Ingeniería e Investigación
versión impresa ISSN 0120-5609
This article presents the design, implementation, virtual simulation and test results for a hardware-software platform for robot soccer. The platform consists of a soccer field, a software-controlled mobile agent (AMC), a user-controlled mobile agent (MCU), a golf ball, an image sensor system and a path-planning controller. The problem for agent AMC is to sense the environment and move to a kicking position to score while avoiding opponents in its way. The soccer field is a bounded area of 1.5m x 2m in which mobile agents move following differential robot kinematics at 25.7cm/s speed; each robot has a kicking system which can speed up a golf ball at 1.43m/s velocity. A position sensor was implemented and optimised in three stages: capturing the soccer field image, converting to HSI format and applying digital filters and calculating the Cartesian components from colour patterns for identifying the mobile robots and the golf ball. Image sensor system processing speed is 30 frames per second. Robot path planning is controlled by a hybrid strategy based on two linearly combined fuzzy logic controllers; one takes agent AMC to the goal position and the other avoids obstacles while it moves to the goal position.
Palabras llave : fuzzy logic; path planning; RoboSoccer; mobile robot; artificial vision; image processing.