RTS engine

Repository: https://codeberg.org/langurmonkey/rts-engine

This is an RTS engine project that I used to enjoy working on. It is a testing sandbox and does not have much focus.

I started playing around with the idea as a time killer some years ago, kicking off the development with a fast version of the A* path-finding algorithm backed not by a grid (as usual) but by a quad-tree. Quad-trees make path-finding superfast because of their hierarchical division of space and their adaptive partition sizes. Even though I used visibility graphs to store the navigable nodes from one given point, quad-trees are also fast for checking the properties/elements of a position’s surroundings, for child nodes are always spatially contained in parent nodes. However, the tree rebalancing operation is too costly, so I ended up implementing a regular grid based approach, which is the one in use today. A* is also quite fast in a regular grid.

Once I got this path-finding on a quad-tree thing up and running, It was time to implement the movement of my entities. I dove a bit into the topic and stumbled upon Craig Reynolds steering behaviors. They turned out to be an excellent method of implementing movement, but they come with caveats. I found these steering behaviors very powerful at producing organic-like movements that do not look forced at all. However, they are usually hard to implement and need A LOT of tweaking to really get them rolling. If you are interested in the topic you can check out Reynolds’ original paper or take a look at the book “Programming game AI by example” by Matt Buckland.

It looks like LibGDX itself provides nowadays an implementation of steering behaviors, which is probably nicer than mine. I implemented my version before, so that’s what is used in this project.

I wrote a blog post about this project.

Current features

• Shiny, not so good looking 2D graphics.
• Backed by a fully 3D model.
• Real time selection and movement of units and groups.
• Unit life bars.
• Fog of war comes in two flavors: tile-based (bad) and mesh-based (awesome).
• Own implementation of steering behaviors that work (with a lot of tweaking).
• Zoom and pan freely.
• Tiled tile map integration.
• Quad-tree and regular grid implementations for spatial awareness.
• Some basic graphical and lighting effects.
• Ashley ECS as entity component system.
• Pretty fast, above 500 FPS on decent hardware. Around 130 FPS on 7th gen Intel laptop graphics.
• Theoretically, Android support.

Running

In order to run the RTS engine on the desktop, first clone the repository:

git clone https://codeberg.org/langurmonkey/rts-engine.git
cd rts-engine

Then, just run the following command:

$ gradlew desktop:run

Video

Here is a video demonstrating a few of the features available as of now.

Licensing

This software is distributed under the MIT license.