Robot Learning

Improving policy adaptation for robot learning

The goal is to find a useful heuristic expression for the dual variable in the problem. This exploration will help find initial guesses to optimize other problems.

Prove mathematically that a printer with 2 non-ideal timing pulleys on the x-axis and y-axis can not achieve self-repair without feedback.

In our lab, we have a robot demonstration team who implements SOTA robot learning technologies on our robots. Here are some projects from the team. Learning Parkour with Simulators (2024) Demo Developers: Hikariu Wada, Momo Hanawa, Koki Fukuda, Yoshihiro Noumi

Development of algorithms for learning control from logged data

Quantifying and comparing RL policies and environments

In this project, research teams from Japan, France, and Germany will collaborate internationally to develop innovative robotic technologies for sustainable agriculture. Additionally, we aim to expand the applicability of robot learning and teleoperation technologies across various fields.

Learn how to apply recent deep learning technologies to robot system (project-based learning)

Learn how to apply recent deep learning technologies to robot system (project-based learning)

Learn how to apply recent deep learning technologies to robot system (project-based learning)