Bingxin Xu

University of Michigan–Ann Arbor, Cornell University

Papers

Total Citations

184

H-Index

About

Bingxin Xu’s research lies at the intersection of computer vision and robotics, with a focus on enabling machines to perceive, reason, and act intelligently in complex environments. Her most impactful contribution is the seminal work “Depth-Encoded Hough Voting for Joint Object Detection and Shape Recovery” (2010, 150 citations), which introduced a novel method for simultaneously detecting objects and reconstructing their 3D shape from depth data—a key advance for autonomous systems operating in cluttered, real-world settings. In robotics, Xu has tackled the challenge of guaranteeing high-level behaviors during exploration. Her papers “Guaranteeing High-Level Behaviors while Exploring Partially Known Maps” (2012, 25 citations; 2013, 9 citations) present a formal approach to automatically synthesizing hybrid controllers that ensure a robot follows user-defined specifications—such as safety or sequencing tasks—even as it incrementally discovers its workspace. This work bridges control theory and temporal logic, offering provable correctness in uncertain environments. By combining robust perception with verifiable planning, Xu’s research has laid essential groundwork for trustworthy autonomous systems, making her a notable voice in the push toward reliable, task-aware robots.