So, where is the A.I.?

Anyone who has read the how-to-play notes, and played GRITS Racing for more than a few minutes, hopefully understands that this game is not just another top-down, shared-screen racing game with canned mechanics.

It's not easy finding depth in shared-screen multiplayer games.  There are easily more than 100 top-down racing games that came before GRITS Racing.  Many of them also shared-screen.  Thus, the basic design for GRITS Racing started with an intent to search for depth in places not yet seen in shared-screen racers.  As a result, we don't use any canned mechanics found in many shovelware racing games (other than a lap counter and clock to trigger the win condition).

Yes, we use a game engine to handle the graphics and platform porting.  We also use a physics engine, audio engine, and gamepad package, to handle other high- and low-level things we don't need to reinvent.  This gives us the maximum amount of time to focus on building original mechanics for things like how the car moves and handles, and how to interact with the environment.  I chose 2D instead of 3D because in 2D I can put about 4 times the number of physics objects on screen at the same processing impact.  This added depth of physics interactions outweighed anything 3D might have given us -- particularly at such small scale.  I basically traded vehicle rollovers for more track objects and debris.  Furthermore, if I really don't need that many more physics objects, I can divert some of that power to other things such as a wider range of game modifiers, special effects, and more interesting AI.

So, that is where the AI is at.  The original plans for the AI calls for AI cars that do more than just chase waypoints around the track like most racing games do.  And I'm not talking about building more strategic AIs either.  The more advanced racing games have that already... and, well, advanced racing strategies is not what GRITS Racing is about.  No, the original game design called for AIs with 4 to 6 different personalities -- and not necessarily racing personalities either.  That will take some significant time to build.

Thus, we made the decision to release GRITS Racing before the AI was built.  That has proven to be a mistake because the lack of AI is not making a good first impression when players test the game alone.  Thus, this "racing" game is now in a Catch-22 position.  Do I release an update sooner with simple AI?  Or, do I wait until the AI itself can make a good first impression when the AI features are finally announced and released?  It seems I should maybe go the simple AI route but the problem there is that it will create false expectations and maybe even upset a few when the AI behavior changes completely.  I will likely look for some middle ground as I dig into the AI capabilities more, and start with semi-interesting AIs that I then improve as we push towards the full release.  We'll see where it goes.

No promises on specific AI features yet.  The interesting AI hopes may prove to be a dud in testing.

Leave a comment

Log in with to leave a comment.