Visual MIMO is based on multiple light emitting arrays and photodiode arrays in visible light communication. Challenged in both user experience and system goodput (accuracy and throughput), we explored factors contributing to screen flicker perception, designed a screen-camera communication system with content-adaptive embedding method to achieve flicker free and high goodput results.

Publication​s

"High-Rate Flicker-Free Screen-Camera Communication with Spatially Adaptive Embedding", in proceedings of INFOCOM 2016, AR: 18.25%,
Viet Nguyen, Yaqin Tang (co-primary author), Ashwin Ashok, Marco Gruteser, and etc. Best-in-Session Presentation Award
[bib] [paper] [slides]

"Content Adaptive Encoding Method for High Frame Rate Screen-Camera Communication". M.S. thesis, January 2016.
[paper] [slides]

Motivated to achieve high accuracy and low energy consumption for indoor localization, we designed a InfraRad system that uses hybrid radio-optical technique. From triangulation, localization can be achieved if knowing some angles or lateral length.

My contributions are:

proposed the algorithm for the receiver to do localization;

designed the strategies for our system deployment in the competition evaluation area.

I worked with Ashwin Ashok and Chenren Xu for Microsoft indoor localization competition - IPSN 2014. Our system is demonstrated using a receiver apparatus where the radio-IR receiver with an array of photodiodes is mounted onto eyeglasses.

To enhance conventional touchscreen user interaction scenarios, there's a novel Swept Frequency Capacitive Sensing technique - "Touche", which provides rich touch and gesture sensitivity to a variety of analog and digital objects. This method is realized by checking the capacitance change in the circuit because of body movement, then classify the signal response into different category. There're applications, such as security password, remote control etc.

My contributions are:

Duplicated the Touche work with launchpad MSP-EXP430G2 and built surrounding circuit on breadboard;

Completed firmware programming for the system.

Related works: Touche, OpenCapSense.

Challenged by low GPS accuracy in dense urban areas, we need to find out new solutions to improve localization accuracy. In general, we propose a system with precise localization by using smartphone cameras capturing its surrounding scenery and take a match with Google Street View images.

My contributions are:

Analyzed camera calibration for Nexus 5.

Tracked image features using Lucas-Kanade method;

Detected feature motion of consecutive images based on pinhole camera model.