Development Neural Network Models on Prediction of Human Purpose

Abstract

As human-robotic (HRI) interactions are the, more main determinants are predicted. However, the large percentage of robotics follows strict guidelines that seriously limit their mobility and functionality. A key fact which contradicts the understanding of HRI is that robots are unable to understand social desires. The goal of this study is to enhance robotic intelligence by educating people to understand the purposes of humans. Contrary to the previous report, this paper offers a mechanism for predicting human intentions before a single procedure is completed. The test of bouncing a baseball at designated targets is carried out to check the process' performance. The deep learning approach proposed shows that neural networks (CNNs) can be completely used in a new situation. Experimental results suggest that three conventional machine training strategies compete with the proposed CNN-vote system. The CNN Vote Predictor achieves high test accuracy with comparatively less data available in the current context.