Exclusive:HavokRelease newAIMiddleware
March 23rd, 2009 Alex J. champandard
2009-3-23, Lai Yonghao (Http://blog.csdn.net/lanphaday) Translation
Original article address:Http://aigamedev.com/middleware/havok-ai-announced
In20On the day (last Friday,HavokInvitationAigamedeve.comVisit their headquarters in San Francisco to visit their newAIMiddleware components. Chief EngineerDave garganAnd software engineersChris ELIONI demonstrated the technologies they spent more than a year and a half working on.Demos.
This article isHavok AIExclusive preview, officially held on Wednesday09Annual game developer conference (23To27(Held in San Francisco)ExpoThe secret. This isGDCFirst releaseAIProducts in the field, welcome to continue with this increasingly popular market.
HavokYesAigamedev.comAnd we chose to expose their latestAIProduct. This article is global exclusive,Havok AIIt will be released tomorrow.
dealing with destruction
David gargan game developers One of the challenges is the dynamic world's pathfinding . Countless people are looking for solutions. As a matter of fact, not only is the world growing (built in games), but it also has to face the destruction of the environment and numerous movable objects. Havok is taken into account when designing its AI middleware component. these requirements, fully integrated with dynamic pathfinding :
"Havok AINot a staticPathfindingAdd a dynamic solution to avoid technology, but build it directly from the dynamic environment. Its unique and innovative solution can process thousands of moving objects in real time and is quite realistic ."
Obviously, one of the key challenges supporting the dynamic world is to efficiently handle inevitable changes. These two methods can be used to achieve this:
- the algorithm that quickly constructs the core navigation grid can process the entire grid in a short time. For reference, Havok demonstrates a castle level during our visit, which takes only a few seconds to process.
- Based on Havok physical experience, they found that changes in the dynamic world are local in time and space. This algorithm can be used to handle dynamic changes to save time.
HavokTheAI demosIntegrated with a full set of middleware, includingHavokPhysical engine andHavokBehavior engine, but in factAIComponents can be connected to any other physical simulation engine.
1: When obstacles are dynamically added and removed in the level, the navigation network is updated. Local control can also help roles avoid collisions (Big chart).
Dynamic navigation Meshes
It is observed that many modern Games are more limited by memory rather than computing power,HavokIt is determined that each level only stores one navigation grid, and then finds the path of any behavior through some running time.HavokBy optimizing itsPathfindingTo run on the coprocessor, suchPS3OfSPUTo support this decision. In the release,HavokSaid:
"Completely scalable and customizablePathfindingSolution. This product provides a multi-threaded and platform-optimized hierarchy for all key game platformsPathfinder."
WhenPathfinderMulti-threaded optimization is often difficult to implement when navigation links and smart objects need to be supported.AI-- FortunatelyHavokAlso provided.
HavokDisplayedDemosThere is a building that contains a navigation link near the window, so soldiers can bow over the building and useHavok BehaviorEngine to process animations. Of course, when a building is damaged,AIThis link will no longer exist if you can handle this change.
In this case, if the object configuration is generally known during development (for example, whether some parts of a building are upright or collapsed), you can use a variety of presentation techniques to efficiently process the underlying system.
2: The windows of a building are connected by navigation to allowAIJump over, just likeHavok BehaviorProcess a smart object like that. After the wall collapsed, the connection in the navigation network will also be updated (Big chart).
Crowds and moving objects
Havok AIAnother characteristic is that it can process the dynamic moving objects in the world:
- active and autonomous roles, such as soldiers or pedestrians.
- passive or physical entities, such as grenades and bombs.
HavokI have introduced this situation to me:
"Achieve the dynamicPathfindingThe advanced local control module provides the ability to predict the movements of obstacles and make adjustments accordingly. When the environment becomes more dynamic, complicated and crowded situations make the movement look more real."
Finally,HavokThe complete official version will be published at tomorrow's press conference.
Figure3:Aigamedev.com HavokThe San Francisco Headquarters. From left Alex champandard ,Dave garganAndChris ELION.