Sicong Pan
Papers
9
Total Citations
93
H-Index
6
About
Sicong Pan is a robotics researcher specializing in autonomous robotic perception, active vision, and 3D object reconstruction. His work sits at the intersection of computer vision, motion planning, and machine learning, with a particular focus on enabling robots to intelligently plan their viewpoints when encountering unknown objects and environments. Pan's most significant contributions center on the view planning problem — how robots can autonomously determine optimal camera positions to reconstruct unfamiliar objects with maximum efficiency. His SCVP framework (2022, 30 citations) introduced a landmark one-shot learning approach that reframes view planning as a set-covering problem, dramatically reducing the iterative overhead of traditional next-best-view methods. Complementing this, his max-flow-based multi-resolution approach (2021, 18 citations) offered a globally optimal solution for complete 3D reconstruction in unstructured settings. Beyond reconstruction, Pan has made meaningful contributions to safe multi-agent reinforcement learning for cooperative robot navigation, vision-guided grasping systems, and agricultural robotics — including innovative work on fruit monitoring through safe leaf manipulation. His 2024 work leveraging 3D diffusion model priors signals a growing interest in integrating generative AI into robotic perception pipelines. With over 90 cumulative citations across a concise body of work, Pan is an emerging voice shaping the future of intelligent, autonomous robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5
- 6Viewpoint Push Planning for Mapping of Unknown Confined Spaces8 citations · 2023
- 7
- 8
- 9Active Implicit Reconstruction Using One-Shot View Planning2 citations · 2024