Yoshikatsu Nakajima

Keio University, Carnegie Mellon University

Papers

Total Citations

H-Index

About

Yoshikatsu Nakajima is a leading researcher in autonomous robotics, with key contributions spanning semantic mapping, terrain perception, and self-supervised learning. His most influential work, "Efficient Object-Oriented Semantic Mapping With Object Detector" (2018, 38 citations), pioneered a system for incrementally building 3D maps annotated with object instances—a critical capability for scene understanding, human-robot interaction, and SLAM extensions. This work has become a foundational reference for researchers seeking to bridge the gap between raw sensor data and semantically rich environmental models. Nakajima has also made significant strides in terrain recognition for mobile platforms. His 2021 paper "Audio-Visual Self-Supervised Terrain Type Recognition" (15 citations) introduced a novel approach that leverages both auditory and visual cues, enabling robots to autonomously classify surfaces without manual labeling. This work addresses a fundamental challenge for social robots, assistive robots, and autonomous vehicles operating in unstructured environments. His follow-up 2020 study on terrain type discovery (4 citations) further advanced self-supervised learning methods, allowing robots to not only recognize but also discover new terrain categories during operation. Through these contributions, Nakajima has demonstrated how combining multimodal sensing with self-supervised techniques can create more adaptive and intelligent autonomous systems. His research continues to shape how robots perceive and interact with complex, real-world environments.