Articles

OPTIMIZING OF BOXING AGENT BEHAVIOR USING ELITISM BASED GENETIC ALGORITHM

Main Article Content

Anang Kukuh Adisusilo Mochamad Hariadi Ahmad Zaini Supeno Mardi Susiki Nugroho

Abstract

OPTIMIZING OF BOXING AGENT BEHAVIOR USING ELITISM BASED GENETIC ALGORITHM aAnang Kukuh Adisusilo, bMochamad Hariadi, cAhmad Zaini, d Supeno Mardi Susiki Nugroho aDepartment of Information Engineering, Faculty of Engineering, University of Wijaya Kusuma Surabaya b,c,dIntelligent Network Expertise Multimedia Department of Electrical Engineering, Faculty of Industrial Technology, Sepuluh Nopember Institute of Technology Surabaya Email: anang@anang65.web.id Abstrak Pola perilaku agen tinju pada permainan tinju dipengaruhi oleh beberapa faktor antara lain teknik gerakan bertinju, jenis pukulan tinju, stamina, dan energi pukulan. Pola perilaku agen tinju secara umum menggunakan variabel random dengan distribusi event dari setiap state secara acak. Penelitian dengan menggunakan FSM (Finite State Machine) berbasis algoritma genetika, menghasilkan nilai fitness 0.96, untuk pola perilaku agen cenderung maju kearah lawan, energi pukulan cenderung sedikit, dan menggunakan jenis pukulan dengan objektivitas tinggi. Penelitian ini menggunakan fungsi elitism pada algoritma genetika untuk dapat menghasilkan nilai fitness yang stabil dan pola perilaku agen tinju yang lebih baik dibandingkan tanpa menggunakan fungsi elitism. Nilai fitness yang dihasilkan dari penelitian ini diantara 3.101748 sampai 3.14738 dan nilai fitness optimal diantara 2.78083 sampai 3.167174, dengan siklus algoritma genetika lebih besar sama dengan generasi ke-25. Pola perilaku agen tinju yang dihasilkan berdasarkan nilai fitness adalah menyerang menggunakan satu jenis pukulan uppercut right dan tiga pukulan jab, dengan energi pukulan diantara 48 sampai 52 dan pola permaianan cenderung maju sambil melindungi wajah (covered). Pola perilaku agen tinju dari nilai fitness adalah menyerang menggunakan satu jenis pukulan uppercut right dan tiga pukulan jab dengan energi pukulan diantara 48 sampai 52 dengan pola permaianan cenderung maju dan melindungi wajah (covered). Kata kunci: Perilaku agen, Algoritma Genetika, Optimasi, Permainan Tinju, FSM. Abstract Boxing agent behavior patterns in the game of boxing is affected by several factors, i.e. the technique of boxing movements, type of punch, stamina, and energy of the punch. Boxing agent behavior patterns in general use variable random event where is state distribution randomly. A study using FSM (Finite State Machine) based on genetic algorithms, resulting fitness value 0.96 for boxing agent behavior patterns that tend to move towards the opponent , used energy to blow is likely small, and it uses the kind of blow that has high objectivity. This study will utilize elitism function in genetic algorithms to produce a stable fitness and better boxing agent behavior patterns than the one use genetic algorithms without elitism function. Fitness value result from this study between 3.14738 and 3.101748 and the optimal fitness value between 2.78083 to 3.167174, with a genetic algorithm cycle equal or more than the 25th. The boxing agen behavior patterns generated from fitness value is to attack using single type of blow, right uppercut punch and jab with a three-punch blow energy between 48 to 52 and patterns game that tend to move foward with covered the face. Key words: Agent behavior, Genetic Algorithm, Optimation, Boxing game, FSM

Downloads

Download data is not yet available.

Article Details

How to Cite
ADISUSILO, Anang Kukuh et al. OPTIMIZING OF BOXING AGENT BEHAVIOR USING ELITISM BASED GENETIC ALGORITHM. Jurnal Ilmiah Kursor, [S.l.], v. 7, n. 2, feb. 2016. ISSN 2301-6914. Available at: <https://kursorjournal.org/index.php/kursor/article/view/42>. Date accessed: 17 dec. 2018.
Section
Articles