A Multimodal Fusion Plane Selection Algorithm for a Multidimensional Intelligent Interaction Pen

In geometry education, traditional methods like chalkboard instruction or geometry software often fail to fully engage students due to their two-dimensional and static nature. Interactive methods commonly used in virtual experiments, such as voice and gesture commands, do not adequately support precise selection required in geometry teaching. Our paper introduces an innovative intelligent pen and a virtual experiment platform specifically designed for geometry education. We have developed a multimodal fusion-based algorithm that interprets user intentions for selecting geometric planes in virtual space, utilizing historical interaction data. Additionally, we propose a dynamic weight updating algorithm that leverages past error information to enhance decision-making, thereby improving accuracy and reducing cognitive load. Our approach has demonstrated a high accuracy rate of 96.00% in experiments and has received positive feedback from users, indicating significant potential for enhancing interactive geometry education.