about | CV 

I got my PhD in Considerate Systems from CMU and was advised by Ted Selker and Ian Lane. I graduated from the ECE BS/MS program at Georgia Tech in Dec 2007. Go Yellow Jackets!

research interests

“The intransigence of human behavior has emerged as the root of most of the world’s biggest challenges.”  — Thomas Goetz, “The Feedback Loop,” Wired July 2011

I am interested in understanding how AI can help in nudging or steering humans to make better economic, social and health choices. For such systems to become a reality, they will need to successfully navigate social spaces, and engage the user with acts of motivation, persuasion and negotiation. In other words, they will need to be socially intelligent. As a step in this direction, my research focuses on Considerate Systems. These add a response layer to context-aware systems, allowing them to behave appropriately with respect to a user’s affective and social state. The next step would be to explore architectures that accommodate affective and social aspects of cognition.

projects

Built an audio-only teleconference system to investigate collaborative decision making, mediated by an agent. The paper presents how the system was successful in getting people to contribute more equally, and also in reducing the disruptions caused by extraneous noise.

Built two android apps to evaluate novel pain assessment technologies, compared with conventionally used paper-based BPI, to demonstrate the feasibility of a flexible and robust extension to the Kinnexxus platform, supporting elders’ everyday monitoring of pain. The project examines issues regarding effectiveness and usability of the novel technology.

Built a light sensor and OOK transmitter at 27 MHz using a TI MSP 430 μcontroller and a custom-built loop antenna. The paper presents the power savings that can be achieved by leveraging the power line as opposed to an on-board power hungry radio, a limiting factor in the wide-spread deployment of wireless sensor networks today.

Designed and built a toy instrumented with FSR sensors, to differentiate between an adult’s and a child’s grip. The paper demonstrates the novel method we used in the design process to inform sensor placement. Pending submission to CHI 2010 Work-In-Progress.

Built a light sensor and OOK transmitter at 27 MHz using a TI MSP 430 μcontroller and a custom-built loop antenna. The paper presents the power savings that can be achieved by leveraging the power line as opposed to an on-board power hungry radio, a limiting factor in the wide-spread deployment of wireless sensor networks today.

Evaluated the temporal stability of leading RF technologies used for fingerprinting in various indoor localization systems, including WiFi, GSM, CDMA, FM, and WPLP. The paper compares the standard deviation for each system over a period of at least four days in three different locations.

Prototyped a human detection system to look into techniques for their deployment on mobile robots. Histogram of Oriented Gradients (HoG) feature detector and a Gentleboost classifier were implemented in Matlab and Python and tested on the INRIA person dataset. An exhaustive literature survey helped to identify future research avenues.

Investigated the deployment of the open source Robotic Operating System as middleware in the Aware Home. Explored the possibility of having multiple ROS Masters (name service) be aware of each other so as to increase reliability and distribute processing. Proof of concept was done using X10 HA sensors.