Ichiro Kobayashi

About

Ichiro Kobayashi is a leading researcher at the intersection of cognitive robotics, natural language processing, and unsupervised machine learning. His work is driven by a fundamental question: how can robots learn to perceive and segment continuous, high-dimensional information—like human motion or speech—into meaningful units, mirroring human cognition. Kobayashi’s major contributions center on developing novel statistical models for unsupervised time-series segmentation. His most cited work, "Segmenting Continuous Motions with Hidden Semi-Markov Models and Gaussian Processes" (56 citations), pioneers a method for robots to autonomously break down continuous movements into recognizable actions. He further advanced this field with "HVGH," a deep neural compression technique for high-dimensional data segmentation (20 citations), and his comprehensive "Survey on frontiers of language and robotics" (53 citations) has become a key reference for researchers. Kobayashi’s impact is evident through his consistent exploration of how robots can learn language and motion from raw sensory data, with recent work extending to spatio-temporal categorization from first-person robot videos. His research is foundational for creating robots that can intuitively understand and interact with the real world.

Research Focus

Key Achievements

Top Papers

1. 1
   Segmenting Continuous Motions with Hidden Semi-markov Models and Gaussian Processes

   56 citations · 2017
2. 2
   Survey on frontiers of language and robotics

   53 citations · 2019
3. 3
   HVGH: Unsupervised Segmentation for High-Dimensional Time Series Using Deep Neural Compression and Statistical Generative Model

   20 citations · 2019
4. 4
   Sequence Pattern Extraction by Segmenting Time Series Data Using GP-HSMM with Hierarchical Dirichlet Process

   15 citations · 2018
5. 5
   Eliciting User Food Preferences in terms of Taste and Texture in Spoken Dialogue Systems

   8 citations · 2018
6. 6
   High-dimensional Motion Segmentation by Variational Autoencoder and Gaussian Processes

   6 citations · 2019
7. 7
   Spatio-temporal categorization for first-person-view videos using a convolutional variational autoencoder and Gaussian processes

   4 citations · 2022
8. 8
   A study on the efficiency of learning a robot controller in various environments

   3 citations · 2013
9. 9
   Soft Computing in Advanced Robotics

   2 citations · 2014
10. 10
    A POMDP-based Multimodal Interaction System Using a Humanoid Robot

    2 citations · 2016

Key Collaborators

Contact & Links