Stephen J. Guy - Homepage

Research Assistant

GAMMA Group, UNC-Chapel Hill

Email: sjguy[AT]cs.unc.edu
Phone: (919) 962-1973
Department of Computer Science
CB #3175 Sitterson Hall, Rm#357
UNC Chapel Hill, NC 27599-3175
United States of America

News

1/1 - I've been accepted at the physics journal Physics Review E! [PDF] [Web]
12/31 - My work with Jur van den Berg on controling complex robotic systems under uncertaintly is accpeted at ICRA. [Web] [MP4]
12/16 - My work with Sujeong Kim on model dynamic agent personalities was accepted at I3D. [Web]
Check it out my updated CV.

Welcome

Hi, I'm Stephen. I'm currently a computer science graduate student at UNC-Chapel Hill where I work as a researcher with the GAMMA group. My research interests are pretty diverse, and I'm excited to do anything as long as the project is engaging. Over the past few years my focus has been in crowd simulation, robotics, sensing & uncertainty, and parallel/data-intensive computing.

Here is a description of some of the projects I've worked on.

Short Bio

Stephen J. Guy is a Research Assistant in Computer Science at UNC-Chapel Hill. He is currently pursuing his Ph.D. under the supervision of Dinesh Manocha and Ming C. Lin, and is expected to graduate in the Spring of 2012. He received his Computer Science M.S. degree from UNC-Chapel Hill in 2009 and his B.S. in Computer Engineering from the University of Virginia in 2006. He has also held several internships including over six months spent at Intel's Throughput Computing Lab.

Bibliometrics (according to Google Scholar)

• h-index: 6 (see why)
• g-index: 12
• publications: 25 (peer-reviewed conference, journals & workshops)

Selected Papers (View All, Google Scholar, DBLP Link)

LQG-Obstacles: Feedback Control with Collision Avoidance for Mobile Robots under Uncertainty J v.d. Berg, D. Wilkie, S. J. Guy, M. Niethammer, and D. Manocha IEEE International Conference on Robotics and Automation (ICRA), May, 2012. [WEB]

Simulating Heterogeneous Crowd Behaviors Using Personality Trait Theory S. J. Guy, S. Kim, M. C. Lin and D. Manocha ACM SIGGRAPH/Eurographics Symposium on Computer Animation (SCA), Aug 2011. [To Appear] [PDF] [WMV]

The Hybrid Reciprocal Velocity Obstacle J. Snape, J v.d. Berg, S. J. Guy, and D. Manocha IEEE Transactions on Robotics (T-RO), vol 27, 2011. [PDF] [MOV] [WMV]

PLEdestrians: A Least-Effort Approach to Crowd Simulation S. J. Guy, J. Chhugani, S. Curtis, P. Dubey, M. C. Lin and D. Manocha ACM SIGGRAPH/Eurographics Symposium on Computer Animation (SCA), July 2010. [PDF] [WMV]

Modeling Collision Avoidance Behavior for Virtual Humans S. J. Guy, M. C. Lin and D. Manocha Autonomous Agents and Multi-Agent Systems (AAMAS), May 2010. [PDF]

ClearPath: Highly Parallel Collision Avoidance for Multi-Agent Simulation S. J. Guy, J. Chhugani, C. Kim, N. Satish, M. C. Lin, D. Manocha, and P. Dubey ACM SIGGRAPH/Eurographics Symposium on Computer Animation (SCA), Aug. 2009. [PDF] [MOV]

Research Projects (View All)

Multi-Agent Navigation

• Distributed collision avoidance on modern CPUs & GPUs (ClearPath)
• Reciprocal collision avoidance (ORCA)
• Least-effort models of navigation (PLE)

Crowd Simulation

• Modeling personality and variation (Personality)
• Data driven crowd simulation (RCAP)
• Simulating very large crowds (Hajj at Mecca)

Robotics

• Navigation and collision avoidance under uncertainty (HRVO)
• Smooth navigation on mobile robots (S-ORCA, AVO)

Software

RVO Library — C++ — Multi-agent navigation library includes agent-agent collision avoidance, agent-obstacle avoidance, and global navigation. [download]

RVO2 Library — C++ — Multi-agent navigation library which improves computation performance of the RVO Library by over an order of magnitude. This library has been licensed by major game development studios for games scheduled for release in 2011 and 2012. [download]

Blob Detection — Matlab — My implementation of an affine-invariant, multi-scale "blob" detection. The method uses a Laplaceian of Gaussian filter to find approximately circle feature regions and orients these features using PCA techniques. [documentation]