Toshimitsu Kaneko

Papers

Total Citations

H-Index

About

Toshimitsu Kaneko is a robotics researcher whose work focuses on the intersection of deep reinforcement learning and vision-based manipulation for industrial automation. His primary research areas include robotic picking and placing, container loading optimization, and offline learning for robotic control. Kaneko’s most notable contribution is his pioneering work on simultaneous planning for item picking and placing using deep reinforcement learning, which addresses a critical challenge in logistics by integrating two traditionally separate planning processes. This approach, published in 2020, has accumulated 12 citations and demonstrates how the condition of picking an item directly influences placement possibilities, leading to more efficient robotic systems. More recently, Kaneko has advanced the field of offline learning for vision-based robotic manipulation, proposing a novel two-stage agent architecture that achieves effective learning with small datasets—a significant breakthrough given that offline learning typically requires massive data collection. His work on behavior correction policies, published in 2024, shows promise for making robotic systems more practical and data-efficient in real-world manufacturing and logistics environments.