K. Hoshino
Papers
2
Total Citations
19
H-Index
2
About
K. Hoshino is a rising researcher at the forefront of safe and reliable autonomous systems, with a primary focus on reinforcement learning (RL) and control theory. Their work directly addresses a critical gap in robotics: ensuring that learning algorithms produce not just optimal, but also safe and physically feasible actions. Hoshino’s most cited paper, “Benchmarking Actor-Critic Deep Reinforcement Learning Algorithms for Robotics Control With Action Constraints” (2023, 17 citations), provides a vital standardized framework for evaluating action-constrained RL. This benchmark is essential for real-world deployment, where a robot’s actions must respect physical limits and safety protocols. Building on this foundation, Hoshino’s latest work, “Physics-Informed Representation and Learning: Control and Risk Quantification” (2024), tackles the even more complex challenge of optimal and safety-critical control for high-dimensional stochastic systems. By integrating physics-informed models, this research promises to advance risk quantification in critical applications like robotic manipulation and autonomous driving. Though early in their career, Hoshino is already making a tangible impact by developing the principled methods needed to bridge the gap between powerful learning algorithms and the stringent safety requirements of the physical world.
Research Focus
Key Achievements
Top Papers
- 1
- 2