Adaptive Robot Soccer Defence Strategy via Behavioural Trail

Adaptive Robot Soccer Defence Strategy via Behavioural Trail

Awang Hendrianto-Pratomo (Center for Artificial Intelligence Technology, Universiti Kebangsaan Malaysia, Malaysia, & University Pembangunan Nasional “Veteran” Yogyakarta, Indonesia), Anton Satria Prabuwono (Center for Artificial Intelligence Technology, Universiti Kebangsaan Malaysia, Malaysia), Siti Norul Huda Sheikh Abdullah (Center for Artificial Intelligence Technology, Universiti Kebangsaan Malaysia, Malaysia), Mohammad Faidzul Nasrudin (Center for Artificial Intelligence Technology, Universiti Kebangsaan Malaysia, Malaysia), Muhamad Syafiq Shohaimi (Center for Artificial Intelligence Technology, Universiti Kebangsaan Malaysia, Malaysia) and Teddy Mantoro (Advanced Informatics School, Universiti Teknologi Malaysia, Malaysia)
Copyright: © 2012 |Pages: 21
DOI: 10.4018/jitr.2012070102
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

Robot soccer is a challenging platform for multi-agent research, involving topics such as real-time image processing and control. A team of robots must work together to put the ball in the opponent’s goal while at the same time defending their own goal. A good strategy for the robot defenders can determine who wins the robotic soccer game. Therefore, the goal of this study is to propose a strategy for the defenders using a production rule based on state diagrams. The rule can facilitate easy and rapid comprehension of certain behaviors with respect to two indicators, such as condition and action. The authors determine five key aspects as conditions, including the positions of two defender robots, the position of the goalkeeper and the ball, and coordination between two defender robots and the goalkeeper robot. Each robot has been set its own defence area and specific actions. They conducted three experiments namely simulator testing, real time testing, and ping pong testing to evaluate their proposed defence strategy. The experimental results show that the authors’ proposed strategy versus three state of the art strategies can defeat up to 92% of all types of attack modes. Meanwhile, in the ping pong testing, their proposed strategy can still protect any goal entering from different attacking modes even though only one or two robots are active in the defence area.
Article Preview

1. Introduction

Robot soccer is a challenging platform for multi-agent research that involves topics such as real-time image processing and control. Robot soccer was introduced in 1994 with a theoretical background to develop multi-robot adaptive, co-operative and autonomous systems that can solve common tasks (Kim & Shim 1997). Furthermore robot soccer is a rich domain for the study of learning issues (Kim, Lee, & Fukuda, 1997).

Each team of robots must work together to put the ball in the opponent’s goal while at the same time defending their own goal, and the team must be able to adapt to new situations and the behavior of various opponents (Pratomo et al., 2010). Additionally, they must learn to work together autonomously and cooperatively (Han, Lee, Moon, Lee, & Kim, 2002; Kim, Hwang, & Kwon, 2003; Wang, Liu, Li, & Yang, 2010; Pratomo et al., 2011). Moreover, the robots also must interact and self-organize autonomously to achieve a common goal (Liu, Liang, & Lin, 2004). Apart from cooperative and coordinated behavior, further technical aspects such as precision of movement and optimization of power efficiency must also be taken into account because robot soccer is well-known as a miniaturization of a complex electro-mechanical system (Prabuwono, Burhanudin, & Said, 2008; Hwang, Tan, & Chen, 2004).

The robots can learn any collaborative or adversarial techniques; however, they must first acquire some low‐level skills that allow them to manipulate the ball (Pozna, Troester, Precup, Tar, & Preitl, 2009). These low‐level skills, for example, dribbling and controlling the ball, though entirely individual in nature, are crucial to develop because these skills are involved in more collaborative tasks, such as sending and receiving passes. In relation to the rich and dynamic nature of the environment in soccer (Zabawi & Omar, 2011), it is either impractical or impossible to directly program the robots to play soccer. Therefore, much of the focus has been shifted to treating the robots as intelligent agents that can learn how to play soccer under different environments and conditions (Hwang et al., 2006; Zhao, Zhang, Wei-Jie, & Li, 2006).

Based on Figure 1, the formation of various strategies that can be used in a robot soccer game depends on the pattern of the game and the types of players. The game pattern that focuses solely on location coordinates will produce a centralized game that is focused on the direction of the ball movement (Plyne et al., 2008; Xiao-Jun et al., 2006). Every robot plays different roles at different locations.

Figure 1.

Robot soccer formation modified (Kim & Shim, 1997; Kim et al., 1997, 2003)

Complete Article List

Search this Journal:
Reset
Open Access Articles: Forthcoming
Volume 10: 4 Issues (2017)
Volume 9: 4 Issues (2016)
Volume 8: 4 Issues (2015)
Volume 7: 4 Issues (2014)
Volume 6: 4 Issues (2013)
Volume 5: 4 Issues (2012)
Volume 4: 4 Issues (2011)
Volume 3: 4 Issues (2010)
Volume 2: 4 Issues (2009)
Volume 1: 4 Issues (2008)
View Complete Journal Contents Listing