Typology of Neural Networks Used in Designing Autonomous Driving Systems for Electric Transport

The role of neural networks in autonomous driving systems, focusing on the application of various neural network architectures such as the PointNet, dynamic graph CNN, recurrent, convolutional, and autoencoder neural networks, is considered. The advantages of using neural networks to process incoming sensory data, including cameras, lidars, and radars, are discussed, which can improve the safety and efficiency of autonomous vehicles. Specific tasks such as object recognition, glitch filtering, route optimization, and adaptation to changing conditions are considered. This paper highlights the need for further research and development of neural network algorithms to improve the reliability and robustness of autonomous systems in real-world conditions.