application introduces standard API to filter large datasets on the metadata. Our interface API interacts with the data provider On your behalf and enable you to explore the data before downloading.

An Extra Scoop

• Prashant Godhwani
• Rohith Ramichetty
• Shruti Asthana
• Praveen Kumar Ramineni
• Shubham Sunita Ambavale

Revolutionize your food business with our all-in-one recipe management and menu planning application.

Online Dictionary

• Hamza Hafiz
• Robin Mathew
• Samar Pratap Singh
• Akhil Surakanti
• Chris Yadala

Online Dictionary is a web application providing definitions and meanings of words in the English language. In addition to word definitions, the site also offers usage examples and pronunciation guides, making it an invaluable tool for anyone looking to expand their vocabulary. Whether you are a student, a teacher, a language learner, or just someone who loves words, this website is a must-visit resource for anyone seeking to deepen their understanding of the English language.

Continuous Checkpointing for Recommendation Systems

• Karthikeyan Rajagopal
• Dhayaneshwar Balusamy
• Sasidharreddy Lankireddy
• Vrashi Shrivastava

Deep Learning models on big datasets usually take hours to train or maybe even weeks. If the model run is interrupted or crashed, a lot of data, time, and effort might be lost. In order to prevent that, we can checkpoint our models during training. This enables us to take a snapshot of the model to prevent data loss of the entire run in case of failure. This will allow us to restart the training from the latest checkpoint. This can be further optimized by making the checkpoint mechanism real-time. We write the updated embedded tables continuously to a log system like Kafka.

On-Demand Event Scheduler

• Lakshaye Vaikunth Ganesh Kumar
• Arjun Ragavendaar Sripathy
• Sri Harshitha Palla
• Divya Sharma
• Venkatesh Mandapati

Our powerful platform makes it easy to poll and plan meetings, book appointments, and more - all in one place with just a few clicks. Try OES today and discover a smarter way to plan!

Capstone Management Tool

• Anvitha Leti
• Jahnavi Kota
• Vaidurya Malathesha
• Vaisaali Murali Krishnan

We are presenting an integrated platform/solution for all the dynamic project management needs. You get detailed bite cards of projects and user profiles for easy tracking. Effortlessly delegate and manage project assignments; notify students of project enrollment status, categorize projects into different types and support a combination of user roles. Also, optimize performance on the go with improved session management and ad-hoc bug fixes.

LifeTreeWNY

• Ranganadh Neelakantham
• Sarthak Jain
• Venkata Sai Saran Putta
• Rohan Rajput
• Rishab Aggarwal

LifetreeWNY is a web-based application that facilitates effective business scaling and management for its users by streamlining various workflows, thereby minimizing the need for manual communication. Through its automation capabilities, the platform enables clients to optimize their operations and achieve increased efficiency.

(CSE 370 : Human Computer Interaction)

Chalk-up/Seas

• Samur Rahman
• Allison Mai
• Dekayla Dubose
• Richard Prosser
• Sebastian Richel
• Jake Guzzardo

An educational based discord style application that works as a hub for students and professors

MyLyfe / aquafit

• Mokshita Gupta
• Taylor Evans
• Shafin Ahmad
• Muhtasim Mushfiq
• Andrew Cao
• Misha Chekhov

This project aims to create a online application for users to document all parts of their life - hence the name MyLyfe. Our goal is to bring our users closer together and create a supportive community for people looking to take steps towards improving their life.

Echidna

• Susan Brethauer
• Wren Martinson
• James Aquilina
• Connor Finn

BetterReads: a web application, similar to GoodReads, for discovering new books to read and making book reviews

(CSE 442: Software Engineering)

Giggle

• Oakley Thomas
• Mausam Vora
• Cory Margarucci
• Abi Abisha
• Aditya Pandya

A talent marketplace website that connects college musicians with gigs posted by the community

Sixerr

• Jake Morrison
• Jeffrey Xu
• Nofle Al
• Timothy Li
• Andy Zhou

Sixerr is a desktop and mobile web application where local communities and/or individuals can interact with one another to exchange goods and services. This app leverages the local community so that even the smallest issues can be amended quickly. For example, students of UB can create a request asking for a study partner, or make an offer to sell a textbook they no longer need. College students buy many things that they do not end up needing. Many of these things may be too trivial and unseen on sites like eBay, Facebook Marketplace, or Amazon. With Sixerr, a student can sell a carton of eggs they no longer want and someone else in the UB area can offer to buy those eggs and maybe offer to trade for a jar of peanut butter. Anything goes (as long as it is safe and legal).

PantryPal

• Mikayla Stevens
• Anthony Huynh
• Jun Lin
• Salim Zedni

Save your grocery shopping inventory and search for recipes that include things you have in stock. Avoid food allergies by adding information to your profile to help filter recipe search results. Ensure you stay on track to use all of your groceries before they expire with a meal planner to save recipe search results to. Keep your kitchen clean with a tracker on expiration dates and a reminder for when it's time to throw stuff out.

(CSE 540 Machine Learning and society)

Team 6

• Suradhya Gadde
• Hannah Jennings
• Gabriela Perez
• Harinee Purushothaman

The goal of this project is to develop a machine learning model that can analyze textbook content and identify phrases and words that perpetuate white supremacist ideologies or promote racial biases. By leveraging natural language processing techniques, the model will assist in moderating textbook standards and questioning the necessity of such content. This project aims to address the issue of one-sided narratives in textbooks, providing a tool to promote critical thinking and inclusivity in education. Additionally, collaborations with teachers and organizations will be fostered to support the adoption and implementation of this tool, ultimately working towards creating more balanced and unbiased educational materials.

(CSE 546: Reinforcement Learning)

Reinforcement Learning for Optimizing Warehouse Management in Complex Industrial Environments

• Aditi Bobde
• Mohit Sai Aravind Nunna
• Shail Rajesh Shah

We are working on an opensource environment RWARE with sparse reward structure and also tried to create our own environment similar to rware .Eligibility Sarsa, flavors UCB with q learning, DQN, but since the reward is so sparse we need to add more depth to the deep network or use a more complex algorithm like SEAC for the RWARE env ,we also created our custom env that we plan to use to teach the agent new to do tasks but training it to do tasks in order like first learning to reach the parcel /city and then increase complexity to first pick the parcel and then reach a goal,  and then increase the number of parcels and/or the number of agents

RC CARS

• Harshavardhan Baira Reddy
• Madhavan Rangarajan
• Dhiraj Gunasheela

The AWS Deep Racer is a self-driving 1/18th scale race car that serves as a platform for evaluating reinforcement learning models through real-world racing on a physical track. By employing cameras to perceive the track and using a reinforcement learning model to govern throttle and steering, the car demonstrates how a model trained in a simulated environment can be effectively deployed in the real world. Equipped with various sensors and cameras, the AWS Deep Racer is capable of perceiving its surroundings, including the layout of the track and any obstacles present. This data is then transmitted to a reinforcement learning model, which utilizes it to compute the appropriate steering and throttle commands for the car based on its current state and the environmental conditions. Not only is the AWS Deep Racer an enjoyable and interactive tool for gaining knowledge about autonomous driving and reinforcement learning, but it also showcases the possible applications of RL in various practical fields, including transportation, logistics, manufacturing, and robotics.

(CSE 540: ML and Society and Rage Against the Machine: HIS 419)

Combating White Supremacy in Hiring Discrimination

• Kashyap Deveshbhai Soni
• Ifeanyichukwu Frank Okafor
• Michael Caliendo
• Dhiraj

Vision We imagine a world without white supremacy would be a world where individuals of all races and ethnicities would have equal opportunities, feel safe, and be treated with respect. A world where the systems and institutions would not be designed to benefit one group of people.

(Reinforcement Learning Project (CSE 546))

Team 16

• Viveka Siva Naga Nithin Annam
• Venkata Sai Divya Pallineni
• Niharika Vikram

Evaluating and Comparing Advanced Reinforcement Learning Algorithms on Cartpole and Pong Environment.

(CSE 540: Machine learning and society)

Improving recommendation algorithms to fight white supremacy

• Aditi Bobde

Impossible project based on the discussions happened in the course CSE 540, throughout the semester. We are trying to solve the bigger problem of white supremacy by choosing a subproblem of radicalization on internet. We aim to solve it by discussing ideas on how to improve recommendation algorithms and dealing with issues caused by the friction between technology and society when not used right.

(CSE 546 : Reinforcement Learning)

Team - 17

• Thilak Reddy Dharam
• Dhanush Kumar Reddy Yarragonda

Performance Analysis on various Multi-Agent Environments

Exploration is good, but to what extent?

• Vivek Kuchibhotla
• Anurima Vaishnavi Kumar

In this research project, we investigate the efficacy of advanced exploration strategies and state space optimization techniques in reinforcement learning, specifically focusing on complex Atari game environments. By integrating the Upper Confidence Bound (UCB) algorithm with Deep Q-Networks (DQNs), we aim to balance the exploration-exploitation trade-off and improve AI agents' performance in high-dimensional state spaces. Our methodology employs an encoder-decoder model to learn a compact representation of the high-dimensional Atari state space, allowing for more efficient state representation and action selection. We incorporate a multi-head attention mechanism in the model to reduce noise and focus on essential components of the state space, enhancing the model's ability to learn relevant features and dependencies. By pre-training the encoder and freezing its weights during the training phase, we optimize the training process, reducing co-variate shift, accelerating convergence, and improving the model's stability and performance. The results of our research demonstrate the effectiveness of combining advanced exploration strategies, such as UCB, with state space optimization techniques like multi-head attention and encoder-decoder models in enhancing reinforcement learning models' performance in complex environments. This approach leads to more efficient training, higher reward returns, and increased adaptability of AI agents across various tasks and environments. Our findings contribute to the growing body of knowledge in reinforcement learning and have implications for the development of sophisticated AI agents capable of addressing a wide range of real-world challenges and applications.

((CSE 546))

Transforming a Robot Car into an agile Ball-Chasing Machine: A Comparative Study of Deep RL Algorithms

• Atharv Dabhade
• Sai Sharan Thirunagari
• Shivani Shingne

In this project, our aim is to enhance the capabilities of a robot car, enabling it to become an agile ball-chasing machine that pursues and reaches its target goal. We employ various state-of-the-art Deep Reinforcement Learning (RL) algorithms, including A2C, DQN, PPO, and DDPG, and conduct a comparative study to evaluate their effectiveness in training the agent. Through this research, we aim to provide insights into the strengths and performance characteristics of each algorithm in achieving the project's objective.

(CSE446)

Team-35

• Adil

Performance Evaluation of Different Multi-Agent Reinforcement Learning Algorithms for Cooperative-Real World Applications

(Independent Study project)

Pick and place robot using computer vision

• Raj Khatik
• Prof. Farshad Ghanei

Detecting objects using computer vision and manipulation of robot arm to pick and place the object

(700 Independent Study)

Educational Web App to Visualize Reinforcement Learning Methods

• Shraddha Shekhar

An interactive Web application for learning Reinforcement Learning methods by exploring different algorithms, environments and hyperparameter settings. The application would serve to simplify the complexities of reinforcement learning, promoting a more intuitive and effective learning process. By providing students with the ability to experiment with various settings, they can gain a deeper understanding of the subject matter, ultimately leading to improved comprehension and retention.