Imagine being in your bedroom, full of furniture, blindfolded, and trying to get to the kitchen. You’d be running into everything, stepping on who knows what. Well, that’s exactly how your AI feels without a proper navigation system.
Now, think about the last time you used Google Maps. But instead of just showing roads, imagine turning on different layers that highlight construction, traffic, or speed traps 😏. Green zones are safe and fast, while red zones are slow or dangerous—all to help you make better decisions on your route.
That’s exactly what an influence map does for your AI—Google Maps for your bot, instead of streets layers, we place influence over the game world. One way we can map out the influence is with a grid.
Building a Grid
Think of a grid like a giant chessboard for AI. Each square has an assigned value or cost that informs the AI about the space. A basic grid might just tell the AI:
Can I walk here? (0 for yes, 1 for no)
But we can take it further—make walls a cost of 20, and assign dangers different values from 1-9, meaning “only go here if you have to or can win.”
Influence around these blasted enemy marines! BAM
You can create multiple influence map grids to inform your AI about different meanings to your AI:
🟢 Pathing Grid: Where your AI can move freely.
🟠 Terrain Grids: Choke points, ramps, climbable cliffs or wall offs to take advantage.
🔴 Combat Zones: Enemy armies, turrets, or contestable areas to avoid.
The magic happens when you layer these grids together, creating a dynamic influence map that helps your AI balance movement, safety, and tactical opportunities—just like a skilled player. Your bots use influence maps to make better, context-aware decisions,—playing in a way that feels far more human. When I spoke to the creators of Tyr, they explained how they plan to design their influence maps to help their army take smarter fights—avoiding bad engagements instead of blindly walking into danger. Just like a pro player… well, the winning part, not the ‘losing your entire army’ part 😅.
🛠️Making Your own Influence Map
Building influence maps from scratch is possible, but real talk—it’s a ton of work. Thankfully, you don’t have to. You can use the popular SC2MapAnalysis plugin to generate grids, or if you’re using Ares, you can enhance prebuilt grids for even more control.
12PoolBot may be the bot with the smallest amount of code in ProBots history with just over 800 lines of code but don’t snooze on its Zergling Micro.
Most Makers default to A since it pairs well with influence maps. But 12PoolBot’s creator took a different route, using a custom Dijkstra implementation to coordinate all units at once—leading to its fluid, swarm-like movement.
