Kazuto Nakashima

About

Kazuto Nakashima is a robotics and computer vision researcher whose work spans 3D LiDAR perception, generative modeling, and intelligent sensing systems for autonomous robots. His research has made notable contributions to several interconnected domains, most prominently LiDAR-based scene understanding and biometric recognition. Nakashima's most impactful work focuses on generative modeling of LiDAR data, including his 2024 paper on denoising diffusion probabilistic models for LiDAR synthesis (22 citations) and a 2023 study on generative range imaging to bridge domain gaps in LiDAR perception. These contributions address critical challenges in scalable simulation and data augmentation for autonomous mobile systems. Complementing this, his research on LiDAR-based gait recognition — including robust 3D recognition across varying walking directions and viewpoint-invariant feature learning — has opened new directions in biometric identification using non-visual sensors. His earlier work on informationally structured environments, including the "Big Sensor Box" IoRT platform and the innovative "fourth-person sensing" concept, demonstrates a broader vision for intelligent spaces where robots, wearable sensors, and embedded infrastructure operate in concert. His Fukuoka datasets for place categorization have provided valuable benchmarks for the robotics community. Across his body of work, Nakashima has consistently advanced robust, real-world perception for autonomous systems.

Research Focus

Key Achievements

Top Papers

1. 1
   LiDAR Data Synthesis with Denoising Diffusion Probabilistic Models

   22 citations · 2024
2. 2
   2V-Gait: Gait Recognition using 3D LiDAR Robust to Changes in Walking Direction and Measurement Distance

   17 citations · 2022
3. 3
   Feasibility study of IoRT platform “Big Sensor Box”

   11 citations · 2017
4. 4
   Fukuoka datasets for place categorization

   11 citations · 2019
5. 5
   Generative Range Imaging for Learning Scene Priors of 3D LiDAR Data

   10 citations · 2023
6. 6
   Learning geometric and photometric features from panoramic LiDAR scans for outdoor place categorization

   4 citations · 2018
7. 7
   Fourth-person sensing for a service robot

   4 citations · 2015
8. 8
   Previewed reality: Near-future perception system

   3 citations · 2017
9. 9
   Learning Viewpoint-Invariant Features for LiDAR-Based Gait Recognition

   3 citations · 2023
10. 10
    Fourth-Person Captioning: Describing Daily Events by Uni-supervised and Tri-regularized Training

    2 citations · 2018

Key Collaborators

Contact & Links