define the reward

[yani-rl-ai]

Hi @ethanabrooks. currently i am working with HSR for reinforcement learning. i have got a lot of insight from your code. I have tried to run your code by using SAC for stable-baseline frameworks, it spent very long time but it was difficult to get the optimum behavior. However, i do not understand how do you calculate and define the reward for pick and place problem(hsr-env). What kind of reinforcement learning algorithm that you used for this environment?
how long did you spend to get the converge results for this environment?
i saw your another repository of HSR environment and i understand how did you define the reward and negative reward, but the repository is quite old. https://github.com/ethanabrooks/hsr
can you explain to me how do you define the reward in this "hsr-env"

[ethanabrooks]

Hi Mohamad. This repository is not maintained any more. Why don't you let me know what you are working on and I can suggest some better options.

[yani-rl-ai]

hi @ethanabrooks thank you for your answer, can you suggest me the better options that relate to HSR robot with gym environment?

[ethanabrooks]

Let me first ask: do you need to use the HSR robot? There are other robotics simulators that are better supported for reinforcement learning.

[yani-rl-ai]

yes, actually i need HSR for RL

[ethanabrooks]

I see. Do you have any restrictions in terms of the deep learning / autograd framework that you use?

[yani-rl-ai]

No, any approach is okay with me.

…

On Thu, 8 Apr 2021, 02:35 Ethan Brooks, ***@***.***> wrote: I see. Do you have any restrictions in terms of the deep learning / autograd framework that you use? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#3 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ALTIY2QTUWB7JFPBWS5TUS3THSJXTANCNFSM42GFHCUQ> .