System configuration of Human-in-the-loop Simulation for Level 3 Autonomous Vehicle using IPG CarMaker

The increasingly automated vehicles (AV) have increased the complexity of the testing methods and number of driven miles required to demonstrate the vehicle system’s reliability. Most modern autonomous driving systems also used deep neural networks which requires a large amount of data to develop. Physical driving alone to collect driving data and test system’s safety is no longer suitable for development of AV as this is costly, time consuming and could harm the road users if the safety system failed. This paper proposed human-in-the-loop simulation testing for evaluation of an autonomous vehicle using a 3D virtual vehicle driving platform that can be used for safety assessment of autonomous vehicle. The aim of this study is to establish human-computer interaction platform that can be used as safety testing for Level 3 autonomous vehicle whereby an emergency takeover is required during critical driving conditions. The proposed platform make use of IPG CarMaker to provide 3D virtual environment with accurate vehicle dynamics model, sensor model and environment model. We are able to interface the IPG CarMaker with Simulink and successfully developed a Simulink model that can interface a steering and pedal driving hardware with the virtual vehicle in the simulation. We can also collect driving data and simulation data from the IPG CarMaker as well as accessing the variable in the IPG CarMaker in real-time using Python. The recorded data can be used to train and fine-tune autonomous system based on machine learning.