ACADSTAFF UGM
| Title | : | SIMULATION OF AUTONOMOUS CAR STEERING CONTROL USING END-TO-END DEEP LEARNING APPROACH WITH CNN METHOD |
| Author | : |
Muhammad Idham Ananta Timur, M.Kom. (1) Prof. Dr. Ir. Jazi Eko Istiyanto, M.Sc. (2) Dr. Andi Dharmawan, S.Si., M.Cs. (3) |
| Date | : | 0 2020 |
| Keyword | : | autonomous car, convolutional neural network, deep learning, AirSim simulator autonomous car, convolutional neural network, deep learning, AirSim simulator |
| Abstract | : | An autonoumous car is a vehicle that is trained to make decisions related to driving using artificial intelligence, in which a computer has a role to take over and fully control the steering. In this research, a model is designed to train a car which can run automatically using end-to-end deep learning method. This car is trained in an environment built by the AirSim simulator. Using a simulator is suitable for training the model because simulators have a potential to produce unlimited amounts of data. The dataset for training is taken by manually driving the car, using the record feature in the AirSim simulator to store the dataset. This dataset consists of vehicle’s condition labels and images which are taken through a single camera mounted on the car’s dashboard. Convolutional Neural Network (CNN) is used to process the image and labels of the dataset while training the model. The results of this research is a simulation that can control the steering wheel of a car, in which the trained model make the predictions for the steering angle based on image input from the camera. The accuracy of the trained model is measured through the RMSE calculation which results in a value of 0.178. This value indicates that the trained model has high accuracy because the low RMSE value means that the variation of predicted values is close to the variation of actual values. |
| Group of Knowledge | : | Ilmu Komputer |
| Original Language | : | English |
| Level | : | Internasional |
| Status | : |
Draft
|
| No | Title | Action |
|---|---|---|
| 1 |
Jurnal End-to-end learning.pdf
Document Type : Bukti Draft
|
View |