search

amir-abdi / artisynth-rl

Reinforcement Learning plugin and models for ArtiSynth
MIT License
5 stars 2 forks source link

reducing action space! #3

Open hosseinahm1367 opened 4 years ago

hosseinahm1367 commented 4 years ago

Hi Amir,

Thanks for sharing the code. very interesting stuff. I was wondering if it is possible to modify the lumbar spine model. Specifically, I am looking to reducee the number of muscles in the action space. How can I go about doing that?

Thanks,

amir-abdi commented 4 years ago

Hi,

Yes that is possible. You can make any modifications in the Artisynth model by updating the source code in java.

On Nov 10, 2020, at 3:52 PM, hosseinahm1367 notifications@github.com wrote:

 Hi Amir,

Thanks for sharing the code. very interesting stuff. I was wondering if it is possible to modify the lumbar spine model. Specifically, I am looking to reducee the number of muscles in the action space. How can I go about doing that?

Thanks,

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or unsubscribe.

hosseinahm1367 commented 4 years ago

Thanks, I figured that out. I do have another question though. Once a policy is trained, can it be generalized to the motion of a scaled model (i.e. different mass or size)?

amir-abdi commented 4 years ago

Well, if it has never seen any variability in the mass and size factors of the state space, I won't expect it to adopt to such change. So: no.

On Nov 14, 2020, at 7:56 PM, hosseinahm1367 notifications@github.com wrote:

 Thanks, I figured that out. I do have another question though. Once a policy is trained, can it be generalized to the motion of a scaled model (i.e. different mass or size)?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub, or unsubscribe.