Search Unity

  1. We are migrating the Unity Forums to Unity Discussions. On July 12, the Unity Forums will become read-only. On July 15, Unity Discussions will become read-only until July 18, when the new design and the migrated forum contents will go live. Read our full announcement for more information and let us know if you have any questions.

MA-POCA and Collaborative

Discussion in 'ML-Agents' started by HadisDev, May 3, 2022.

  1. HadisDev


    Mar 2, 2022

    I am trying to understand how MA-POCA is collaborative. During the training, we have the centralized baseline and value function which do contain information on other agents.

    But in the evaluation / Testing, we only have the policy network which takes in the local observation. If no agent has any information on the whereabouts of the other agents, how can they collaborate?

    For example, my experiment has a list of objects which the agents need to collect. I use a Buffersensor to pass the relative positions to the agents as input.

    I want them to collaborate in a way that minimizes the total distance traveled. Would it be a good idea to also supply another BufferSensor with the relative positions of each agent?
  2. iffalseelsetrue


    May 3, 2018
    I thought the MA-POCA takes care of the inter-agent communications for you already? sorry if I'm wrong, still learning...

    I know they are writing the documentation of MA-POCA atm.