Masaki Saito

Preferred Networks (Japan), Kyushu University

Papers

Total Citations

H-Index

About

Masaki Saito is a robotics researcher whose work bridges the gap between machine learning and real-world robotic manipulation. His most impactful contribution, "End-to-End Learning of Object Grasp Poses in the Amazon Robotics Challenge" (2020, 20 citations), demonstrates a pioneering approach to training robots to autonomously grasp diverse objects in unstructured environments—a critical challenge for warehouse automation and domestic robotics. This work showcases his expertise in deep learning for robotic perception and control, directly addressing the demands of human-robot coexistence. Earlier, Saito laid foundational groundwork with his research on "Model-based motion tracking system using distributed network cameras" (2010, 4 citations), which tackled the essential problem of detecting and tracking human motion to enable safe, collaborative human-robot interaction. By integrating distributed camera networks with model-based tracking, he contributed to the vision of a society where robots can seamlessly and securely work alongside people. His research trajectory—from human motion tracking to end-to-end grasp learning—reflects a sustained commitment to making robots more perceptive, autonomous, and collaborative partners in everyday environments.