|
Computer Science
Department Stanford University Gates 133, 353 Serra Mall Stanford CA 94305-9010 |
(650) 725-8814 (Work) (650) 567-9409 (Home) (650) 725-1449 (Fax) murali@cs.stanford.edu http://robotics.stanford.edu/~murali |
| Ph.D. | Computer Science, Brown University, Providence, Rhode Island, June 1998. |
| Sc.M. | Computer Science, Brown University, Providence, Rhode Island, May 1993. |
| B.Tech. | Computer Science and Engineering, Indian Institute of Technology, Madras, India, August 1991. |
| Model building | I have developed algorithms for incrementally constructing three-dimensional models of urban environments using sensors mounted on multiple, coordinated robots. These algorithms minimise the total time needed to complete the reconstruction. They do not need any a priori knowledge of the environment and take sensor limitations, sensor errors, and errors in estimating sensor positions into account. |
| Binary space partitions | The binary space partition is a geometric data structure that represents a hierarchical decomposition of a set of objects. It is used in various applications such as visualisation, solid modelling, and surface simplification. I have developed algorithms that construct binary space partitions of provably near-linear size and logarithmic depth for architectural models, urban landscapes, and terrain-like data sets. In practice, the algorithm for architectural models constructs a smaller binary space partition than most known algorithms. I also developed the first provably-efficient algorithm for maintaining a binary space partition for a set of moving segments in the plane. |
| Hidden-surface removal | I developed the object complexity model for hidden-surface removal. I am currently implementing a new algorithm for hidden-surface removal in walkthrough applications that renders complex environments efficiently by maintaining a small superset of the set of visible objects. |
| Geometric data repair | Many geometric data sets contain geometric and topological errors. I developed and implemented a novel approach based on spatial partitions for generating topologically-consistent solid models from arbitrary polygonal data. |
| Contour-line extraction | An important problem in geographic information systems is computing the contour of a terrain at a given height. Since terrain data sets often do not fit in main memory, algorithms operating on them should minimise the number of disk accesses they perform. I developed the first algorithm for contour extraction that uses an optimal number of disk accesses. |
| Algebraic decision trees | Constructing algebraic decision trees is a fundamental machine learning problem. I designed algorithms that construct trees of near-optimal size for points in two and three dimensions. |
| October 1998-present | Post-doctoral researcher with Profs. Leonidas J. Guibas and Jean-Claude Latombe in the Computer Science Department at Stanford University. |
| 1993-1998 | Research Assistant for Prof. Jeffrey S. Vitter at Department of Computer Science, Duke University. |
| Summer 1996 | Internship with Prof. Thomas A. Funkhouser in the Multimedia Lab, Bell Laboratories, Lucent Technologies, Murray Hill, New Jersey. |
| Summer 1995 | Internship in the Performer group, Advanced Systems Division, Silicon Graphics Inc., Mountain View, California. |
| Spring 1993 | Teaching Assistant for Discrete Mathematics in the Department of Computer Science at Brown University. |
| 1991-1992 | Research Assistant for Prof. Franco Preparata in the Department of Computer Science at Brown University. |
| 1986-1991 | Scholarship from the National Council for Education, Research, and Training, India. |
| 1989 | Rajalakshmi Krishnamurthy English Prize, Indian Institute of Technology, Madras, India. |
| 1987 | The first Indian participant in NASA's International Summer Students Program. |