CAIAS Simulator: Self-driving Vehicle Simulator for AI Research

This paper presents a simulation environment which includes virtual structures of a low-cost embedded designed car for the autonomous driving test, tracks, obstacles, and environments. A cross-platform game engine, Unity 3D, empowers the embedded designed car to check and trial new tracks, parameters and calculations in the 3D environment before the real-time test. The virtual environment fabricates the domain such like that it is the mimics of the activity of a genuine car and Unity 3D are utilized to incorporate the embedded designed car into the test situation while the car’s movements and steering angle can serve as an examination premise. Distinctive driving situations were utilized to analyze how the sensors respond when they are connected to genuine circumstances and are also utilized to confirm the impacts of other parameters on the scenes. Options are available to choose flexible sensors, monitor the output and implement any autonomous driving, steering prediction, deep learning and end-to-end learning algorithm.