Paul Vernaza

Me, standing in front of a large film print.

Welcome!

I am currently a postdoctoral fellow with the Robotics Institute at Carnegie Mellon University. I work primarily with Prof. J. Andrew Bagnell on problems lying at the intersection of machine learning and robotics. I recently completed my PhD at the GRASP Laboratory of the University of Pennsylvania, working under the supervision of Professor Daniel D. Lee.

I also enjoy photography.

Contact

pvernaza (at) cmu.edu

Research

My main research focus is reasoning about and synthesizing trajectories in high-dimensional spaces by exploiting the compressible structure of these problems. In my PhD thesis, I developed a set of techniques for this purpose that can be applied both to sequence analysis problems in machine learning and complex motion planning problems in the robotics domain. I am currently working on stochastic extensions of these tools that will (on the learning side) better model the stochastic nature of trajectories in the real world, and (on the planning side) exploit the power of random sampling to solve very difficult motion planning problems.

Curriculum Vitae

Download (PDF format)

Publications

PhD Thesis

[1] Paul Vernaza. Efficient learning and inference for high-dimensional Lagrangian systems. PhD thesis, University of Pennsylvania, May 2011. [ bib | slides | .pdf ]

Journal Articles

[1] Paul Vernaza and Daniel D. Lee. Learning and exploiting low-dimensional structure for efficient holonomic motion planning in high-dimensional spaces. International Journal of Robotics Research, 31, December 2012. [ bib | .pdf ]
[2] Jonathan Bohren, Tully Foote, Jim Keller, Alex Kushleyev, Daniel Lee, Alex Stewart, Paul Vernaza, Jason Derenick, John Spletzer, and Brian Satterfield. Little Ben: The Ben Franklin Racing Team's entry in the 2007 DARPA Urban Challenge. Journal of Field Robotics, 25(9), 2007. [ bib ]

Refereed Conference Papers

[1] Dmitry Yershov, Paul Vernaza, and Steven M. LaValle. Continuous planning with winding constraints using optimal heuristic-driven front propagation. In IEEE International Conference on Robotics and Automation, 2013. [ bib | .pdf ]
[2] Venkatraman Narayanan, Paul Vernaza, Maxim Likhachev, and Steven M. LaValle. Planning under topological constraints using beam graphs. In IEEE International Conference on Robotics and Automation, 2013. [ bib | .pdf ]
[3] Paul Vernaza and J. Andrew Bagnell. Efficient high dimensional maximum entropy modeling via symmetric partition functions. In Neural Information Processing Systems, December 2012. [ bib | .pdf ]
[4] Paul Vernaza, Venkatraman Narayanan, and Maxim Likhachev. Efficiently finding optimal winding-constrained loops in the plane. In Robotics: Science and Systems, July 2012. [ bib | .pdf ]
[5] Paul Vernaza and Daniel D. Lee. Learning dimensional descent planning for a highly-articulated robot arm. In Proceedings of the International Conference on Intelligent Robots and Systems, 2011. [ bib | slides | .pdf ]
[6] Paul Vernaza and Daniel D. Lee. Learning dimensional descent for optimal motion planning in high-dimensional spaces. In Proceedings of the 25th AAAI Conference on Artificial Intelligence, 2011. [ bib | slides | .pdf ]
[7] Paul Vernaza and Daniel D. Lee. Efficient dynamic programming for high-dimensional, optimal motion planning by spectral learning of approximate value function symmetries. In IEEE International Conference on Robotics and Automation, 2011. [ bib | slides | .pdf ]
[8] Paul Vernaza, Daniel D. Lee, and Seung-Joon Yi. Learning and planning high-dimensional physical trajectories via structured Lagrangians. In IEEE International Conference on Robotics and Automation, 2010. [ bib | slides | .pdf ]
[9] Paul Vernaza and Daniel D. Lee. Scalable real-time object recognition and segmentation via cascaded, discriminative Markov random fields. In IEEE International Conference on Robotics and Automation, 2010. [ bib | code | slides | .pdf ]
[10] Paul Vernaza, Maxim Likhachev, Subhrajit Bhattacharya, Aleksandr Kushleyev, and Daniel D. Lee. Search-based planning for a legged robot over rough terrain. In IEEE International Conference on Robotics and Automation, 2009. [ bib | .pdf ]
[11] Paul Vernaza, Ben Taskar, and Daniel D. Lee. Online, self-supervised terrain classification via discriminatively trained submodular Markov random fields. In IEEE International Conference on Robotics and Automation, 2008. [ bib | slides | .pdf ]
[12] Sachin Chitta, Paul Vernaza, Roman Geykhman, and Daniel D. Lee. Proprioceptive localization for a quadrupedal robot on known terrain. In IEEE International Conference on Robotics and Automation, 2007. [ bib | .pdf ]
[13] Paul Vernaza and Daniel D. Lee. Robust GPS/INS-aided localization and mapping via GPS bias estimation. In 10th International Symposium on Experimental Robotics, 2006. [ bib | .pdf ]
[14] Paul Vernaza and Daniel D. Lee. Rao-Blackwellized particle filtering for 6-DOF estimation of attitude and position via GPS and inertial sensors. In IEEE International Conference on Robotics and Automation (ICRA), 2006. [ bib | code | slides | .pdf ]
[15] Yuanqing Lin, Paul Vernaza, Jihun Ham, and Daniel D. Lee. Cooperative relative robot localization with audible acoustic sensing. In IEEE International Conference on Intelligent Robots and Systems (IROS), 2005. [ bib ]

Technical Reports

[1] Paul Vernaza, Daniel D. Lee, and Ben Taskar. Learning a kernel for discriminative, low-dimensional embedding of partially labeled data. Technical report, GRASP Laboratory, 2008. [ bib | .pdf ]