• Advertisement


  • Content count

  • Joined

  • Last visited

Community Reputation

3756 Excellent

1 Follower

About IADaveMark

  • Rank
    Moderator - Artificial Intelligence

Personal Information


  • Twitter
  • Twitch
  • Steam
  1. OpenCV, AI

    Really more as a curiosity. For the most part, they aren't used in game AI... and for good reason.
  2. OpenCV, AI

    Not really a game AI topic, is it?
  3. Boids, a way not to check flock every update?

    Also, just because the tick time is 250-500ms, doesn't mean you should necessarily change your vector that often. There's all sorts of things you can do to make them seem... er... less than capable. Perhaps some tips on randomness in here: http://www.gdcvault.com/play/1014496/Using-Randomness-in-AI-Both
  4. Boids, a way not to check flock every update?

    Actually, I usually update my behaviors (of all types) about every 250-400ms. If you set each zombie's next check at a range about that big, the randomness will spread their updates out automatically so they aren't all on the same frame. Also, the variability leads to a more organic feel on a per zombie basis.
  5. Boids, a way not to check flock every update?

    The most famous of flocking birds, starlings, only pay attention to their 7 nearest neighbors and yet the group, as a whole, makes for a spectacular display.
  6. R&D Techniques of selling AI game

    Ok... you're not being terribly clear here. Do you have a prototype yet or are you trying to sell an idea so that you can make a prototype?
  7. Controlling a Very Intelligent AI

    Ummm... you realize that this forum is for game AI, right?
  8. How to create a dwarf fortress like game

    Uh oh... you've got your work cut out for you.
  9. AI Topic for External Blog

    Wouldn't it make more sense for you to decide on a topic that you are interested in and/or knowledgeable about?
  10. Choosing Flowfield Methods

    Not necessarily a direct answer, but certainly worth your while. https://www.gdcvault.com/play/1018262/The-Next-Vector-Improvements-in http://www.gameaipro.com/GameAIPro/GameAIPro_Chapter23_Crowd_Pathfinding_and_Steering_Using_Flow_Field_Tiles.pdf
  11. Yeah, I was being overly simplistic before moving on to the fact that A* is just "better Dijkstra".
  12. Wow... a lot of facepalm here. I can't see how you would need to do "research" on a topic that 30 minutes worth of Googling would solve for you. Dijkstra is uninformed and therefore kind of wandery. A* is informed with a heuristic and therefore more directed. It is also guaranteed to find you the shortest path if a path exists provided an admissible heuristic. It also supports arbitrary geometry shapes such as navmesh polys so you don't have to be entirely grid based. You can also arbitrarily change the edge costs between polys to represent something other than distance -- for example terrain type -- so you can bias the results easily to match the world. There are plenty of optimizations as long as the environment fits. e.g. if you are grid based, you can use things like JPS (Jump Point Search) and JPS+. Trying to do a survey and "research" on this is like polling the Math and Physics forum to see what the best way to find the area of a circle is.
  13. Tactical Pathfinding - Unity3D

    One method is propagating information into an influence map and then checking the map underlying individual navmesh polys as you are running the pathfind. In this case, if you register enemy visibility of squares into the map, then you can look up those values and increase the cost of those polys as you run your find. The agent would then avoid those and, in this case, move behind the obstacles instead.
  14. Yeah, not sure why you would overlay a grid onto something that is inherently edge-based like a road network. In fact, road networks are a dream for pathfinding. It's open areas with arbitrary geometry that are a PITA. That said, there is something to the notion of putting it into a hierarchy. Cluster graphs are the best bet. That's where you find groups that are close together and largely interconnected and treat them as a single "meta node". You then find which meta nodes are connected to each other and store the aggregate distance between them. You can then do a high level pathfind knowing full well that lower level ones will work. I'm oversimplifying, of course, but I'm wondering if you actually need that. Depends on how many intersections (nodes) you have on your map.
  15. Game AI architecture for a RTS game

    That said, there are all sorts of benefits to using a decision tree (or more specifically, a behavior tree) than using a "rule based system". Since you seem very new to game AI (another reason this is a big challenge for you), read the article that Alvaro linked to (it's mine). That will give you an overview of the different types of architectures.
  • Advertisement