Kensuke Kondo

Papers

Total Citations

H-Index

About

Kensuke Kondo's research lies at the intersection of robotic mapping, computer vision, and spatial data compression. His primary contributions focus on solving the critical problem of scalable map retrieval for mobile robot self-localization. Kondo pioneered the application of the multi-scale Bag-Of-Features (BOF) approach to large-scale environment maps, enabling robots to efficiently identify and retrieve previously built maps from vast collections. His most cited work, "Multi-Scale Bag-of-Features for Scalable Map Retrieval" (2012, 11 citations), established a novel framework that significantly improved retrieval accuracy by analyzing map features across multiple spatial scales. Building on this foundation, Kondo also explored innovative grammar-based compression techniques for robotic maps, leveraging Manhattan world priors to represent pointset data through context-free grammars. This work addressed the practical challenge of reducing map storage requirements while preserving essential structural information for localization tasks. Though his citation counts are modest, Kondo's research represents an important early exploration of how computer vision techniques like BOF could be adapted for the unique challenges of robotic mapping, contributing to the broader field of autonomous navigation and spatial understanding.