Kazuya Yoshida

Extraterrestrial construction moves from concept to hardware reality: reconfigurable modular robots can independently inspect, mechanically lock, and 3D-map inflatable lunar habitats in analogue terrain.

Autonomous robots can now traverse zero-g space station cabins without dedicated locomotion tracks by framing discrete handrail grappling and continuous path planning as a unified, manipulability-aware graph search.

Reliable cooperative cargo transport in lunar missions is now achievable through a novel phase-decomposed reinforcement learning framework that ensures safety and efficiency.