Research
Current research
I’ve been deeply influenced by my advisor Dave’s research style, and I’m trying to explore and address fundamental problems in robotic state estimation using tools from optimization, differential geometry, probability theory, and information theory. Coming from an engineering background, I also strive to translate these ideas into practical systems — such as high-performance C++ libraries and field robotics experiments.
The dream seems a little bit ambitious, and I’m still learning along the way:)
Certifiable Estimation
1, Trying to make certifiable estimation implementation easier and used in more real robotics application (ongoing…), and brings more developed feature (like incremental estimation and robust estimation) to certifiable algorithm. (ongoing…)
2, Robust certifiable estimation. (learning and exploring…)
Distributed Optimization
1, (More efficient and faster) Distributed bundle adjustment solver. (ongoing…)
2, Distributed certifiable estimation. (ongoing…)
Based on our certifiable factors and distributed second-order optimization algorithm, trying to achieve faster convergence than the currently leading RBCD-based methods.
Other research interests
Large-scale and long-term autonomous operation
Edges/nodes Sparsification.
Lifelong localization/mapping
New map/environment representation.
Information-based algorithm
information-based planning, mapping, calibration…
Active SLAM…