Deep reinforcement learning for modeling human locomotion control in neuromechanical simulation.
Seungmoon Song, Łukasz Kidziński, Xue Bin Peng, Carmichael Ong, Jennifer Hicks, Sergey Levine, Christopher G Atkeson, Scott L Delp
PMID 34399772WHAT IT FOUND
This paper reviews computer algorithms that taught a digital skeleton to walk and turn.
It contains no patient data, no clinical trials, and no findings that apply to human rehabilitation.
What this paper is
This is a review of computer science methods for simulating human movement. It describes a machine learning competition where algorithms learned to control a digital skeleton, but it reports no data from human patients or clinical outcomes. There is nothing here to change your practice.