Masahiko Hoshi

Papers

Total Citations

H-Index

About

Masahiko Hoshi is a robotics researcher whose work advances the reliability and accuracy of autonomous mobile robot navigation. His primary research areas include simultaneous localization and mapping (SLAM), graph-based optimization, and localizability estimation. Hoshi’s major contributions address a critical challenge in SLAM: the accumulation of pose estimation errors that distort maps. In his highly cited 2022 paper, “Graph-based SLAM using architectural floor plans without loop closure” (8 citations), he pioneered a method that leverages pre-existing floor plans as constraints, eliminating the need for traditional loop closure—a technique that often leaves residual errors. He further refined this approach in 2024, integrating wall detection to enhance map consistency without loop closure. Additionally, his 2022 work on “Localizability Estimation based on Occupancy Grid Maps” (5 citations) introduced a framework to predict localization reliability across environments, enabling robots to assess where they can trust their sensors. By addressing both mapping accuracy and localization confidence, Hoshi’s research has practical implications for robots operating in structured indoor spaces, such as warehouses or hospitals. His innovative use of architectural constraints marks a notable shift toward more robust, real-world SLAM solutions.