Ziqi Lu

Papers

Total Citations

H-Index

About

Ziqi Lu is a leading researcher at the intersection of robotics, computer vision, and 3D scene understanding. Their work centers on developing robust, object-level perception systems for autonomous agents, with key contributions in discrete-continuous optimization for simultaneous localization and mapping (SLAM) and novel applications of 3D Gaussian Splatting. Lu’s highly cited 2022 paper, “Discrete-Continuous Smoothing and Mapping” (16 citations), introduced a general framework for maximum a posteriori inference in hybrid state spaces, a foundational advance for handling ambiguity in real-world environments. This work directly informs their 2021 study, “Consensus-Informed Optimization Over Mixtures for Ambiguity-Aware Object SLAM” (7 citations), which tackles the challenge of multiple probable object poses due to symmetry or occlusion. Demonstrating a commitment to closing the simulation-to-reality gap, Lu’s 2022 paper “SLAM-Supported Self-Training for 6D Object Pose Estimation” (10 citations) leverages SLAM to generate self-supervised training data, enabling robots to adapt to novel environments. Most recently, their 2025 paper “3DGS-CD” (7 citations) pioneers the use of 3D Gaussian Splatting for detecting physical object rearrangements, opening new avenues for long-term scene monitoring and manipulation. With a growing citation impact and a clear trajectory from theoretical foundations to practical deployment, Ziqi Lu is shaping how robots build and maintain persistent, actionable 3D maps.