The work aims to propose a research method of cartoon character modeling design based on emotional experience so as to provide an optimal design idea for the character modeling design of intervention APP in the face of the increasing number of ASD children in China. Firstly, combined with literature, 8 representatives were screened out from 20 HRV indicators, and the data of the 8 indicators were collected through HRV experiments. Secondly, based on the improved semantic scale, the emotional experience scores were arranged. Then, stepwise regression analysis was used to screen out the 3 main HRV factors that had the greatest impact on emotional experience scores. Finally, multiple regression equations were established to explore the regression relationship between emotional experience score and 3 HRV indicators. It was found that ASD children had the most obvious emotional experience feedback on the three types of characters, and the characters with main outline in square and circle and trapezoidal and circular components could improve LF/HF and SDNN and reduce M–HRT, thus obtaining a pleasant emotional experience. Integrating the improved semantic scale to establish a multiple linear regression model between emotional experience and three HRV indicators provides a theoretical basis for the design and scheme evaluation of APP cartoon character modeling of ASD children's intervention, and the character modeling design in line with ASD children's emotional experience can stimulate children's interest, improve interactive experience and enhance the effect of intervention.