Immersive learning behavior analysis based on UWB positioning and innovative education mode

The rapid development of intelligent education and human-computer interaction has given rise to the demand for immersive learning, which has raised higher requirements for indoor spatial positioning accuracy and learning behavior recognition ability. To achieve centimeter level three-dimensional positioning and multi-dimensional behavior pattern recognition, a high-precision immersive learning behavior analysis framework for innovative education models is studied and constructed. Methodologically, a data acquisition and fusion system based on ultra wideband technology and nine axis inertial measurement units is constructed, and a two-layer robust Kalman filter combined with adaptive particle swarm optimization is introduced to achieve centimeter level position estimation; Simultaneously, a dual spatiotemporal graph convolutional network-Transformer model is proposed to deeply model typical behaviors of learners, such as collaboration, discussion, and presentation, from both spatial topology and temporal dependence perspectives. The experimental results show that the proposed immersive learning behavior analysis method has a root mean square error of only 3.5 ± 0.4 cm in three-dimensional positioning in an open classroom environment, an attitude angle error of about 1.0 ± 0.2 °, a behavior recognition accuracy of 96.8 ± 0.6%, and an F1 value of 96.1 ± 0.7%; Even under severe non-line of sight conditions with 60% occlusion, the F1 value remains at 91.5 ± 1.3%, and the average inference delay is only 2.2 ± 0.3 s. The research results show that the proposed method can achieve high-precision and strong robustness in learning behavior recognition under innovative education mode, providing a feasible solution for immersive smart classroom construction and personalized teaching intervention.