Object Recognition Using Multi-Input Sensor with Convolutional Neural Network

Abstract

Object recognition is applied in many areas, such as robotic systems, introducing many techniques. This article presents object recognition using the multi-input sensor with the Convolutional Neural Network (CNN). The tactile sensor and flex sensor are investigated to use in this system. The objects input uses ten different objects, and the CNN is designed to be the recognition technique. The results show that the multi-input sensors have combined tactile sensor and flex sensor can provided the best recognition efficiency rate is 97.60% when compared with single sensor.