Abstract:The rapid evolution of new technologies in the automotive sector is driving the demand for advanced simulation solutions, enabling faster software development cycles. Developers often encounter challenges in managing the vast amounts of data generated during testing. For example, a single Advanced Driver Assistance System (ADAS) test vehicle can produce several terabytes of data daily. Efficiently handling and distributing this data across multiple locations can introduce delays in the development process. Moreover, the large volume of test cases required for simulation and validation further exacerbates these delays. On-premises simulation setups, especially those dependent on High-Performance Computing (HPC) systems, pose several challenges, including limited computational resources, scalability issues, high capital and maintenance costs, resource management inefficiencies, and compatibility problems between GPU drivers and servers, all of which can impact both performance and costs.To address these issues, transitioning to cloud-based simulations using AWS services—such as S3, Batch, Auto Scaling Group (ASG), and Elastic Container Registry (ECR)—allows for parallel execution of Software-In-the-Loop (SIL) tests. Multiple compute instances can be deployed simultaneously, significantly reducing testing times. Additionally, GPU-based simulations can be executed on cost-effective instances, potentially reducing simulation costs by up to threefold. The Docker container used in these simulations packages the necessary software algorithms, simulation tools, and test recordings. Key driving functions, including Blind Spot Detection (BSD), Lane Change Assist (LCA), and Occupant Safety Exit (OSE), are tested using real-world driving scenario data to ensure their effectiveness. The workbench offers an innovative environment for developing software solutions using virtual ECUs, even before the physical hardware becomes available. Furthermore, the automotive operating system utilized for software development provides robust middleware, ensuring secure and standardized communication between computers and the cloud.

ADAS Simulation on Cloud

Simulating Cloud Environments of Connected Vehicles for Anomaly Detection

AADS: Augmented autonomous driving simulation using data-driven algorithms

Distributed Simulation Platform for Autonomous Driving

Diagnosing and Predicting Autonomous Vehicle Operational Safety Using Multiple Simulation Modalities and a Virtual Environment

EdgeDrive: Supporting Advanced Driver Assistance Systems Using Mobile Edge Clouds Networks

Driving Simulation Platform Applied to Develop Driving Assistance Systems

An Approach to Systematic Data Acquisition and Data-Driven Simulation for the Safety Testing of Automated Driving Functions

Cyber-Automotive Simulation and Evaluation Platform for Vehicular Value Added Services

Cloud Services Enable Efficient AI-Guided Simulation Workflows across Heterogeneous Resources

Leveraging Cloud Computing to Make Autonomous Vehicles Safer

Closed-Loop Simulation Test Method for Autonomous Vehicles Based on Adversarial Scenarios

Customized Co-Simulation Environment for Autonomous Driving Algorithm Development and Evaluation

Simulation-based Safety Assurance for an AVP System incorporating Learning-Enabled Components

An Advanced Framework for Ultra-Realistic Simulation and Digital Twinning for Autonomous Vehicles

Off-Road Autonomy Validation Using Scalable Digital Twin Simulations Within High-Performance Computing Clusters

Hardware-In-the-Loop for Connected Automated Vehicles Testing in Real Traffic

LGSVL Simulator: A High Fidelity Simulator for Autonomous Driving

Vehicle-in-the-Loop Intelligent Connected Vehicle Simulation System Based on Vehicle-Road-Cloud Collaboration

Testing driver warning systems for off-road industrial vehicles using a cyber-physical simulator