Michihisa Hiratsuka

About

Michihisa Hiratsuka is a leading researcher in robot learning from human demonstrations, with a focus on enabling non-expert users to intuitively teach complex skills to robots with unknown kinematics. His core contributions lie in developing nonlinear manifold alignment and mapping frameworks that allow arbitrary robots to imitate human motion trajectories in real time, bypassing the need for explicit kinematic models. In his highly cited 2016 work, “Trajectory learning from human demonstrations via manifold mapping,” Hiratsuka introduced a method that learns the underlying structure of human demonstrations and transfers that skill to robots with diverse morphologies. He extended this approach in his 2018 paper, “A Non-Linear Manifold Alignment Approach to Robot Learning from Demonstrations,” which further refined the alignment process to handle greater kinematic differences between teacher and robot. Each of these foundational papers has garnered 8 citations, reflecting their growing influence in the fields of imitation learning and human-robot interaction. Hiratsuka’s work is particularly notable for its practical impact: it lowers the barrier for deploying robots in non-laboratory environments, from homes to factories, where robots must adapt to new tasks without specialized programming. His research continues to shape how robots learn from human guidance, making skill acquisition more accessible and scalable.

Research Focus

Key Achievements

Top Papers

1. 1
   Trajectory learning from human demonstrations via manifold mapping

   8 citations · 2016
2. 2
   A Non-Linear Manifold Alignment Approach to Robot Learning from Demonstrations

   8 citations · 2018

Key Collaborators

Contact & Links