I am particularly interested in developing SLAM-aware robots that can learn persistently in an environment from visual experience. My research attempts to understand the capabilities at the intersection of object and scene understanding and Simultaneous Localization and Mapping (SLAM).
- ICRA ‘17: Our paper on Centralized Graph Databases for Mobile Robotics got accepted to ICRA ‘17
- ICRA ‘16: Our paper on High-Performance and Tunable Stereo Reconstruction got accepted to ICRA ‘16
- Our research on Monocular SLAM Supported Object Recognition has been featured on MIT News.
- RSS ‘15: Our paper on Monocular SLAM Supported Object Recognition got accepted to RSS ‘15
- CVPR ‘15: Our paper on Line-Sweep: Cross-Ratio for Wide-Baseline Matching and 3D Reconstruction got accepted to CVPR ‘15
Self-Supervised Place Recognition in Mobile Robots
S. Pillai and J. Leonard
Learning for Localization and Mapping Workshop, IROS, 2017
[ pdf ]
SLAMinDB: Centralized graph databases for mobile robotics
D. Fourie, S. Claassens, S. Pillai, R. Mata, J. Leonard
International Conference on Robotics and Automation (ICRA), 2017
[ pdf ]
High-Performance and Tunable Stereo Reconstruction
S. Pillai, S. Ramalingam and J. Leonard
International Conference on Robotics and Automation (ICRA), 2016
[ pdf, video, bib ]
- Experimental Implementations of Stereo Matching Algorithms in Halide: I helped advice [Min Zhang] at CSAIL, MIT on stereo algorithms and their implementation using Halide, a Domain-Specific Language for Image Processing. See MEng. Thesis, Code for more implementation details.