Kazuki Mizuta

The University of Tokyo, University of Washington

Papers

Total Citations

H-Index

About

Kazuki Mizuta is a rising researcher at the forefront of safe and intelligent autonomous robotics, specializing in the intersection of control theory, machine learning, and motion planning. His work is defined by a central challenge: how to enable robots to navigate dynamic, human-populated environments with guaranteed safety. Mizuta’s major contributions lie in integrating **Control Barrier Functions (CBFs)** and **Control Lyapunov Functions (CLFs)** with modern learning-based approaches. In his highly cited 2022 paper (9 citations), he pioneered a method for safe persistent coverage control, using **Sparse Bayesian Learning** to infer safety constraints directly from sensor data, allowing robots to explore unknown spaces without pre-programmed obstacle maps. Building on this, his 2024 work, **CoBL-Diffusion** (8 citations), represents a significant leap forward. By fusing diffusion-based generative planning with CBFs and CLFs, he created a framework that generates robot trajectories that are not only efficient but provably safe in crowded, multi-agent settings. This work directly addresses a critical bottleneck in deploying robots alongside humans. Mizuta’s research is rapidly gaining recognition for its elegant synthesis of rigorous safety guarantees with the flexibility of data-driven learning, marking him as a key innovator shaping the future of trustworthy robot autonomy.