Utilizing Advanced Deep Learning with Dynamic Capsule Networks & Self-Attention for Steering Angle Prediction
This project implements an end-to-end autonomous driving system using deep learning-based behavioral cloning for steering angle prediction. The model leverages Dynamic Capsule Networks and Self-Attention Mechanisms to enhance feature extraction and decision-making. The final model is deployed for real-time inference to simulate autonomous driving behavior.
End-to-End Learning – Uses raw images as input and predicts optimal steering angles.
Advanced Deep Learning Models – Implements Dynamic Capsule Networks for spatial awareness and Self-Attention Mechanisms for improved context understanding.
Real-Time Inference – Deploys the trained model for live autonomous driving simulation.
Behavioral Cloning – Learns from human driving behavior to make safe and smooth driving decisions.
Edge Deployment Ready – Optimized for efficient inference on embedded devices like NVIDIA Jetson.