Object Recognition for Humanoid Robots Using Full Hand Tactile Sensor

With the development of human-like robotic technologies, robots have been advanced to sense and recognize objects, which is one branch artificial intelligence mimic human reactions. Although high-performance computer technologies are used, ability object recognition by touch still low due lack proper sensors. Therefore, this research studied learning system through robot touches developing an sensory acting as electronic skin with tactile The Tactile Sensor developed in research, consisting 15 Arrays palm’s touchpoints. Furthermore, analysis was on Bag Word (BoW) Convolution Neural Network (CNN) algorithms. BoW technique, using Support Vector Machine (SVM) a classifier Moment Analysis Descriptor (MA) provided highest accuracy, showing more than 80.15% accuracy from five grasping object. CNN approach, InceptionNetV3 98.28% only capture