dSPACE partners with AWS to enhance its portfolio of solutions for Automotive development and testing. By collaborating with AWS, dSPACE is combining the premium quality of its industry-leading solutions with the power, flexibility, and scalability of the AWS cloud to support modern data-rich and computation-intensive automotive developments. dSPACE and AWS are working together to accelerate the automotive development and testing cycle by delivering the most innovative, scalable, and efficient set of data acquisition, simulation and testing solutions to automotive engineers across the world.
SIMPHERA powered by AWS
Web-based, scalable solution powered by AWS for the simulation and validation of autonomous driving functions
SIMPHERA is a software solution for simulating and validating applications for autonomous driving. The software brings them to life at an early stage and helps you analyze them and handle their complexity. How? By giving you easy access to powerful functions for simulating vehicle dynamics, traffic, and more. So you can try new algorithms, such as for vehicle control, in the virtual world: simpliCity.
SIMPHERA Reference Implementation for AWS
The SIMPHERA Reference Implementation for AWS enables AWS customers to easily setup their system and deploy SIMPHERA. It contains everything needed for a customer to know how to configure the different sub-systems like the storage, the computation nodes and the databases that are needed for a seamless use of SIMPHERA. The reference implementation guides the customer and enables and quick start into the world of SIMPHERA.
Customer Success Stories
As in many research labs and autonomous driving companies, the Queensland University of Technology measurement car generates terabytes of sensor data while driving. The difference is that the Intempora Validation Suite (IVS) has drastically improved our ability to manage and search the data efficiently and to facilitate access to specific and relevant corner cases for our project by adding bespoke algorithms to the existing labels to identify them better. When we modify the software components of our automated driving system, the data management solution hosted on AWS cloud helps us extract discrepancies in already performed driving situations, allowing us to provide a broader safety analysis.
The symbiotic fusion of dSPACE and AWS has revolutionized the capabilities of our autonomous racing team, AI Racing Tech. dSPACE's cutting-edge computer solution allows us to seamlessly log data and deploy our software stack in real-time, while AWS empowers us to store, analyze, and simulate our race car's performance before deploying it back to the dSPACE computer. This collaborative synergy has enabled us to push the boundaries of innovation in autonomous vehicles and racing.
SIL in the cloud with ARM HPC
Today's E/E architectures consist of different types of ECUs, often described as HPCs (High Performance Computers), Zonal and Edge ECUs. There are several approaches to virtualize Zonal and Edge ECUs that typically build atop standards like AUTOSAR Classic Platform. However, adding HPCs to simulations is still challenging as they make use of more complex Operating Systems (like QNX, Linux, or Android) and Middlewares (like AUTOSAR Adaptive Platform) and bring new complexity when adjusting the build chain for simulation purposes (e.g., switch from ARM to x86).
AWS provides ARM instances in the cloud that can run the software stack of HPCs including infotainment applications. Together with the strong capabilities of VEOS to simulate Zonal and Edge ECUs, we can create simulation systems that contain all types of ECUs of current and upcoming E/E architectures.
To achieve this possibility, we couple VEOS with AWS Graviton instances and execute the whole simulation in the cloud.
High performance Data Replay using AWS Cloud
Testing perception and data fusion is an essential step in validation of Autonomous Vehicles. Undetected failures and malfunctions that occur at early stage of the data processing chain can be easily propagated and have a strong impact on safety and reliability of subsequent driving functions like prediction,motion planning or control. In addition, the perception of AV is mostly based on machine learning algorithms which must be validated by replaying big volumes of real sensor data.
The connection of flexible and scalable dSPACE data ingestion and validation solutions with the AWS Cloud opens up completely new opportunities of high-performance data replay testing for globally operating development teams. Data collected by a fleet can be ingested into AWS Cloud with up to 100 Gbps high speed connectivity and low latency.
Once available in the cloud, huge sensor data sets can be rapidly downloaded to dSPACE replay systems located in data centers across the globe for testing perception and data fusion algorithms of ADAS/AV ECUs under real-world conditions with highest accuracy and correctness. The joint solution reduces ADAS development and testing times, accelerates time-to-market, and simultaneously creates a repeatable ADAS testing framework for DevOps to deploy globally across different locations.
Accelerating Validation – Smart Logging and Targeted Data Logging Campaigns
In this joint showcase, dSPACE and AWS demonstrate smart fleet management that enables targeted data acquisition campaigns for ADAS/AD development and validation.
The solution aims to drastically decrease the turnaround time of a data collection task from definition of requirements until data is made available in the data center, boosting the efficiency of the data-driven development process.
The core hardware component of the solution is dSPACE AUTERA AutoBox, an in-vehicle data logger that provides the high computing power, bandwidth, and storage capacity required for the development of autonomous systems.
This high performance is showcased in the demo by ingesting and processing multiple raw cameras and high bandwidth lidar, where object identification algorithms are running on the onboard GPU to enable scene understanding.
The scene understanding gained in combination with CAN signals is used to trigger the recording process, according to a user’s set criteria or threshold, such as the number of pedestrians or vehicles in the vehicle’s field of view.
The user defines the threshold within IVS, a web-based tool for managing global data lakes and when the specified threshold is exceeded, the AWS FleetWise component running in Intempora RTMaps triggers an upload of the recording into the AWS cloud. After the data is ingested into the cloud, the data can be viewed and manipulated within IVS. The benefit of the system shown is that only the precise data which is relevant for engineering development and testing is stored, ingested, and selected, which saves time and money.