Overgrowth takes place in the savage world of Lugaru where rabbits, wolves and other animals are forced to use paws, claws and medieval weaponry to engage each other in battle. Combining 3rd person adventure platforming with intricate melee combat, Overgrowth achieves a unique feel. Overgrowth also benefits from Wolfire's brand new Phoenix Engine which has been built from the ground up to allow the use of cutting edge graphics, animation, and physics. Add to these exciting features Overgrowth’s realistic artificial intelligence and streamlined control system and the result is an astoundingly immersive experience.

Report article RSS Feed Pathfinding with Detour

A few days ago I posted that we've added automatic navigation meshes using Recast, and now I'd like to show some navigation paths using its companion library, Detour.

Posted by jeffr on Jun 3rd, 2010

A few days ago I posted that we've added automatic navigation meshes using Recast, and now I'd like to show some navigation paths using its companion library, Detour. These libraries come with some really nice-looking debug visualization utilities, so I thought I would show them off here. Here are a few quick pathfinding tests to see how Detour handles some common trouble spots:


Those look pretty accurate to me! It may look like the corners are cut so closely that the character would bump into a wall, but this is not the case; the navigation meshes are already taking the character's collision radius into account. Now let's see how it handles longer paths:


It seems to work well on paths of any length, but sometimes shows a slight preference for going through tile corners. The navigation mesh is divided into tiles so that Recast only has to voxellize a small part of the scene at a time, and the tile connections may be biasing the navigation heuristics a little -- this should be pretty quick to sort out.

The next step is to get some AI characters to run along these paths, and see how it goes. There is a lot of interesting work to do to combine pathfinding with actual combat strategy and teamwork, and these libraries will help us get to that a lot sooner.



Track us
on ModDB (visit our page)

Please join us here too:
Facebook icon ModDB icon Steam icon Twitter icon YouTube icon

Post comment Comments
Isundir
Isundir Jun 3 2010, 4:09am says:

lol this reminds me I really like the animal (rabbit?) in armor artwork that you can see if you look up in the right lower corner of the article :-P

+3 votes     reply to comment
moci
moci Jun 3 2010, 4:17am says:

Hey guys! I love the fact that you're getting more and more into game play instead of the graphics side of things now! Good job.

I do have a question, if you have 2 enemies. Will they use the same path? For instance in the top image, on the right side. A second blue cylinder could try to flank the red cylinder.

Or would that be part of your AI logic, making them think 'my buddy is going that way, let's find a new path for me that goes a bit different'?

Also, on the bottom image. I would maybe try to go around the obstacles instead of making all those turns. Will the path finder always try to make a straight line?

-Or will you add 'weights' to areas? (Maybe a weight tool in the editor where level designers can add/subtract weight, or automated as well). Where normal ground is 'blue' and more dense areas get more 'red', then the pathfinder would try to stay in the blues wherever possible?

+8 votes     reply to comment
Awesome_ninja
Awesome_ninja Jun 3 2010, 1:21pm replied:

you made my head spin and overheat with all these questions xD

+1 vote     reply to comment
formerlyknownasMrCP
formerlyknownasMrCP Jun 3 2010, 7:19pm replied:

I would think they'd use the same path and try to avoid running into each other- this is why apparently navmesh is good for Crowd dynamics. To make them use separate paths I think you'd have to code the AI to do that, not so much the Pathfinder.

+1 vote     reply to comment
SliderFF
SliderFF Jun 3 2010, 5:02am says:

mmm, i have a sense of some video, of characters with AI racing, comes))

+2 votes     reply to comment
Dragonlord
Dragonlord Jun 3 2010, 1:33pm says:

Just out of curiosity, what kind of path finding they use? Some A* type algorithm?

+1 vote     reply to comment
Arxae
Arxae Jun 3 2010, 3:40pm replied:

Blog.wolfire.com
check that post, they go into more detail there :)

+1 vote     reply to comment
formerlyknownasMrCP
formerlyknownasMrCP Jun 3 2010, 7:14pm replied:

judging by the demo I'd say yes.

+1 vote     reply to comment
hogan_skoll
hogan_skoll Jun 4 2010, 10:14am says:

Ty overgrowth interesting shizzle

+1 vote     reply to comment
Post a Comment
click to sign in

You are not logged in, your comment will be anonymous unless you join the community today (totally free - or sign in with your social account on the right) which we encourage all contributors to do.

2000 characters limit; HTML formatting and smileys are not supported - text only

Icon
Overgrowth
Platforms
Windows, Mac, Linux
Developer & Publisher
Wolfire Games
Contact
Send Message
Official Page
Wolfire.com
Release Date
TBD
Game Watch
Track this game
News
Browse
News
Report Abuse
Report article
Related Games
Overgrowth
Overgrowth Indie Single & Multiplayer Adventure
Related Engines
Phoenix Engine
Phoenix Engine Commercial TBD
Related Groups
Wolfire Games
Wolfire Games Developer & Publisher with 3 members