Autonomous maneuvering decision research of UAV based on experience knowledge representation

An autonomous maneuvering decision algorithm based on experiences of pilots and tactics theories in one-to-one air combat is proposed. Firstly a method of situation assessment based on Naive Bayes is addressed, in which the historical situation is taken into consideration when predicting the next mo...

Full description

Saved in:
Bibliographic Details
Main Authors: He, Xuemei, Zu, Wei, Chang, Hongxing, Zhang, Jie, Gao, Yang
Format: Conference Proceeding
Language:English
Subjects:
Air Combat
Aircraft
Algorithms
Assessments
Atmospheric modeling
Bayes
Decision theory
Experiential Knowledge
Games
Mathematical model
Mathematical models
Nash equilibrium
Pilots
Tactics
Unmanned aerial vehicles
Unmanned aircraft
Weapons
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:An autonomous maneuvering decision algorithm based on experiences of pilots and tactics theories in one-to-one air combat is proposed. Firstly a method of situation assessment based on Naive Bayes is addressed, in which the historical situation is taken into consideration when predicting the next moment situation. Then preferences of experienced pilots and tactics theories are mathematically modeled. This model is more flexible and extensible than that models made up from a bunch of rules, because the mathematical model is easier to code and can adaptively adjust flying state of aircraft according to the output of situation assessment. Finally a maneuvering scoring function and a maneuvering matrix are described. The whole process of combating is regarded as a static complete information game where an optimal maneuvering solution by Nash equilibrium solution exists. Simulations are devised by one unmanned aircraft running the proposed algorithm against one manned aircraft adversary. Results show that the unmanned aircraft will reach an advantageous situation regardless of starting from offensive or defensive state.
ISSN:1948-9447
DOI:10.1109/CCDC.2016.7530973