Optimal Kernel Classifier in Mobile Robots for Determining Gases Type

Abstract

The use of TGS sensor and Arduino could create a robot to be capable of detecting and classifying some gases. In this research, 3 kinds of SVM Kernel Classifiers are investigated. Robots equipped with 3 TGS sensors are used to classify methanol and acetone. Xbee modul is used as a communication medium between robots and server. The robots are run in the experimental environment. When they detect the gas, they will get closer to the source and classify the gas type. The classified gas data is then sent to the server. From this research, it can concluded that polynomial and RBF have better performance in classifying methanol and acetone.