Humans use a lot of sensory information (e.g. vision, haptic and equilibrium) not only for stabilizing the body, but also for walking on unstructured environment.
This capability has been very challenging to reproduce with conventional robotics, despite the application of powerful computation to solve state-estimation, full-body models etc.
However, are we solving these kinds of complex problems when we walk? I think I'm not. For us, bipedal walking is easier than multiplication or calculating the velocity of a falling apple. I believe this is because a large part of walking control is not done by the brain, but rather by simple, localized controllers, such as central pattern generators (CPGs) and reflexes.
We are studying the basic principles for achieving human-like bipedal locomotion with these simple, decentralized controllers by using a simplified model of the body and control system. In the future, we aim to develop a humanoid robot that can adapt it's locomotion to the circumstances, like us.
I graduated from the electronic control engineering department in Sendai national college of technology. After that, I completed a bachelor's degree in engineering at Tohoku university. I am currently a graduate student in department of electrical engineering.
I enjoy playing and developing computer games. I also like digital drawing.