Click on the title of each project to get detailed description/report

 (Successfully defended on Nov 10, 2015)
 Human Factors Analysis of Automated Planning Technologies for Human-Robot Teaming

Co-Chair: Dr. Subbarao Kambhampati
Co-Chair: Dr. Yu Zhang
Committee: Dr. Georgios Fainekos
Committee: Dr. Nancy Cooke

Graphite as a Service.pptx

2015 SUMMER INTERN PROJECT @ Electronic Arts Inc

During my summer internship @ EA, I created a Service for the open source architecture Graphite using vRealize Automation Application Services 6.2, where I automated the complete application provisioning in the cloud including deploying, configuring and updating the application's components and dependent middle ware platform services on infrastructure clouds.

In this project, I investigate the incorporation of a general planning capability into the robot to facilitate peer-to-peer human-robot teaming, in which the human and robot are viewed as teammates that are physically separated. My results show that peer-to-peer teaming reduces cognitive loads from objective measures and it does not reduce situation awareness for short-term tasks.

Most of the mobile phones currently in use have less processing capability. A trivial solution to this resource poverty problem is compute offloading.In this project I have implemented a communication mechanism for providing value added services on an opportunistically formed Stationary Vehicular Cloud by exploiting the idle vehicular resources that, now has become a thing of commonality in cities.

The automatic question generator for Tamil is a Language processing system which is used to generate questions for valid Tamil sentences which follows the grammatical rules and constraints imposed by the language. This Natural Language Processing (NLP) system combines the knowledge about the language and application domain to automatically generate questions for Tamil sentences.

This project explores a security threat -DoS (Denial of Service Attacks) that can affect cloud systems and interconnected components on Internet. In my experimental setup I have tried to simulate DoS attacks from an attacker VM i.e. a particular host machine inside the Mininet environment. The attack procedure involves sending a large volume of traffic to the target host, a behavior which is not normal, or using varying packet size to load a target virtual machine.

In this project, I developed a secure banking system which ensures safe online banking that provides customers a secure banking facility by incorporating different security techniques such as one time password, public key infrastructure in addition to TLS/SSL. This project was designed as a web application using JSP, Java Spring and Hibernate frameworks, and MySQL database.

In this project I implemented the Vector Space Model (using just TF weights) to rank the documents and I also implemented the Authorities/Hub computation. I finally implemented K-means algorithm to obtain the top- N documents using TF/IDF and clustered the results using the simple K-means algorithm which randomly picks the initial k centroids.

The main aim of this project is to develop a tenant-aware SAAS application which enables multi-tenancy, customization and scalability. I also designed a database schema, system architecture and GUI to support SAAS with emphasis on same code base for multiple tenants with customizable services and tenant-aware front end components.

There are M transmitter and N receiver stations. Given a matrix that keeps track of the number of packets to be transmitted from a given transmitter to a receiver. If the (i; j)-th entry of the matrix is k, it means at that time the station i has k packets for transmission to station j. During a time slot, a transmitter can send only one packet and a receiver can receive only one packet. The problem is to find the channel assignments so that maximum number of packets are transferred from transmitters to receivers during the next time slot.

 The main aim of this project is to demonstrate through experiments that it is possible to construct a histogram intersection kernel SVMs (IKSVMs) with run time complexity of the classifier being equal to log (number of support vectors) as opposed to linear complexity that is achieved through the standard approach.

In this project, the main aim is to build a Recommendation Engine for the Netflix Movie Data-set and address the problem of cold-start, i.e. addressing the problem of recommendation when there is a sparsity of past user data. The recommendation engine is based on Collaborative filtering method and it
uses Online Matrix Completion technique to overcome the data sparsity problem of collaborative filtering method.

The main aim of this research project is to perform an analysis of human factors on the effectiveness of proactive support for human aware planning in human robot teaming. We performed our evaluation in a simulated USAR task, in which the efficacy of teaming is not only dependent on individual performance, but also on teammates’ interactions with each other. In this task, the human teammate is remotely controlling a robot while working with an intelligent robot teammate ‘Mary’. Our main result shows that the subjects generally preferred Mary with a Goal and Intent Recognition - GIR ability (compared to Mary without this ability), with which the human cognitive load was indeed increased (albeit not significantly) even though the subjects appeared to interact less with Mary.


The main aim of this project is to familiarize myself with the NAOqi SDK of Webots simulator for the Nao Robot. I used that simulator and made use of the Mini RoboCup environment on Webots to implement a basic task (locate, move to and kick a ball to score a goal) in a robot soccer game.

Mini Robo cup

Mini RoboCup Standard Premier League - Webots simulation demo video


13) Mining Association Rules in Folksonomies

Mining Association Rules in Folksonomies

14) Integrating Folksonomies with the Semantic Web

Integrating Folksonomies with the Semantic Web