Adjuvant Therapy For Children With ASD Cognitive Affective Method Of Human-computer Interaction

The autism spectrum disorders is a children’s pervasive developmental disorder, it has seriously hindered their and other people’s learning and living. Due to the complexity of ASD etiology and individual differences, there is no effctive drug. The study found that these children are more likely to accept simple and single things, the behavior and characteristics make them easier to like some computer software or system. Therefore, it have a great significance to establish a harmonious,affective interaction system.In the further study of the intelligent robots and artificial emotion theory, emotional interaction model framework was established. Constructed emotional intelligence based on emotional cognitive evaluation theory and emotional decision-making ability. It get information from outside by attention model. The outside information processing was completed by emotion module to get the emotional state of interactive object. Put the emotional state as an emotional input of cognitive modules, and take advantage of the emotion transfer model based on markov. Make the agent to achieve the same emotional state with interactive object under the stimulus of the outside world information. Then the emotional decision module make a corresponding emotional behavior according to the current emotional state.Facial feature location and extraction was completed by improved AAM algorithm. The algorithm using the shape parameter to combine the parameters of ASM and AAM into a single one.In the emotional cognitive decision module, Agent improve the traditional Q learning algorithm based on reinforcement learning theory. They also considered the agent internal incentives and rewards from the environment. While using the BP neural network in the Q-learning Q function. Put the emotional behavior role to people through the compute and get facial features of people. To help these children to facilitate understanding of their own emotional world and improve their ability of expression.The experiments results show that the proposed feature point location and feature extraction algorithm can improve the stability and accuracy of system. Through the simulation experiment proves that the emotional intelligence has good emotional decision-making ability, verify the feasibility of the system.