Max Robotics — Global Service Robot Directory + US Certification

About

Yu-Shen Liu is a leading researcher at the intersection of computer vision, robotics, and 3D perception. His work focuses on enabling intelligent agents to understand and navigate complex environments through cross-modality learning and vision-language reasoning. Liu’s major contributions include pioneering differentiable registration techniques for aligning 2D images with 3D LiDAR point clouds, a critical capability for autonomous driving and robotic mapping. His 2023 paper on VoxelPoint-to-Pixel Matching (9 citations) introduced a novel end-to-end framework that replaces traditional PnP estimation with learned correspondences, significantly improving registration accuracy. In 3D place recognition, Liu developed Pyramid Learnable Tokens (5 citations), a method that outperforms prior work by leveraging hierarchical features from real-world LiDAR scans rather than accumulated 2D data. His most recent work, Zero-Shot Object Navigation with Vision-Language Models (9 citations), demonstrates how large language models can guide robots to find objects in unseen environments without task-specific training. Liu’s research has direct applications in autonomous navigation, SLAM, and service robotics, with his papers collectively cited over 24 times in just two years. His goal-driven transformer for robot behavior learning further showcases his commitment to scalable, data-efficient robot learning from human play data.