In today’s fast-paced world, individuals are constantly seeking ways to optimize their time and safety, especially during commutes and travel. Time management, traffic congestion, and safety are critical aspects of driving that have driven researchers to develop autonomous vehicles. A critical challenge in developing autonomous vehicles is achieving reliable lateral motion control. This paper addresses this issue by leveraging the deep learning techniques, and specifically explores the benefits of the NVIDIA Udacity car simulator. Our approach involves training a convolutional neural network (CNN) model to simulate the human driving behavior by utilizing a dataset of images and corresponding steering angles recorded during manual driving sessions in the simulator. It also emphasizes the importance of preprocessing the visual data to enhance the accuracy of the steering predictions. Our model is capable of managing complex driving scenarios within a simulated environment without the risks associated with real-world testing. A Mean Squared Error of 0.0123422 and an R-squared value of 0.8587687 suggest that our model is highly effective at navigating the car across different unfamiliar tracks.
Autonomous Vehicle Utilizing Deep Learning Methods
21.11.2024
920859 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
AUTONOMOUS VEHICLE SYSTEMS UTILIZING VEHICLE-TO-VEHICLE COMMUNICATION
Europäisches Patentamt | 2020
|Autonomous Vehicle Simulation Using Deep Reinforcement Learning
TIBKAT | 2021
|CONTROLLING AN AUTONOMOUS VEHICLE WITH DEEP REINFORCEMENT LEARNING
British Library Conference Proceedings | 2019
|