This project explores how Equivariant Graph Neural Networks (EGNNs) enhance predictive accuracy in point cloud data analysis. The results aim to advance the processing of unstructured 3D data for autonomous vehicle perception and forecasting.