Alexander Noack
b-plus technologies GmbH

In the development and validation process of new functions for AD/ADAS applications, OEMs and Tier1s are looking for complete solutions that can be integrated with each other. They are less interested in dealing with detailed questions, but instead want to get information from measurement technology experts on how the customer-optimized setup should be designed in its totality, which components should be chosen with which pros and cons. The goal is to be able to test functions live in the vehicle as well as to generate data sets that can be used for further tests in HiL methods or even for AI training.

From consulting, recording, management of the recorded data, live analyses through to data optimization and storage solutions in the data center - this is the complete system for ADAS/AD validation. Together with our partners, we now offer a wide range of services in the data-driven development environment.

3 Main Takeaways

• Creating the right measurement technology setup for data and test drives
• Target-optimized recording processes incl. data management and scenario detection
• Data security and data integrity

Johannes Zangerle
b-plus technologies GmbH

b-plus knows that the portfolio must be complemented as a package with software solutions, whether with our own components or those of our partners. Johannes' presentation covers the needs of the development department, the efficient use of existing data, defines the term "campaign" and explains which components can be used in Hardware and Software in and outside the vehicle. He also discusses how these large amount of data can be transferred to a cloud environment, for example, and how they can be made available to the test engineers there.

3 Main Takeaways

• Campaign Definition and Execution for efficient Testfleet Management
• Health Monitoring and Remote Management of Components
• Modular Definition of Recorder and High Performance Computer

Peter Kirsch
Solectrix GmbH

Powerful video grabber and playback systems are required for the validation of camera-based ADAS/AD systems. Modern vehicles with AD driving functions according to SAE Level 3 often have 12 or more high-resolution cameras installed, which transmit their raw data to the processing ADAS/AD ECU via protocols such as GMSL, FPD-Link, etc. Enormous video data streams of up to 6 Gbit/s per camera must be captured lossless and timestamped synchronously during test drives and fed to the data logging host system with lowest latency.

What is needed is a powerful, flexible and cost-effective video grabber and playback system that supports all common technologies in the automotive world and can be easily expanded to include future technologies. Uncomplicated system integration and setup as well as intuitive operation are further important requirements for such a system.

This presentation will illustrate how the video grabber and playback system SX proFRAME from Solectrix as part of the b-plus data logging platform DATAlynx, fulfills all these requirements.

3 Main Takeaways

• Modular approach of proFRAME safes costs and time, changeable adapter modules support all important automotive protocols
• proFRAME already pre-integrated in b-plus DATAlynx und Brick logger systems is close to ready2use
• Example configurations for a wide range of cameras (SerDes, sensor) available, perfect starting point for a project specific logger setup

Christoph Michael
Quasara GmbH

Quasara has introduced Plato Intelligence Anywhere, an innovative AI system for autonomous driving data collection at ADAS&AV Tech Expo in Stuttgart. This advanced technology, running on measurement cars, utilises the latest GenerativeAI technologies to enhance smart, real-time data acquisition.

The solution detects an unlimited number of objects, situations, and edge cases directly on the vehicle, triggering recording only where it’s needed. Each car performs specific detection tasks and alerts fleet managers upon capturing significant data points. The system also provides statistical analysis during and after drives, aiding in efficient playbook planning. This technology enables companies to reduce driving distances by at least 50% while collecting equivalent valuable data, improving efficiency when recording and increased cost-effectiveness.

By exposing critical edge cases in real-time, Plato Intelligence Anywhere significantly advances the development and deployment of autonomous vehicles, enhancing the reliability of ADAS and autonomous driving.

3 Main Takeaways

• Smart Recording instead of “always-on” recording
• Detect an unlimited variety of objects, situations and edge cases - no pre-training needed
• Capture the right data on the vehicle in real-time

Adrian Bertl
b-plus technologies GmbH

In today's metrology environment, comprehensive software solutions are essential to ensure a complete and secure logging infrastructure. Software components play a critical role in standardizing data formats, ensuring time synchronization, and implementing data encryption. By overcoming the challenges of data source integration and secure data management, our advanced software solutions provide a reliable framework that improves the quality and usability of recorded data, ultimately driving efficiency and innovation in measurement processes.

3 Main Takeaways

• Time Synchronization: Accurate time synchronization is a critical aspect of data acquisition. With our own XTSS technology, b-plus ensures that all recorded data is correctly timed to guarantee the integrity and accuracy of the analyses.
• Data Encryption: To ensure the security of sensitive data, b-plus uses robust encryption methods. These protect data not only during storage, but also during transmission, especially when using cloud services.
• User Usability: The software solutions need great emphasis on user-friendliness in order to promote the acceptance of the technologies. Intuitive user interfaces and clear processes are crucial to facilitate the handling of complex measurement technologies and to increase efficiency in daily use.

Steffen Metzner
AVL List GmbH

Despite all efforts to carry out tests of vehicle automation functions using simulation technology, a major part of tests and validation is still necessary to be executed with real vehicles. For an efficient development and verification of sensor and perception systems, despite the possibility to log and record high data-rates, highly accurate and reliable ground truth reference sensor devices are necessary. Without such systems, troubleshooting and objective evaluation of vehicle functionalities behaviour become very expensive and often even impossible. The solution discussed in this presentation aims at supporting the development of vehicle automation functionalities during the entire development process. The focus lies on usually underestimated or even missed functionalities like GT-sensor diagnosis and monitoring, time synchronization, calibration (EOL and in-field) and vehicle-setup support. Finally we discuss how online processing can be used to minimize data and prepare the refinement of data evaluation during post-processing.

3 Main Takeaways

• Data collection with or without reference sensors needs direct feedback from the vehicle to the operators of the test-vehicles and to the central office to deliver high value data
• Especially ground-truth sensors have a very much higher failure rate than mass production sensors and need to be monitored continuously to avoid data waste
• The calibration of the whole system is crucial for online- and post-processing and is one of the most underestimated topics in the automotive (test-) industry

Mario Henrique Cruz Torres
IVEX

Automotive OEMs and Tier1’s spend huge amounts of resources testing and validating ADAS functions by collecting real-world driving data during long collection campaigns driving for thousands of kilometers.
Accessing, analyzing and evaluating the data to find specific challenging driving situations requires a big effort, often reserved for highly technical people, which slows the learnings and restricts communication to other colleagues and managers.
IVEX helps automotive companies with the testing and validation of ADAS features through a modular software platform incorporating flexible pipelines, allowing all levels of users to navigate through vast amounts of data collected and automatically generating simulation ready files for specific driving scenarios.
The Software allows to easily pinpoint events through customized metrics and to categorize hours of test data into specific interesting driving scenarios such as Euro NCAP Cut-in.
Further assessment can be performed by navigating, evaluating and easily visualizing the scenarios based on different parameters.

3 Main Takeaways

• Resource-Intensive ADAS Testing: Automotive companies spend significant resources and effort on collecting and analyzing driving data for ADAS validation, requiring highly technical expertise
• Need for Efficient Data Analysis: Most of the collected data is never used. It is really needed to enable easy event identification and scenario categorization of recorded data, as well as enhancing further assessment and visualization from different technical and not-technical levels
• IVEX Software Platform: IVEX offers a modular platform that simplifies data analysis, scenario-based testing and data navigation as well as enhances collaboration between teams

Mario Ehrenberger
SVA System Vertrieb Alexander GmbH

Stefan Jährling
SVA System Vertrieb Alexander GmbH

Tilmann Schröder 
NetApp

In the fast-paced automotive industry, engineering teams often work across different geographical locations on complex and data-intensive tasks such as vehicle design, crash tests and simulations. The challenge is to efficiently manage and process large amounts of data so that teams around the world can collaborate in real time. Our solution combines the benefits of on-premise infrastructures with the scalability and innovation of the cloud, supported by technologies such as NetApp's FlexCache. This enables engineering teams to share data from vehicle tests and simulations in near real-time and work efficiently on vehicle safety tests. b-plus plays a central role in implementing our solution in real automotive environments through advanced in-car storage and ingest stations.

3 Main Takeaways

• Efficient management and processing of large amounts of data in a hybrid multi-cloud environment
• Real-time collaboration and analysis of vehicle data worldwide through innovative technologies
• Integration of b-plus technologies to seamlessly capture and store vehicle data for advanced ML analytics

Frank Kraemer
IBM Germany

Join us to explore the revolutionary IBM Storage Ceph storage platform, a 100% open-source software-defined solution providing distributed block, file, and object storage from a single cluster. Uncover its scalability, flexibility, and data safety features, ensuring services without single points of failure. We will delve into exciting advancements such as Ceph Management, NVMe-over-Fabric, Multi-region bucket replication, and S3-Select. Gain a comprehensive understanding of Ceph's architecture and capabilities, and witness how it breaks boundaries in storage technology, shaping the future of Automotive mass data storage in the on-premises and cloud.

3 Main Takeaways

• Scalability: Ceph is highly scalable and can manage storage of any size, making it a great choice for Automotive mass storage. It’s designed to work on commodity hardware
• Fault Tolerance: Ceph provides excellent fault tolerance through its self-healing capabilities
• Unified Storage System: Ceph offers a unified storage system, meaning it can handle block storage, object storage, and file storage in the same system

Johann Führmann
b-plus automotive GmbH

Gather the optimal insight into the variety of applications in the field of data replay and realtime injection of simulated data. b-plus offers useful information beginning with the differences of closed and open loop setups, the challenges of simulation based realtime injection and how to migrate HiL systems to the cloud infrastructure.

3 Main Takeaways

• Open and closed loop coexist
• Time synchronization is the magic sauce for multichannel simulation based HiL systems
• How to scale generic HiL Systems up for data center usage

Matt Daley
rFpro

The development of ADAS and AV systems no longer needs to rely on real-world data collection alone... Synthetic data is now able to complement and expand the range of safety critical edge case datasets that are used for tuning sensor systems, training perception systems and testing full stack ADAS and AV solutions.

The combination of the industry's leading driving simulation solution from rFpro and data handling solutions from b-plus - combine to form solutions that are able to rapidly accelerate the way that ADAS and AV systems are developed and delivered. Join us to explore how synthetic data delivers both high fidelity open loop data playback and closed loop real-time testing solutions that are changing the way that the next generation of vehicles are being developed.

3 Main Takeaways

• Synthetic data can used for both open loop & closed loop hardware in the loop testing
• rFpro & b-plus have integrated solutions for hardware in the loop testing
• rFpro & b-plus both have industry leading products for this application

David Stengel
b-plus automotive GmbH

The quality of data is crucial in the development of robust and reliable autonomous driving systems. This presentation delves into the essential role of Data Quality Assurance throughout the lifecycle of autonomous driving datasets, covering processes from specification writing to data annotation and verification. We will address the challenges of managing multi-sensor data and the various types of errors commonly found in training datasets. By providing detailed insights into dataset quality, our CONIX.dpc solution enhances the dataset's utility for training advanced models, reduces risks and enables long-term savings in time and resources.

3 Main Takeaways

• Explore the significant impact of data quality on the development of autonomous driving systems, focusing on key Data QA processes including specification writing, data annotation and data quality checks.
• Examine the complexities of managing and ensuring the quality of multi-sensor data, and how CONIX.dpc identifies and addresses errors across various data modalities, ensuring the reliability of integrated datasets.
• Understand how thorough Data QA processes lead to long-term benefits, including reduced development challenges, cost savings, and a better basis for verification of autonomous driving systems.

Tim Stahl
Driveblocks GmbH

Sensor fusion in autonomous driving combines data from multiple sensors to provide a stable and comprehensive model of the environment. While individual sensors have unique perspectives that may include occlusions, misdetections, or accessibility issues, a representation that fuses data from multiple sensors over time can effectively mitigate these individual shortcomings. This presentation delves into the underlying principles and real-world application of geometrically interpretable stochastic filtering in object fusion and real-time driving corridor generation. The advantages of this technology lie in the straightforward addition and removal of sensors from different domains and the enhanced human interpretability. Unlike encapsulated end-to-end deep learning technologies, this interpretability facilitates the validation and certification processes. Results are showcased across various domains, both on and off-road, highlighting the robustness and versatility of this approach in diverse driving environments.

3 Main Takeaways

• Improved Environmental Modeling: Understand how fusing data from multiple sensors over time creates a stable and comprehensive model of the environment, overcoming issues like occlusions, misdetections, and accessibility limitations of individual sensors
• Versatile and Scalable Technology: Learn about the advantages of geometrically interpretable stochastic filtering, including the easy addition and removal of sensors from different domains and its applicability in both on-road and off-road scenarios
• Geometrically Interpretable: Discover how the human interpretability of this technology, as opposed to encapsulated end-to-end deep learning methods, paves the way for validation and certifications