Thanks to all our BOOM 2017 project teams!
Rowena Chen, Nupur Kale, Tapasya Kothapally, Crystal Liu, Anagha Todalbagi
We implemented all the steps involved in human computer interaction design through this project to build a solution to find accessible entrances for users to assist in independent living. Individuals with mobility difficulties have been found to face a lot of difficulty while commuting from one place to other. This problem has become a major impediment to their independence in executing their daily activities and travel. On conducting extensive user interviews for understanding specific problems and goals of these users, it was realized that individuals with mobility difficulties bound on a wheelchair faced severe problems in their commute in terms of accessibility and navigation. To address this problem, we have come up with niche accessibility and navigation mobile application ‘Accessimaps’ to empower our users with accessibility points and navigation routes of specific buildings for them better and pre-plan their commute to various destinations. The application also lets our users save particular routes to let others access them without prior search feed and thereby help the community.
Saloni Joshi, Nupur Kale, Luis Fernando Plaz, Yilu Sun
Obesity is the most outstanding problem during pregnancy, resulting in significant consequences, such as difficult deliveries and gestational diabetes. In the US 70% of women that are overweight before pregnancy end up in the obese category. To tackle this issue we propose an Alexa Skill to help these women manage and track their weight as well as connect with the pregnant community.
Benefits of Resource Disaggregation in Datacenters
The slowdown of Moore’s law has led to surfacing of several fundamental limitations of today’s server-centric architectures. As a result, a new computing paradigm is emerging — a disaggregated architecture, where each resource type is built as a standalone “blade” and a network fabric interconnects the resource blades. While beneficial from the computer architecture perspective, conventional wisdom suggests that decoupling of resources will lead to performance degradation for legacy systems and applications. Thus, widespread adoption of such architectures is currently gated on systems community building systems and applications optimized for this emerging computing paradigm. We present preliminary evidence that the above blockage may be unfounded. Our main insight is that disaggregated architectures offer finer-grained resource multiplexing and improved “packing” of jobs compared to server-centric architectures. We show that, for many legacy systems and applications, performance degradation due to resource disaggregation can be completely offset by improvements due to higher resource utilization in disaggregated architectures. (Paper submitted to the Usenix HotCloud '17 workshop)
Charisse Foo, Sarah Le Cam, Eric Jung, Chris Martin, Alex Pomerenk, Katie Stinebruner
Brought to Light is a desktop game that combines the puzzle and thriller genres, dropping you in a ruin with no map and no way to deal with enemies except candlelight. Instead of favoring light or shadow, strategically switching between light and dark is needed to explore the environment, avoid and interact with enemies, unlock doors, and eventually escape each level. The light and dark offer different advantages over the environment, with monsters turning from helpful tools to deadly enemies when the light changes. Alternating between the light and dark states to solve the puzzles and find the escape, you are never safe.
CiceroDB: Optimizing Voice Output of Relational Data
Mark Bryan, Jiancheng Zhu
Recent research on data visualization aims at automatically identifying the best way to represent data on visual interfaces. Supported by a Google faculty Award, the Cornell Database Group is currently studying the complementary problem of "data audiolization". The goal here is to optimize the way in which structured data is represented via audio interfaces. This problem setting is motivated by emerging devices such as Google Home or Amazon Echo that interact with users primarily over voice based interfaces. In this demo, we present a very early prototype of CiceroDB, an experimental database system that will offer different modules to translate data into voice output. In the current prototype, we model voice output optimization as an NP-hard optimization problem with the goal to reduce speaking time for outputting a given data set (subject to user-defined output precision constraints). The current prototype applies either exhaustive or polynomial time algorithms with approximation guarantees to solve the voice output optimization problem and reads out the optimized results. We currently reduce speaking time by removing redundancies which is appropriate for outputting relatively small data sets. In future work, we will explore methods to transmit information that allow users to get insights about the distributions underlying larger data sets.
Sacheth Hegde, Daniel Liu, Gautam Ramaswamy
While many cloud providers appear to provide identical services, actual performance may vary greatly based on user applications, underlying technologies, and other factors. In this project, we explore the difference between many cloud services such as Fractus, Amazon EC2, and Microsoft Azure and try to understand why those differences exist. We will explore performance related to both the network and storage. We also mimic real world usage with a test application called Ceph.
Dheeraj Chakilam, Peter Mocarski, Karun Singh
Constableaux is an automated theorem prover and educational tool based off of the method of analytic tableaux. Given a formula in propositional logic, Constableaux proves whether or not it is a tautology, and generates an interactive, step-by-step tree visualization of a proof or counter-example. Our project aims to help students who have trouble grasping the underlying intuition behind proofs of propositional logic by presenting them through a visual medium. Propositional logic is the foundation for theoretical computer science, but we've found that many students, including us, have trouble following proofs with case analysis and heavy branching. The size of a page forces textbook writers to present proofs in a linear fashion, when the intuition behind such proofs is anything but linear. Our tool presents proofs in a tree-like structure, which more closely models the way we reason about such proofs. Additionally, students can run through a proof step-by-step, zooming into specific portions of the proof with explanations along the way. Constableaux was created as a cummulative project for MATH/CS 4860: Applied Logic, in Fall 2016, under the supervision of Professor Bob Constable.
Dae Won Kim, Jared Lim, Amit Mizrahi, Kenta Takatsu, Chase Thomas
The Cornell Data Science Training Program, a one-credit course offered through the Cornell Data Science project team, is a twelve-week unofficial course that focuses on manipulating data, visualizing trends, and implementing machine learning algorithms. Students, who are not expected to have any programming experience, are exposed to concepts through the R programming language, which is used to complete four assignments and two real-world data science projects. The goal of the training program is to give students practical hands on experience with data science, and to equip them with the tools needed to take on their own projects. Topics include linear and logistic regression, decision trees, bayesian classifiers, clustering algorithms, text analytics, big data tools, and ensemble methods. Over 100 students are currently enrolled.
Cristian Alonso, Gillian Boehringer, Daniel Fayad, Keshav Iyer, Todd Kirchhoff, Alana Marzoev, Shiv Roychowdhury, Alexandra Voinea
We are an engineering project team competing in SpaceX's Hyperloop Pod Competition.
Varun Belur, Emma Carpenter, John Draikiwicz, Matthew Filipek, Jacob Glueck, Hunter Goldstein, James Haber, Scott Holmdahl, Seetha Kolli, Nicole Polemeni-Hegarty, Andrew Showers, Sean Viswanathan
The Cornell Mars Rover project team designs an innovative Mars rover to compete in the University Rover Challenge. We are an interdisciplinary student-run team that brings together talented minds from engineering, science, and business. Together, we foster creativity through the development of our competitive design.
Matthew Barker, Daniela Gottesman, Jesse Mansoor, Vinisha Mittal, Divyansha Sehgal
A course review website for Cornell students, by Cornell students.
Wenhui Feng, Peter Ferenz, Scott Hamill, Jonathan Jalving, Peter Moegenburg, Paige Trexel, Kai Weng Wong, Le Yuan, Jiahao Zhang
This project competes in the RobotArt competition. Our team collaborates with Cornell Autonomous Systems Lab (cornell-asl.org) to implement an autonomous, artistic behavior on KUKA Youbot. The implemented system is able to create visually interesting paintings on canvas.
Cindy Chen, Michelle Ip, Shoshaunah Jacob, Kevin Mao, Yeon Soo Park, Ken Shimizu
Cornell Steel Bridge is an interdisciplinary project team that designs, fabricates, welds, and constructs a 20’ multi-member steel bridge for the annual ASCE Student Steel Bridge competition. During our design phase, we extensively use modeling software such as MASTAN and Autodesk Fusion 360 to analyze element forces and capacities, connection strength, and failure modes in buckling and lateral stability. MASTAN is a matrix structural analysis program that serves as our primary tool for linear elastic finite element bridge modeling, while Fusion 360 functions as a key component for optimizing our bridge connections. Using these applications, we completed our bridge design for the 2017 competition during the fall semester, and both programs have optimized the iterative process for the analysis phase of our work.
Tennyson Bardwell, Zander Bolgar, Cuyler Crandall, Julia Currie, Kerri Diamond, Joseph Featherstrom, Kevin Guo, Alan Lee, Artina Maloki, Mihir Marathe, Kristina Nemeth, Cora Peterson, Noel Picinich, Anthony Viego, James Wu
We design and build fully autonomous vehicles for competition and research.
Nikita Ermoshkin, Troy Joseph, Shirley Kabir, Alexandra Katague, Kevin Kruempelstaedter, Rahul Madanahalli, Grant Mulitz-Schimel, Brendan Quinn, Joshua Williams
We are an interdisciplinary project team that combines aspects of computer science, engineering, and business.
Joseph Antonakakis, Annie Cheng, Daniel Li, Amit Mizrahi
It's tough to find interesting events on campus since sources of information are decentralized: most people rely on Facebook event recommendations. We've created a system that uses Facebook data (and other event data sources if need be) as the basis of an IR system for searching public Cornell events by time, location, organization, keyword, etc. In addition to the events finder, we built a system that will apply machine learning to events data to help event organizers intelligently plan on-campus events by determining the location and time of the event that maximizes attendance.
Abenezer Lemma, Yeon Soo Park, Tiffany Zheng, Kenzie Zou
We designed, built, and programmed an economical and easy-to-use cupcake frosting machine. Our cupcake froster can create 3 major cupcake designs: spiral, flower, and Pokéball. It utilizes a Photon microprocessor, DC motors, stepper motors and drivers, laser-cut acrylic, 3D printed parts, and ready-made structural parts. Frosting is pumped from a syringe by using a lead screw and an acrylic plate. To create patterns, we have a Lazy Susan rotating base and a sliding platform, both powered by stepper motors. Together, we have a simple and elegant machine that creates interesting designs.
Depth Perception for Autonomous Underwater Vehicles
Yuji Akimoto, Ryan Butler, Nolan Gray, Dae Won Kim
Our project is based on our first-place project in the BigRed //hacks Sp '17 Data Science competition. The idea is very simple: using footage from the CUAUV team's underwater vehicle to create an machine-learning algorithm that can correctly identify static target objects in the water (colored buoys) and determine the machine's position and displacement. The project involves using several clustering algorithms and statistical techniques to enhance the object recognition capabilities, which would help complement or replace the sensors that already collect relevant information on position and movement. Our goals in the project were to achieve a very accurate and reliable algorithm that can classify objects based on their color, write a script that can calculate relative distance based on the size of the recognized object in the camera's visual field, and be able to run these processes in real time - to ensure full-autonomy of the vehicle.
Cameron Drew Chafetz, Jonathan Chen, Kenneth Lee, Kyle Sampson, Susan Wenshan Zou
Magic Studios, a team of 6 undergraduate students, collaborated to develop this desktop game. Deserted is a turn-based tactics game revolving around the character Dara who forms alliances with spirits to fight in a tournament challenging the upper class.
Extending Cultural Comprehension
Hana Cummings, Juliana Hong
As part of research in the Intercultural Communication Lab, we created two Google Chrome extensions designed to increase understanding among speakers of different languages on Facebook. We use IBM Watson's Alchemy API and TextRazor API to conduct sentiment, emotion, and entity analyses on foreign posts to provide users with information to improve comprehension of cultural context.
The FactChecker: Verifying Natural Language Summaries of Relational Datasets
Daniel Liu, Niyati Bharat Mehta, Weicheng Yu
Fake news are perceived as a growing problem and purely manual approaches to fact checking fail to scale. The Cornell Database Group has recently started to develop semi-automatic tools that support users in checking natural language descriptions of relational data. This scenario is more and more common nowadays as newspapers, articles as well as blog entries refer to statistics that are derived from large datasets. We analyze the text surrounding numerical statistics to extract keywords that hint at the SQL queries underlying the postulated numbers. We efficiently merge the computation of alternative query candidates and analyze the match between text and data. We finally envision a tool similar to a spell checker that marks up verified or suspicious text passages for users while at the same time requesting user feedback about text passages that are difficult to analyze.
Fridge Friend - A Food Waste Solution
Victoria Beall, Lola Legrand, Michael Luzmore
Our project is focused on reducing consumer-side food waste by allowing people to better visualize the contents of their refrigerator. By raising awareness of food consumption habits, we hope to better inform purchasing behavior and reduce the environmental ramifications of food waste.
Ishaan Jhaveri, Margaret Meehan, Mike Merrill
GraffitiBot is a portable device that can be used to create art in public places. It consists of a drawing head, two wall-mounted stepper motors controlled via stepper drivers, and a Photon microcontroller. It is capable of drawing on virtually any smooth surface including blackboards, whiteboards, glass, paper, and even painted walls. GraffitiBot can also draw with nearly any implement: pens, markers, pencils etc. GraffitiBot can be used to inspire education, advertise a project, make public announcement, or create art. The project is useful because it promotes artistic and technological literacy while sparking a dialogue about the boundary between art and technology. Is it a tool or a creator? Is its work derivative or original? GraffitiBot revisits old questions about the role of machines in our lives.
Sagar Akre, Patrick Baginski, Charmi Sunil Mehta, Adisa Soren, Kaiwen Zhong
HealthPass is a digital application that allows for patients to take control of their own healthcare data, learn about their health status and seamlessly transfer their medical information to any healthcare facility around the world. We developed this application in response to data gathered during interviews with international students who expressed their frustration in dealing with the healthcare system in the United States. These patients told us about not being able to seamlessly transfer their basic medical information such as immunization and allergy records and not fully understanding what practitioners were communicating to them. To develop the application, we employed the use of a user-centered design process that included gathering insight from our target user population through feedback sessions. This allowed us to improve on our prototypes and through several iterations, we were able to design an application that most effectively responded to the needs of our target users.
Hippo: a HIPAA-compliant Video Chat App
Sonia Appasamy, Frank Chan, Erin Chen, Albert Chung, Ben Edwards, Lillyan Pan
Cornell Engineering World Health is building a HIPAA-compliant telemedicine platform for doctors and patients to conduct virtual appointments through video conferencing. The platform aims to provide health care to patients in remote or under-resourced areas.
Marc Nicholas Choueiri, Schuyler Duffy, Sanjay Guria, Anagha Todalbagi, Yixiao Wang
To assist with independent living, we have built a soft robotics device to pick up things for people who are unable to use their fingers to pick up things either because of old age or an accident. This is a wearable pneumatic device activated by a touch sensor which will be in the user's sleeve. The-Helping-Hand will ensure that they will be able to pick up things as before. We will also show The-Tilting-Table, a device that assists in independent living by helping the user tilt anything that is placed on this table, either a water jug to pour water or a shoe to bend and tie the shoelaces.
Erin Chen, Han-Wen Chen, Annie Cheng, Brett Clancy, Adam Wang, Simon Wang, Yiting Wang
An interactive website integrated with PredictionIO to predict taxi demand based on past usage and analyze trends depending on time, location, and weather.
Mengjia Guo, Michael Henning, Joel Hoover, Mauricio Moreyra, Nicole Pirringer, Max Rademacher, Reid Michael Wade
Laser Penguins is a multiplayer networked game for iOS and Android devices. It is a top-down shooter with bullet ricochet and a fast paced, bite-sized play time.
Anant Agarwal, Ryan Butler, Joseph Chuang, Jared Lim, Amit Mizrahi, Amol Tandel, Chase Thomas, Zhihan Wang, Kevin Yang, Yungton Yang
The Kaggle Competition Team is a subteam of the Cornell Data Science project team. We developed an ensemble model for predicting 2017 March Madness tournament results on Kaggle, an online data science competition platform. Each team member developed their own logistic regression model with a unique set of engineered features, using metrics such as the ELO rating system and Massey Ordinals. We then combined these into one logistic regression model that uses the outputs of each individual's model as features.
Memory Disaggregation for Rack Scale Computers
Mohammad Saifee Dohadwala, Amardeep Singh Manak, Shaan Ramesh Shetty
Modern computers have processors coupled with memory. But memory utilization is not the same across all servers. Depending on various applications running on server, some may be under-utilizing the memory while others may be in need of extra memory. With recent advances in the speed of networks, we can now start thinking of decoupling the processor with the local memory. The servers would all be connected on a network and would share memory. This makes it easier to scale memory as per the application. This is the idea behind Memory dis aggregation.
Catherine Boutwell, Jacob Cooper, Daniel Sainati, Apurv Sethi, Melody Jean Spencer, Hezekiah Thompson
The player takes control of Yuri, a Russian rock climber who wants to follow in the footsteps of his namesake and reach the stars. Using only his own body, Yuri must scale treacherous terrain, managing both his energy and unwieldy limbs, to reach space.
Napkin Folding Machine
Olav Imsdahl, Sierra Stone, Olivia Wherry
We built a prototype in Rapid Prototyping that folds napkins. It starts with a flat napkin, and has wooden flaps passing it down an assembly line. It pops up a napkin at the end.
Ryan OHern, Yanir Nulman, Sean Viswanathan
The field of swarm robotics is increasingly popular, however, human swarm interaction is poorly understood. Our goal is to improve understanding of how one person can control a swarm; how many people can work efficiently alongside swarms; and what metrics can be used to evaluate such systems. Specifically, based on an old system for autonomous collective construction of 3D structures, we have developed autonomous agents able to assemble structures towards a user-specified goal in the online game “Minecraft”. Minecraft is a natural environment for experimenting with construction techniques, and has a set of open API's to allow external intervention. We have developed a simple browser tool to let users design their desired structure; we then compile this into a custom data structure that can be used by the simulated robots. We also allow human users to join the game to work alongside the agents, and introduce a human foreman who can observe and manage the workforce through an overlay GUI to ensure completion of the correct structure. This framework will lay the groundwork for future studies of how robots and humans may collaborate efficiently in complex scenarios, and how we can enable a layman user to operate a large swarm of ubiquitous robots.
Optimal Review Scheduling for Human Learning
Igor Labutov, Sid Reddy
In the study of human learning, there is broad evidence that our ability to retain information improves with repeated exposure and decays with delay since last exposure. This plays a crucial role in the design of educational software, leading to a trade-off between teaching new material and reviewing what has already been taught. A common way to balance this trade-off is spaced repetition, which uses periodic review of content to improve long-term retention. Though spaced repetition is widely used in practice, e.g., in electronic flashcard software, there is little formal understanding of the design of these systems. Our paper addresses this gap in three ways. First, we mine log data from spaced repetition software to establish the functional dependence of retention on reinforcement and delay. Second, we use this memory model to develop a stochastic model for spaced repetition systems. We propose a queueing network model of the Leitner system for reviewing flashcards, along with a heuristic approximation that admits a tractable optimization problem for review scheduling. Finally, we empirically evaluate our queueing model through a Mechanical Turk experiment, verifying a key qualitative prediction of our model: the existence of a sharp phase transition in learning outcomes upon increasing the rate of new item introductions.
Sophie Huang, Daryl Sew, Charles Tark, Amy Wang, Ning Wang, Peaky Yuter
A paper airplane has escaped from its cold, claustrophobic office prison and aims to explore the open world. Interested in the little plane’s adventure, you accompany it using your wind powers to guide it through the unfamiliar and sometimes hostile environment. We created this game in CS4152 with Walker White, and are continuing to work on it.
Plant Operations Smartphone Tracker
Jose Castro, Matthew Limjoco, Fawn Wong
The Plant Operations Smartphone Tracker (POST) is a suite of applications for the collection, analysis, optimization, and visualization of water quality data from rural water treatment plants in central america. With POST, plant operators and implementation partners can easily track plant parameters and visualize performance over time. Data collection is fully robust to intermittent internet connection, and optimal for use in remote laboratory settings. POST is installed in several water treatment plants in Honduras and has collected over 4000 submissions from plant operators since Spring 2016.
Ian Arawjo, David Li
Israr Mahmood, Alexander Rucker
In our project, we investigate the architectural implications of tunneling the AXI bus over a network between nodes, for the purpose of memory and I/O disaggregation. We will test our assumptions using a prototype running on an FPGA.
Resolve & Vent
Sean Kim, Dou Mao, Cassie Wang, Ally Wu
A stress management app that creates an anonymous community for people to resolve problems and to vent about their stress.
Christina Cicileo, Rory Giszter, Sophie Nicolich-Henkin, Kofi Otseidu, Jake Padilla, Kevin Yihkai Shih
Shoal Storm is a stealth-action side scroller developed by Intertidal Games. Shoal Storm is written in Java using libgdx, Box2D, and Gradle. The game features dynamic collision detection, enemy AI, and elegant fish swarming.
Nupur Kale, Sritapasya Kothapally
Scheduling and making a note of announcements and deadlines through various learning management systems (LMS) is a cumbersome process. It is difficult to browse through different LMS for every coursework to schedule or create events on your generic calendar. The web application ‘Smart Planner’ addresses this problem and provides a convenient one click solution for it. With this application, we have automated the process of searching deadlines through a search and get algorithm in gmail to gather the data and put it in the app. For this project, we have built a web based android application called ‘Smart Planner’ to schedule intelligently while containing time. To come up with this end result, we conducted three rounds of user interviews, design iterations and finally developed it on ionic frameworks using angular.js. In addition to this we integrated google sign in functionality, gmail API search (fetch deadline data from user’s gmail) and google calendar API (to auto-populate on calendar). We have incorporated client recommendations and user feedback at every stage of our project. When you open this application it will automatically list your upcoming deadlines, announcements from the various LMS without actually opening them. We have also provided a calendar that would be automatically populated on hitting the calendar view feature on this application. In addition, we have also incorporated a checkbox feature for the list of deadlines to check them done on completion.
Stephanie Chan, Jeffrey Huang, David Landy, Arlena Wu, Jiaming Zhang
Sync is a desktop computer game created for presention at the Game Design Initiative at the Cornell 2016 Showcase. The objective is to guide the lonely half note Minim through a world of music by syncing in time with rhythmic obstacles.
Austin Astorga, Matthew Barker, Mihir Chauhan, Annie Cheng, Derrick Ho, Sahil Khoja, Monica Ong, Shiv Roychowdury
An open-source iOS app designed and developed from the ground up to help Cornell students use the TCAT transit system.
Felix Chen, Wenhui Feng, Shiyu Wang, Yiting Wang, Le Yuan, Leezel Zamidar
Teddington is a Puzzle Platformer computer game built by a team of 4 programmers and 2 designers during the course CS3152. Game description: Memories fade with age. Staring into the empty gaze of the old lady lying next to him, Teddy wishes his owner recognizes him. Determined to build a commemorative photo album, Teddy ventures through the 19th century city Teddington to collect beautiful but forgotten memories. With your power to place items anywhere in Teddington, design the paths to help Teddy complete his quest.
Logan Allen, Mihir Chauhan, Jesse Chen, Annie Cheng, Dennis Fedorko, Ilan Filonenko, Sanjana Kaundinya, Ji Hun Kim, Jon Lee, Keivan Shahida, Tiffany Zheng
Tempo is the second app from CUAppDev. It allows users to post their song of the day and discover what others are listening to.
X-Docker: Truly Isolated Docker Containers
Nerla Jean-Louis, Nishad Mathur
The cloud network access to a seemingly infinite amount of compute and storage. What underlies the cloud is the notion of virtual machines, a layer of software that allows physical hardware processors to be multiplexed over many "virtual" machines. As a consequence, Virtual machines, and lighter-weight "containers" now power the cloud! This project aims to integrate a very popular container technology called Docker with with a new container called X-container. Docker is extremely popular but does not support true isolation which can lead to security concerns. X-container is a Xen based virtual machine infrastructure that attempts to blend the best of both worlds, the low overhead of containers with the security of virtual machines. This project aims to integrate Docker with the X-container runtime; providing truly isolated docker containers while minimising the overheads from spawning a virtual machine for each container. By using an X-container back-end with Docker we can maintain the advantages provided by lightweight containers while increasing security and isolation.