Sicong Pan

MS-MEM achieves higher mapping accuracy while minimizing scene disturbances, showcasing the power of integrating multiple manipulation skills in robotic perception.

SDPP achieves superior shelf organization and density while ensuring accessibility, outperforming traditional methods in both simulated and real-world environments.

Uncertainty can be harnessed to enhance the reliability of depth-only 3D semantic segmentation without compromising privacy.

People want blurry robot vision, but *how* blurry depends on how close the robot is and how nosy they think it's being.

3D diffusion models can be harnessed to generate object priors for one-shot view planning, enabling robots to efficiently reconstruct objects by focusing on complex regions.