INTELLIGENT ASTEROID AGENTS COMBINING PATHFINDING AND STEERING ALGORITHMS

INTELLIGENT ASTEROID AGENTS
COMBINING PATHFINDING AND
STEERING ALGORITHMS

Carlos Barboza

Kenny Barron

Kevin Cherry

Tung Le

Daniel
Lorio

ABOUT THE GAME


Avoid enemy ships as they chase after you


Every second alive adds points (10)


Killing an enemy adds points (100)


Player can wrap around screen, enemies can’t


Enemy can only propel forward, player can
move forward and in reverse


DEMONSTRATION

OBJECTIVE


Maximize agent’s performance in fully
-
dynamic, multi
-
agent environment with limited
knowledge of environment


Determine performance through the use of
different AI algorithms and parameters


Performance is gauged by score at end of
game

ORIGINAL GAME


“
FRENZY SURVIVOR”


User controlled
player


Limited ship
following


Disorganized
in pursuit of
player ship

http://berfenfeldt.com/

ENVIRONMENT



Partially Observable
–

Limited to set regions

´
Strategic
–

Moves based on location of enemy

´
Episodic
–

Experience based on perception of
enemy

´
Dynamic
–

Enemy constantly moving/regenerating

´
Discrete
–

Agent responds with set action based
on perception of enemies in viewed regions

´
Multi
-
Agent
–

Steering algorithm applied to
enemies, and pathfinding for player

P
–

Score

E
–

Grid


A
–

Moves


S
-

Grid Regions

ADAPTING THE ENEMY


Adapted open source project
OpenSteer in C#


Enemy determines velocity
and direction based on
players location


Steer for seek allows enemies
to converge around player


Steer for flee allows enemies
to distance themselves away
from player in any direction

ADAPTING THE PLAYER


Loosely based on A*
Pathfinding


Combines heuristic,
a
ctual
cost, and utility function to
quantitate each move choice


Agent chooses maximum
move value



ALGORITHM EVOLUTION



1
st

Agent was a simple reflex agent, not
partially observable, random movements and
actions, and it was not rational


2
nd

Environment was partially observable,
agent was a simple reflex agent, and partially
rational


3
rd

Agent was goal based agent, partially
observable environment, and partially rational

HEURISTIC FUNCTION (H)


Creates grid to discretize the game world

´
Each grid cell has bitmask that holds information
on cell contents


If enemy is in cell


If cell is part of a region

HEURISTIC FUNCTION (H)


Creates 5 regions from grid

´
Multiplies enemy presence with
proximity to player in each region
using a cubic scale.

«
A
–

Accelerates Forward

«
F
–

Flees Backward

«
L
–

Turns Left

«
R
–

Turns Right

«
S
–

Shoots

L






R







S








F









A

ACTUAL COST FUNCTION (G)


Concept of look ahead
implemented as a move tree for
actual score


1 point per move simulates
survival time

´
Prune on dead state (count dead
states)

´
Total dead states counted for each
move’s
subtree

´
Final move score:


MaxDescendentScore

* W
1

-

TotalDescendentDeadStates

* W
2

root

L

A

S

F

R

L

A

S

F

R

...

UTILITY FUNCTION (U)


Acts as a multiplier for move values, and tilts
the behavior of the player towards passive or
aggressive


A = 1.4


B = 1.0


F = 1.2


L = 0.8


R = 0.8


S = 1.2

FORMULA


f(m) = (h(m) + g(m)) * u(m)


where m = move


Overview:


h(m): Evaluate for each move and choose the one
with the largest value


g(m): Simulate gameState for each child, prune
dead states, and select move with highest


u(m): Utility function adds custom factor to each
value

RESULTS

1

2

3

4

5

6

7

8

9

10

AVG

BEST

H,U(8)

1204

1303

1105

1305

1304

1305

1304

1209

1611

1205

1285.5

1611

H(8)

1204

1208

1203

110

1230

816

1212

1304

1304

1330

1092.1

1081

H,U
-

EC(4)

1806

1705

2810

1911

1606

1708

1906

1806

5316

2709

2328.3

5316

H,U
-

Double(8)

1004

2416

4814

2306

2606

2605

2307

4412

4923

2607

3000

4923

H
-

Double(8)

2708

1606

3710

2508

2609

5218

1009

1408

2506

1805

2508.7

5218

H
-

Double(4)

2507

1205

1509

1103

2309

2107

1707

1505

4315

2409

2067.6

4315

0
1000
2000
3000
4000
5000
1
2
3
4
5
6
7
8
9
10
Avg
Best
H,U(8)
H(8)
H,U - EC(4)
H,U - Double(8)
H - Double(8)
H - Double(4)
DEMONSTRATION