Search Unity

Is it common to use Q learning and reinforcment learning for navigation, obstacle avoidance,etc?

Discussion in 'Navigation' started by mahdiii, Feb 16, 2018.

  1. mahdiii

    mahdiii

    Joined:
    Oct 30, 2014
    Posts:
    856
    Hi all
    I would like to know whether it is common to utilize Q learning and reinforcement learning for navigation and obstacle avoidance in dynamic environment, path finding, car movement and selecting actions(shoot,chase,retreat,...) by AI enemies in TPS, FPS shooter game or RTS?
    Please cite papers and introduce games which used it.
    Thank you
     
    Last edited: Feb 16, 2018
    mahdiii, Feb 16, 2018
    #1

  2. MadeFromPolygons

    MadeFromPolygons

    Joined:
    Oct 5, 2013
    Posts:
    3,982
    No it is not common, and likely a waste of time unless you have some serious graphics-server level hardware to train it long enough to do anything useful in "big" and "modern" scale + detail games. If you have the time and hardware and knowledge to create and then train agents to do this, then sure go for it.
     
    MadeFromPolygons, Mar 6, 2018
    #2
    Harinezumi and mahdiii like this.

  3. neoshaman

    neoshaman

    Joined:
    Feb 11, 2011
    Posts:
    6,493
    No it's overkill
     
    neoshaman, Mar 6, 2018
    #3
    mahdiii likes this.