Giter VIP home page Giter VIP logo

anyanglepathfinding's Introduction

Any-Angle Pathfinding

This package is a lightweight any-angle pathfinding algoirthm that depends on SharpMath2. This implementation tends to have faster performance and a simpler implementation compared to Theta* Pathfinding.

This package does not require a graph system and does not give optimal paths, however it seems that for a 2D rts style game this is sufficiently fast and gives sufficiently good paths. Importantly, the difficult of calculating the path does not depend on the map size but rather the complexity of the path produced. This ensures that even when the paths are slow it is intuitive why the paths are slow and thus easier for level designers to avoid pathological cases.

Videos

The following video demonstrates paths from this library. Click for higher quality

Paths example

Installation

Open either a git for windows terminal (windows) or the standard terminal (linux, mac). Move into the solution folder (i.e., cd into the folder containing the .sln folder for the project you want to use this pathfinder in). Run the following:

mkdir SharpMath2
cd SharpMath2
git init
git remote add origin "[email protected]:Tjstretchalot/SharpMath2.git"
git pull origin master
cd ..
mkdir AnyAnglePathfinding
cd AnyAnglePathfinding
git init
git remote add origin "[email protected]:Tjstretchalot/AnyAnglePathfinding.git"
git pull origin master

Then open up your project, right click the solution, "Add existing project", navigate to the sharpmath2.shproj file and add it. Repeat for the AnyAnglePathfinding package.

Right click "References" inside your main project. Under shared projects (left), check "SharpMath2" and "AnyAnglePathfinding".

Add the NuGet package High-Speed-Priority-Queue

Quickstart

var map = new RectPartitionAAMap<AACollidable>(200, 100);

var entity = new AACollidable()
{
    Position = new Vector2(50, 70),
    Bounds = ShapeUtils.CreateCircle(radius: 10, segments: 7)
};
map.Register(entity);

var entity2 = new TSCollidable()
{
    Position = new Vector2(100, 70),
    Bounds = ShapeUtils.CreateCircle(radius: 7, segments: 5)
};
map.Register(entity2);

var excludeIds = new HashSet<int>();
excludeIds.Add(entity.ID);

List<Vector2> path = new AAPathfinder(
        map, entity.Bounds, entity.Position,
        new Vector2(150, 70), excludeIds, 0
    ).CalculatePath();

Console.WriteLine($"entity 1 at {entity.Position}");
foreach(var loc in path)
{
    Console.WriteLine($"moved entity 1 to {loc}");
}

Usage

You will need a map class that implements AAMap<T>. This class will need to be able to trace polygons as they move through your world. In general your implementation can mirror the SimpleAAMap<T> class, except you should incorporate some space partitioning system in order to prevent the n^2 collision detection costs. Luckily you can get a great amount of benefit from the general implementation provided with RectPartitionAAMap<T> which will automatically partition your map into rectangles calculated from the location of entities:

Partitioning example

The most direct way to incorporate ThetaStarSharp into an existing project:

  • Have your entity class extend (or include) AACollidable
  • Have your map class include a RectPartitionAAMap<T> instance where T is your Entity class
  • Calculate paths via new AAPathfinder(map, bounds, start, end, excludeIds, excludeFlags).CalculatePath()
  • Move entities with Map.Move(int id, Vector2 pos) - some work needs to be done to keep the partitions accurate. Failing to do this will cause cascading errors.

Common Issues

When using the RectPartitionAAMap<T> it is essential that you do not modify the position of collidables directly, instead using Move(int id, Vector2 position) where id is the id of the collidable to move. If you see the pathfinder attempting to path through another entity, then the partition has broken. First use Visual Studio Code to find all references to the Position for your AACollidable class and verify it's not being modified anywhere. If it is not, turn on the compiler flag PARTITION_DEBUG and re-run. This will verify the state of the partition before and after each operation, which will allow you to isolate where the problem is occurring. If it happens on the second check within a function, it is a bug with this project. If it happens on the first check within a function (i.e., before it has done anything), it is a bug with your project.

If it doesn't catch any issues, then it must be something else. For this particular bug, verify that the entity is actually added to the pathfinder. This is as simple as checking Map.CollidablesLookup.ContainsKey(ent.ID).

Comparison to other pathfinding algorithms

This is not a grid-based pathfinding algorithm. Furthermore, tracing polygons across arbitrary distances requires polygon collision detection which is not trivial to implement on your own. By incorporating this project you get the flexibility of any-angle pathfinding on a plane in just a few lines of code. Furthermore you no longer have to worry about grid size as a hyperparameter in your pathfinding performance, which improves generalizability.

This package is not suitable if any of the following apply:

  • Movement is not identical on a plane (i.e. sidescrollers cannot use this package)
  • Your entities can only move in a limited number of directions (i.e. cardinal directions)
  • It is more expensive to move over some parts of the map then others, and you want this taken into account during pathfinding
  • Optimal pathing is more important than quickly finding paths
  • There are relatively few paths that will get you to the destination. In other words, the world is not relatively open.
    • Navigating through a maze will work but is not going to be fast using this algorithm

This package is meant for you (and you should submit issues to address any problems) if:

  • Your entities can move in any angle (i.e. an RTS)
  • Collidables are relatively sparse
  • You prefer human-looking paths to optimal paths
  • Paths will tend to have only a few nodes or be impossible
  • Some classes of objects do not collide with other classes of objects
    • For example, you want to move to attack a building and ignore any enemy that is in the way
  • Some paths will not collide with a small set of objects
    • For example a door that only one unit can walk through

Flags example

// define your flags by incrementing the number on the right.
// this gives you up to 63 flags. You can use an enum with [Flags]
// which subtypes long as well
long TEAM_1 = 1 << 0; // 2^0
long TEAM_2 = 1 << 1; // 2^1

var map = new RectPartitionAAMap<AACollidable>(2000, 1000);
var entity = new AACollidable()
{
    Position = new Vector2(50, 70),
    Bounds = ShapeUtils.CreateCircle(radius: 10, segments: 7),
    Flags = TEAM_1
};
map.Register(entity);

var entity2 = new AACollidable()
{
    Position = new Vector2(100, 70),
    Bounds = ShapeUtils.CreateCircle(radius: 7, segments: 5),
    FLAGS = TEAM_2
};
map.Register(entity2);

var excludeIds = new HashSet<int>();
excludeIds.Add(entity.ID);

List<Vector2> path = new AAPathfinder(
        map, entity.Bounds, entity.Position,
        new Vector2(150, 70), excludeIds, TEAM_2 // ignores team 2 (i.e., entity2) -> direct path
    ).CalculatePath();

Console.WriteLine($"entity 1 at {entity.Position}");
foreach(var loc in path)
{
    Console.WriteLine($"moved entity 1 to {loc}");
}

anyanglepathfinding's People

Contributors

tjstretchalot avatar

Stargazers

 avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar  avatar

Watchers

 avatar  avatar  avatar  avatar  avatar

anyanglepathfinding's Issues

Replace collision-only checks with GJK algorithm

It should be possible to improve performance significantly by replacing the locations in the overall pathfinding which use the SAT algorithm but only need collision yes/no answer with the GJK algorithm.

In particular, the SimpleAAMap should use a GJK algorithm for traces:

This change will likely require changes to SharpMath2 to support GJK

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.