An Intelligent Error Correction Algorithm for Elderly Care Robots

With the development of deep learning, gesture recognition systems based on neural network have become quite advanced, but application effect in elderly is not ideal. Due to change palm shape elderly, rate most people only about 70%. Therefore, this paper, an intelligent error correction algorithm game rules proposed basis AlexNet. Firstly, paper studies differences between palms and young people. It also analyzes misread by using probability statistics method establishes a misread-gesture database. Then, misreading-gesture library, maximum channel number different gestures fifth layer studied similar curve Pearson algorithm. Finally, completed under rule. The experimental results show that can be improved more than 90% elderly-accompanying robot understand people’s intentions accurately, which well received users.