xbpeng / DeepMimic

Motion imitation with deep reinforcement learning.
https://xbpeng.github.io/projects/DeepMimic/index.html
MIT License
2.27k stars 485 forks source link

when run a pre-trained policy , is it possible to set the input state/action change follow the input motion clips? #188

Open visonpon opened 1 year ago

visonpon commented 1 year ago

Hi @xbpeng ,after trained a policy with motion clips that contain walk and run use amp, it can walk or run normal, but I need to press r to reset the initial state in order to change the motion type(e.g, from walk to run). I wonder in test phase, if the initial state can follow the input motion sequence frame by frame rather than random select a pose from motion file as initial pose, so that the charactar can imitate the input motion clips in the test phase?