Self Driving Car in a Constrained Environment - Computer, Communication, and Signal Processing Access content directly
Conference Papers Year : 2022

Self Driving Car in a Constrained Environment

B. Harish
  • Function : Author
  • PersonId : 1334344
D. Thenmozhi
  • Function : Author
  • PersonId : 1117233

Abstract

The purpose of the research is to build a machine learning model which can drive a car on the tracks of Udacity’s Car simulator without any human intervention. This is achieved by mimicking the human driving behaviour in the training mode on a track. A dataset is generated by the simulator based on the human driving behaviour in the training mode and a deep learning model is built using this dataset which is then used to drive the car autonomously on any unseen track. Initially the model performed well only on the already seen track and failed to perform well on new unseen tracks. The simulator track in which the car was trained with didn’t consist of any sharp turns or elevations or any other road barriers, but the real world tracks do contain them, so in order to overcome this problem image processing techniques like zooming, changing brightness, flipping images, panning were used and in order to avoid over-fitting problem more dataset was generated using data augmentation techniques. Finally a model was built which was able to generalise the tracks and drive the car autonomously on the unseen track of the simulator.
Embargoed file
Embargoed file
0 7 13
Year Month Jours
Avant la publication
Wednesday, January 1, 2025
Embargoed file
Wednesday, January 1, 2025
Please log in to request access to the document

Dates and versions

hal-04388146 , version 1 (11-01-2024)

Licence

Attribution

Identifiers

Cite

B. Harish, D. Thenmozhi. Self Driving Car in a Constrained Environment. 6th International Conference on Computer, Communication, and Signal Processing (ICCCSP), Feb 2022, Chennai, India. pp.325-339, ⟨10.1007/978-3-031-11633-9_23⟩. ⟨hal-04388146⟩
12 View
0 Download

Altmetric

Share

Gmail Facebook X LinkedIn More