Max Robotics — Global Service Robot Directory + US Certification

About

Yuki Kadokawa is a robotics researcher whose work bridges simulation, real-world deployment, and energy-efficient hardware for autonomous systems. His primary research areas include sim-to-real reinforcement learning, laboratory automation, and edge-computing robot control. Kadokawa’s most impactful contribution is “Cyclic policy distillation,” a sample-efficient method for sim-to-real transfer that leverages domain randomization, enabling robots to adapt policies from simulation to physical environments with minimal real-world data (16 citations). He also advanced laboratory automation with “Learning Robotic Powder Weighing from Simulation,” where his team developed a robotic system capable of milligram-level precision in powder dispensing—a notoriously complex task due to material variability and dynamic powder behavior (15 citations). On the hardware frontier, his work on “Binarized P-Network” demonstrated deep reinforcement learning for image-based robot control on FPGAs, achieving power-efficient inference suitable for edge devices (6 citations). More recently, his “Robust iterative value conversion” method further optimized neurochip-driven edge robots for real-time decision-making. Kadokawa’s research consistently tackles the practical challenges of deploying learning-based robotics in real-world settings, from labs to resource-constrained platforms.