question

Mnik avatar image
0 Likes"
Mnik asked Joerg Vogel commented

Looking for Assistance RL model for Private Use Willing to Pay

I am currently working on a small project focused on task assignment and I aim to implement multi-agent reinforcement learning (MARL) using FlexSim. I am looking for experienced freelancers who specialize in using reinforcement learning tools in FlexSim to assist me in developing this model. If you have the relevant expertise and are interested in a paid collaboration, please send me a private message or email me at [email protected]. Thank you

FlexSim 24.1.0
reinforcement learning
· 3
5 |100000
Toggle Comment visibility. Current Visibility: Viewable by all users

Up to 12 attachments (including images) can be used with a maximum of 23.8 MiB each and 47.7 MiB total.

Joerg Vogel avatar image Joerg Vogel commented ·

@Mnik,

Please begin with:

https://docs.flexsim.com/en/24.1/ModelLogic/ReinforcementLearning/KeyConcepts/KeyConcepts.html

https://www.flexsim.com/videos/ai-ml-reinforcement-learning-in-flexsim/

and search for given answers here for about two years. And then request answers for questions of your own problems again. Thank you for your patience and acceptance of academic integrity standards.

0 Likes 0 ·
Mnik avatar image Mnik Joerg Vogel commented ·

@Joerg Vogel Thank you for your prompt reply and the helpful links.

I am working on a project that involves task assignment in assembly line workstations, which need to dynamically adapt to different situations. For this purpose, I am defining global tables for task precedence rules and the current task assignments to workstations. I have used the tutorials provided in the links you shared.

My question is: Is it possible to use multi-agent reinforcement learning (MARL) with the platform provided by FlexSim and its communication with Python? If so, should I define two separate reinforcement learning tools?

Thank you for your help!

0 Likes 0 ·
Joerg Vogel avatar image Joerg Vogel Mnik commented ·
When Reinforced Learning was introduced, there had been a limitation on instanced Python code execution to one single main thread on a windows system. If you want to take actions of several simultaneously learning instances you need to find an answer how they get right parameter allocation to those instances in one thread from FlexSim for multiple agents, because there is a chance that they may influence mutually each other. There has been a paradigm change towards instanced code in Python. This depends on version. And you need to find out when this has happened and if it is supported for libraries of FlexSim interfaces.
0 Likes 0 ·

0 Answers