Comparison Between A* And Obstacle Tracing
Pathfinding In Gridless Isometric Game
Lailatul Husniah
Informatics Department
Universitas Muhammadiyah Malang
Malang, Indonesia
husniah@umm.ac.id
Eko Cahyono
Informatics Department
Universitas Muhammadiyah Malang
Malang, Indonesia
ekobudi@umm.ac.id
Rizky Mahendra
Informatics Department
Universitas Muhammadiyah Malang
Malang, Indonesia
rizky.ade.mahendra@gmail.com
Ali Sofyan Kholimi
Informatics Department
Universitas Muhammadiyah Malang
Malang, Indonesia
kholimi@umm.ac.id
Abstract— The pathfinding algorithms have commonly
used in video games. City 2.5 is an isometric grid-less game
which already implements pathfinding algorithms. However,
current pathfinding algorithm unable to produce optimal route
when it comes to custom shape or concave collider. This
research uses A* and a method to choose the start and end
node to produce an optimal route. The virtual grid node is
generated to make A* works on the grid-less environment. The
test results show that A* be able to produce the shortest route
in concave or custom obstacles scenarios, but not on the
obstacle-less scenarios and tight gap obstacles scenarios.
Keywords—game, pathfinding, A*, grid-less, and isometric
I. INTRODUCTION
Video games genres and styles have evolved in the past
decades. The current technologies make it easier for a
developer to implement or combine one genre to another.
One of the genres that we studied is 2D simulation games
with isometric perspective. In the past time, most the 2D
isometric games rely on the grid as its base. Almost every
game-objects shown on that game has bounded to that grid.
Grid usage also makes it easier to implement pathfinding and
mapping. Common A* algorithm could have implemented
very well.
Unity3D is a game engine to make a high-quality 3D and
2D games [1]. Although it could produce a 2D game, all
game objects occupy a 3D space. Thus a default 2D game
would not become a grid based game. However, it still
possible to develop a grid-based 2D game by generating the
grid itself. Based on such occasion, to solve pathfinding
problems, we can either generate a virtual 2D grid to
implement basic A* pathfinding or create a custom
pathfinding method to adapt grid-less environment.
“City 2.5” is a grid-less 2D isometric city-building
simulation game. Since we are part of the developer who
develops this game, we can access and modify every part of
this game. City 2.5 use a custom pathfinding method called
Obstacle Tracing (OT) [2]. However, this method is not
optimal since it cannot always guarantee the shortest path.
The tendency of NPC (Non-Player Character) to encircling
the obstacle could make the path even longer. NPC is also
relying on obstacle collider convex shape which is means
when NPC found a concave collider the path taken would be
longer.
This previous study [3] use Fuzzy Mamdani Logic to
solve navigation problems that made better navigation on the
robot hexapod fire extinguisher. Other studies [4] use A* to
solve pathfinding problems in the 2.5D isometric game. The
game is grid-based, which preferable for A* algorithm. The
primary goal of that study is to investigate and determine the
optimal pathfinding strategy based on several measures such
as steps and time have taken to reach the goal using A*
algorithm [5].
A* itself is favorable as a pathfinding method as it is
simple to implement, is very efficient, and has lots of scope
for optimization [6,7]. A* able to generate the shortest path
because it uses a heuristic function to estimate the distance of
any point to target point [8]. A* implementation on Unity3D
has been done by [9], and so we need to adapt the
implementation method for this game.
A* is an algorithm which measures the heuristic distance
between a given point, while the pathfinding itself rely on the
search space on how the A* graph represented in the game.
In this study [10,11] there are several ways to represent the
search space. Previous research [12] use A* to solve
pathfinding method for a grid-based graph. In this research,
we will use the common A* search space representation, a
rectangular grid. The advantage of using the grid is it is
easier to generate automatically [13] and easy to implement
map representation [14].
Some studies about comparing pathfinding methods have
been done by [11,12,13,14,15]. Unfortunately, none of those
paper discusses pathfinding on grid-less environment
explicitly. What makes this paper different from previous
studies is the case study, grid-less environment, and the
method Obstacle Tracing, which is rarely used in other paper
and research The goal of this paper is to solve NPC
pathfinding problems on a grid-less 2D isometric game.
Therefore, A* is implemented and compare it with current
Obstacle Tracing method. This paper also focuses on how to
implement A* properly without altering the unique rules and
game design itself. Thus we will limit not to include the
computational complexity of the two algorithms, but
compare how the path has taken and the speed of
computation.
Proceeding of EECSI 2018, Malang - Indonesia, 16-18 Oct 2018
978-1-5386-8402-3/18/$31.00 ©2018 IEEE
489