CARLA Hardware-in-the-Loop Simulation Environment for Autonomous Driving

This article suggests a validation methodology for autonomous driving. The goal is to validate front camera sensors in advanced driver-assist systems (ADAS) based on virtually generated scenarios. The outcome is the CARLA-based hardware-in-the-loop (HIL) simulation environment (CHASE). It allows the rapid prototyping and validation of the ADAS software. We tested this general approach on a specific experimental application/setup for a vehicle front camera sensor. The setup results were then proven to be comparable to real-world sensor performance. The CARLA simulation environment was used in tandem with a vehicle CAN bus interface. This introduced a significantly improved realism to user-defined test scenarios and their results. The approach benefits from almost unlimited variability of traffic scenarios and the cost-efficient generation of massive testing data.