Physics Teaching System Based on Three Dimensional Convolutional Neural Network with Virtual Reality Technology

In this era, physics teaching system is required to visualize and comprehend various concepts such as energy conservation, collisions, and momentum transfer that need dynamic and interactive environments to bridge the gap between theoretical principles and real-world applications. The integration of three dimensional (3D) Virtual Reality (VR) technology helps to overcome the issue of computational complexity. The motive of this paper is to develop a system for effective teaching of physics using VR by categorizing the inherited data into three types based on collisions; elastic collision, inelastic collision and Complex systems like multiple object collisions. Initially, the data is inherited from the PhysX cohort that contains various physics related data such as collisions, dynamics and other physics phenomena. The collected data is processed through Bounding Volume Hierarchies (BVH) and Separating Axis Theorem (SAT). Further, the processed data is embedded into 3D Convolutional Neural Network (CNN) model since it has the ability to extract optimal spatial features. In the end, the extracted features are translated into VR model using robust tools such as Unity and NVIDIA. The proposed 3DCNN-VR model achieves greater results with accuracy of 98.68% when compared with Immersive VR (IVR).