Autonomous Surface Vehicle (ASV) Obstacle Avoidance Using Fuzzy Kohonen Network (FKN)

Abstract

Autonomous Surface Vehicle (ASV) is robot boats which can navigate autonomously to avoid obstacles in their path direction. The ASV has designed to detect and measure the distance of the position of the obstacle. The Fuzzy Kohonen Network (FKN) method is applying to the ASV as its brain to divine the manoeuvre what should do. The FKN is getting the information (Crips) from two sonar sensors, where are located in front of the ASV. In this experiment the FKN has four (4) pattern scenario which is three pattern normal condition, and one pattern danger condition. The three normal conditions define as if no obstacle detecting by two sensors, or either one of the sensor has detected an obstacle. The last condition pattern (danger) will occur when each sensor has detected the obstacle. These pattern condition is coding in C# and embedded into ATMega328. The range sensor is setting for 50-210 cm with the error rate is less than 1%. The manoeuvre of this ASV is providing two DC motors by controlling the PWM value. As the results of the experiment,The manoeuvre is quicker and smooth without a crash to obstacles.