Papers
10
Total Citations
379
H-Index
5
About
Jingnan Shi is a robotics researcher whose work spans simultaneous localization and mapping (SLAM), robust estimation, and the integration of deep learning with physics-based optimization for autonomous systems. Affiliated with MIT, Shi has made significant contributions to some of the field's most pressing challenges, particularly in enabling robots to navigate and perceive complex, real-world environments reliably. Among Shi's most influential contributions is the LAMP 2.0 multi-robot SLAM system (161 citations), which enables heterogeneous robot teams to map and localize in large-scale, perceptually degraded underground environments—a critical capability for search-and-rescue operations. Complementing this, Shi contributed to ROBIN (70 citations), a graph-theoretic framework for outlier rejection in robust estimation, addressing scenarios where up to 90% of measurements may be corrupted. The Loc-NeRF system (85 citations) further demonstrates Shi's versatility, combining Neural Radiance Fields with Monte Carlo localization for real-time, vision-based robot positioning. Shi also co-developed PyPose (35 citations), a library bridging deep learning and physics-based optimization to improve generalization in robotic applications. Beyond research, Shi has contributed to robotics education through MIT's open-source Visual Navigation for Autonomous Vehicles course, reflecting a commitment to both advancing and democratizing the field.
Research Focus
Key Achievements
Top Papers
- 1
- 2Loc-NeRF: Monte Carlo Localization using Neural Radiance Fields85 citations · 2023
- 3
- 4PyPose: A Library for Robot Learning with Physics-based Optimization35 citations · 2023
- 5Kimera: From SLAM to spatial perception with 3D dynamic scene graphs12 citations · 2021
- 6
- 7
- 8PyPose: A Library for Robot Learning with Physics-based Optimization3 citations · 2022
- 9
- 10