SteeraTool: Exploiting the potential of digital twin for data generation

Autonomous driving heavily relies on accurate lateral control to ensure safe and reliable operation. In this paper, we present a novel approach to exploit the potential of digital twins for steering angle prediction in autonomous driving. Our method combines the use of a custom-built tool, SteeraTool, for generating a high-fidelity dataset, SteeraSet, of steering angle data with the implementation of a simple deep neural network architecture. The dataset was collected through simulations in a diverse range of scenarios. The neural network model was trained and evaluated on the generated dataset, and achieved promising results. Our work lays the groundwork to leverage the potential of digital twins in the area of lateral control. Moreover, SteeraTool can be used as a testbed in this area.