Multi-Action Situational Response by Jason Madden 5/1/2008.
-
date post
20-Dec-2015 -
Category
Documents
-
view
213 -
download
0
Transcript of Multi-Action Situational Response by Jason Madden 5/1/2008.
Motivation
Autonomous agents used in real world applications
Need methods to control without having to interact with the agents
Can Improvements be made?
The Problem
•What is X-Pilot?
–2-D multi player space combat game.
•What makes X-Pilot a good test bed.
–Finite, simplified( in respect to the physical world) environment
Problem Definition
Direct input Relational Input
Represents a direct readof the state variables
● Velocity
● Direction
● Wall distance
● Enemy distance
Allows knowledge of thedifference between the ship'sdirection and another angle
including:
● target enemy direction
● bullets predicted contact
angle
● enemy's heading
● bullet tracking
● projected firing angle
● and wall structure.
Goal
•To develop a autonomous controller with direct and relational input, that maximizes the time the agent is
alive.•Allowable actions: Thrust, turn, and shoot
Solution Method•Alteration of a GA developed by Matt Parker, Gary Parker and Timothy Doherty.1. Agent is approaching wall and is very close to wall.2. Agent is approaching wall and somewhat close to wall.3. Agent is approaching wall and at a moderate distance from wall.4. Agent is approaching wall at any distance and a bullet is incoming.5. Agent is approaching wall at any distance and an enemy is close and closing.6. Agent is approaching wall at any distance and an enemy is close but not closing.7. Bullet is incoming and very close to striking.8. Enemy is close and incoming bullet is close.9. Enemy is at a medium to far range and incoming bullet is close.10. Enemy is detected medium to close.11. Enemy is detected medium to far.12. Agent is entering a corner clockwise.13. Agent is entering a corner counterclockwise.14. Agent detects no walls, bullets or enemy agents.15. Wall detected 90° to left of agent.16. Wall detected 90° to right of agent.
Solution Method (changes)
•Changing the representation of a gene.
• Inspired from physics and biology
–Add some uncertainty for a situation
Maximum Fitness Results
01
23
45
67
89
1011
1213
14
0
5
10
15
20
25
30
Max Fitness this Generation
EA Max Fitness
Generation
Tim
e in
Seconds
Conclusions and Observations
•Evolution is slow•Competitive and Cooperative
•GA method strays away from a self building program.
•Determine a fitness function that prevents the reward for inactivity.
Problem Statement
Given a spaceship within the X-Pilot environment and a set of environmental inputs, create an automated control for the space ship. The inputs include direct input which represent a direct read of the variables used in X-Pilot and relational input. Relational input allows for the comparison of two angles.
This will find the difference between the ship's direction and another angle including target enemy direction, bullets predicted contact angle, enemy's heading, bullet tracking, projected firing angle, and wall structure. The automated control is allowed to turn the ship, thrust, and shoot bullets.