Fig. 2
From: Deep reinforcement learning for modeling human locomotion control in neuromechanical simulation
Musculoskeletal models for studying human movement. a Models implemented in OpenSim [1] for a range of studies: lower-limb muscle activity in gait [13], shoulder muscle activity in upper-limb movements [14], and knee contact loads for various motions [15]. b A Hill-type muscle model typically consists of a contractile element (CE), a parallel elastic element (PE), and a series elastic element (SE). The contractile element actively produces contractile forces that depend on its length and velocity and are proportional to the excitation signal. The passive elements act as non-linear springs where the force depends on their length