Atlanta, GA - *****
Joan Devassy accdad@r.postjobfree.com
joandevassy.me/
Objective
Find a full-time position in Computer Vision Research and Development.
Education
MS, Computer Vision & Robotics, Georgia Tech, Atlanta, GA.
2011–2013
Advisor: Professor Henrik Christensen
BE, Computer Science and Engineering, Anna University, India.
2006–2010
Advisor: Professor R. Prabhakar
Industry Experience
May Robotics Research Intern, Honda Research Institute, Mountain View, CA.
2013–Aug Implemented a Driver Assistance System with integration of person head-tracking with RGBD
2013 camera, GPS-IMU pose estimation with Kalman Filtering and spoken driver-dialog. The
system was first tested in simulation before porting successfully to a Honda Pilot car.
May Robotics Software Intern, Neato Robotics, Newark, CA.
2012–Aug Interned with the R&D team at Neato, to design and implement of a simulation software suite
2012 for the Neato XV series of robots which allowed scene simulation and logging to make testing
easier. The system was deployed in the test suites of the company and a partner firm.
Academic Experience
Current Research Associate, Georgia Tech Research Institute.
Working on Data Analysis and Machine Learning algorithms for Worker Safety (Georgia State
Research Grant)
Aug Graduate Research Assistant, Aerospace, Transportation and Advanced Systems
2013–Dec Laboratory, Georgia Tech Research Institute.
2013 Machine Learning and Data Analysis algorithms for a Worker Safety assistive technology
system. (Georgia State Research Grant)
Aug Graduate Research Assistant, Center for Robotics and Intelligent Machines, Geor-
2012–May gia Tech, Atlanta, GA.
2013 Pose-estimation for moving objects using an inertial unit and camera coupled by a Kalman
Filter. (Boeing Grant)
Jan Graduate Research Assistant, Mobile Robots Laboratory, Georgia Tech.
2012–May Algorithms and mathematical framework for robot navigation inspired by primates, using a
2012 monocular camera (Office of Naval Research Grant)
Aug Graduate Research Assistant, Aerospace Systems Design Laboratory, Georgia
2011–Dec Tech.
2011 Smart Grid Power-System Analysis. (General Electronics grant)
Publications
Teruhisa Misu, Antoine Raux, Ian Lane, Joan Devassy, Rakesh Gupta "Situated
Multi-modal Dialog System in Vehicles", CHI ’14, ACM Human Factors in Computing
Systems (submitted ).
Arkin, R.C., Dellaert, F., and Devassy, J., 2012. "Primate-inspired Mental Rotations:
Implications for Robot Control", IEEE International Conference on Robotics and
Biomimetics (ROBIO 2012).
R.C. Arkin, F. Dellaert, J. Devassy, "Envisioning: Mental Rotation-based Semi-reactive
Robot Control", Technical Report GIT-MRL-12-03, 2012
Anush S., Joan D. Mary, "Quantum Computing - An insight into present technologies
and future applications", International Conference on Digital Factory, 2008.
Technical Experience
In order of proficiency
languages C, C++, Scheme/LISP, Python
systems Linux (Ubuntu), Windows
tools ROS (Robot Operating System), Matlab, R, Octave, Emacs, L TEX, Bash Scripting,
A
Git
libraries OpenCV, PCL, OpenGL, OpenNI
robots Baxter, Pioneer P3DX, Willow Garage Turtlebot, Kuka Industrial Robot Arm, Kuka
Youbot, Scorbot ER III
Relevant Coursework
Georgia Tech, Atlanta
computer Computational Science and Engineering Algorithms, Computational Complexity
science
computer Computer Vision, 3D Reconstruction for SLAM and Mapping
vision
machine Machine Learning, Probabilistic Graphical Models
learning
robotics An Introduction to Robotics Research, Robot Planning, Autonomous Robotics, Applied
Estimation for Robotics, Artificial Intelligence
controls Linear Systems and Controls, Nonlinear Systems
Honors
Graduate Research Assistantship for the duration of Master’s program (2011–2013),
Georgia Tech
Kerala State Merit Scholarship for the duration of Bachelor’s program, Anna University
Best University Final Project of the Year 2009-2010 for ’Multilevel Image Querying
Using Discrete Wavelet Transforms’ (team of 4)
Best Project of the Year 2008-2009 for ’A Chess Playing Robot’ (team of 2)
Master’s Research (2012-2013)
Real-time Pose-Estimation Using Visual-Inertial Sensor Fusion - An IMU-
Asus camera sensor array that can be mounted on mobile robots to obtain drift-
corrected pose estimates in indoor environments over long distances.
Undergraduate Research (2009-2010)
Multilevel Image Recognition Using Discrete Wavelet Transforms - Use
Wavelet Decomposition and machine learning to implement a pyramid-based technique
for a real-time image search engine.Won the Best Undergraduate Project Award in
Anna University, 2010
Academic Projects
IMU Enhanced Monocular Visual SLAM for Mobile Devices - Developed a
parallel tracking and mapping algorithm to perform 3D reconstruction using an IMU
and a monocular camera.
Low Cost Lane Detection for a Mobile Robot with a Monocular Camera -
Combined Hough line detection and a bag-of-words approach to detect lane markings
and predict the best action for a moving vehicle.
Sketch-based Image Retrieval using Visual Similarity using PCA, HOG
and SVM - Used HOG descriptors extracted from a sketched image to learn and
retrieve the most similar images from the PASCAL VOC 2007 dataset.
Easy Hand-tracking using Shape Context Classification and RANSAC -
Wrote a hand-tracking algorithm using computer vision and machine learning for
detecting hands in a video and tracking them from frame to frame.
A Greedy Clustering Algorithm for Speeding up Plane Detection on
Cuboidal Objects - A greedy algorithm that took advantage of object geometry to
exploit plane constraints for object detection.
A Computer Vision based Planning Strategy for teaching a Humanoid
Robot Sword-fighting - Used a RGBD camera for tracking a sword wielded by a
human opponent and tried to predict the next best ’play’ for the robot.
A Poker Game Engine Using Bayesian Inference Built a poker game engine
modelling the game using Bayesian Inference.
SLAM in Dynamic Environments with Factor Graphs and Levenberg-
Marquadt Optimisation - Modelled dynamic environments for a moving robot
using factor graphs and then built a map by doing error-optimization with landmark
measurements.
Implementation and Analysis of Cache-Aware Hash Tables - Exploited cache
characteristics for getting a high-performance hash table that was faster than cache-
oblivious ones.
A Chess Playing Robot aided by Artificial Intelligence and Computer Vi-
sion - Wrote and designed a chess engine that made game strategy using coin-
recognition on the chessboard.