Xuanpeng Li

Papers

Total Citations

H-Index

About

Xuanpeng Li is a researcher working at the intersection of computer vision, robotics, and autonomous systems, with particular expertise in 3D scene understanding and trajectory forecasting. His most influential contribution, "Semi-Dense 3D Semantic Mapping from Monocular SLAM" (2016), addresses a fundamental challenge in robotic perception: achieving rich, semantically meaningful 3D reconstruction using only a single monocular camera, circumventing the cost and inflexibility of stereo or RGB-D sensor setups. This work, which has garnered 47 citations, demonstrates how geometric and appearance information can be jointly leveraged to enable practical autonomous navigation. Building on his interest in intelligent systems, Li later turned his attention to motion prediction, proposing a Hierarchical Motion Encoder-Decoder Network for trajectory forecasting (2021), which captures both fine-grained movement trends and high-level driving intentions — factors previously overlooked in the field. This work highlights his commitment to developing more human-aware and context-sensitive autonomous agents. Across his research portfolio, Li consistently tackles real-world constraints in robotics and autonomous vehicles, bridging the gap between theoretical computer vision and deployable intelligent systems.