In the webinar series "The Data Sushi Lessons" this was shown by b-plus together with the partners Bertrandt, Deutronic, IBM, Incenda AI and Zukunft Mobility (ZF).
The exquisite components for development tools of autonomous driving and ADAS functionalities were explained to the visitors in presentations and Q&As. Because just like sushi, the individual "ingredients" and the processing of measurement data for test systems have an enormous impact on the quality of the final product.
The speakers presented how raw sensor data is acquired, stored, processed, replayed and simulated into valid test data. They also demonstrated other process steps in raw data management such as precise test vehicle setup and advanced fleet management.
3 Main Takeaways
Data is the central element in advancing autonomous driving. They accompany the development and validation process in various forms (e.g. raw data, metadata) starting with the first bit, still in the sensor itself or from the ADAS control unit. In order to be profitably further processed, reliably high quality and, in particular, temporal synchronization are essential.
Unlike in the past, this further processing already begins in the vehicle. Our lessons follow this data path from the source to the data center memory and beyond.
Our first "ingredient" is therefore the extraction of data from the ECU software and a flexible path directly into the measurement technology by means of the Measurement Data Service.
3 Main Takeaways
The complexity of autonomous systems and the associated sensor technology e.g. Cameras, Radars and Lidars is increasing continuously. Therefore, the complexity and scope of measurement data involved in the development and application of future systems must be managed accordingly. A powerful and reliable measurement technology is a prerequisite for the development and validation of systems and is a critical and safety-relevant part of autonomous mobility. It is also decisive how well this measurement technology is integrated in the overall vehicle system. A powerful energy supply, a safe and user-friendly operating concept (HMI), and reliable cooling of the measurement equipment are key factors for reliable performance, ensuring a high quality result.
b-plus and Bertrandt have developed and built a latest generation technology test carrier together; MAX. In this presentation, Julian Kapitel has been taking the audience through the entire history of the development and assembly of such a technology carrier.
3 Main Takeaways
Aiming to get its customers safely into series production, b-plus automotive shows how metadata in combination with a holistic management approach can increase the efficiency of a test fleet. Beginning with the preparation, the realisation and the evaluation of test drives, a comprehensive best-practice workflow is showcased. It improves the development process qualitatively and quantitatively.
The use of meta information along the development process reduces the workload and speeds up the cycle from harvesting data during test drives and analyzing raw data.
3 Main Takeaways
In addition to the powertrain's electrification, tomorrow's mobility will be defined above all by the self-driving vehicle. Due to their large number of sensors and processing hardware, advanced driver-assistance systems (ADAS) require enormous amounts of electrical energy. A central role for the energy supply of ADA systems in electric vehicles is played by the high-voltage converter, which provides essential data mainly in the test and validation phase of vehicles. This data is used to hedge the vehicle electrical system and the sustainable improvement and optimization of individual components.
The speakers from b-plus, Bertrandt, Deutronic, IBM, Incenda.ai and Zukunft Mobility (ZF) are experts when it comes to pioneering future mobility. The single presentations gave interesting and inspiring insights in the toolchain for ECU development. In total, they showed us a holistic approach on pushing autonomous driving.
3 Main Takeaways
ADAS require extensive validation at various development levels. Hardware-in-the-loop tests of ECUs are an established test method, allowing for assessment of the robustness and performance on the target platform. The radar ECUs developed by ZF are intensively tested at early development stages by re-injection of real radar data in order to achieve maximum functional reliability and quality. High data rates, electrical robustness and exact synchronization are important parameters for the test.
In the presentation, the HIL system developed by Zukunft Mobility GmbH has been presented, in which the B-HIL from b-plus plays a central role for data re-injection.
3 Main Takeaways
Thinking backwards: Testreports of the validation of ADAS/AD sensors and ECUs are “the reason” for huge challenges in the area of Hardware-in-the-loop. Not only for the 24/7 validation in HiL farms it is worthwhile to reprocess the data, but already at the beginning of the development at the developer’s desk. Using recorded data from end-to-end, removes a lot of hurdles by design.
3 Main Takeaways
Advanced driver-assistance systems (ADAS) suites have revolutionized the auto industry, but they have also generated a great deal of complexity, new technologies, and costs. AD providers use artificial intelligence (AI) as one key component. Researchers and developers who can deliver insights faster while managing rapid infrastructure growth will be the industry leaders. Following the process of data acquisition from test drives via ingest to the data centre and processing of big data, many challenges occur. To handle this unstructured data, seamless scaling systems and toolsets are necessary from the sensor to the data centre with storage and computing systems.
This presentation provided a case study of how synchronized data capturing in test drives, big data computing, storage and archiving are integrated in today’s AD development workflow.
3 Main Takeaways
It’s all known that there’s the need to record a huge amount of data for the development of nowadays and future vehicles. Therefore, it’s increasingly important to carefully choose the “relevant” data, in order to efficiently develop and test complex automated driving functions.
Pre-Processing the data with AI methods in parallel to raw data recording can give precious insights to what is being recorded, and therefore increase efficiency. This presentation gave a short overview about AI methods that are in use at b-plus for the purpose of data enrichment.
3 Main Takeaways
The behaviour of Artificial-Intelligence-Based systems highly depends on the quality of the training data. Therefore, all the processes relevant for the data acquisition must face comprehensive quality assurance – data quality already starts during the data collection. In the Data Garage project, Incenda AI teams up with b-plus tackling this challenge. Incenda AI’s intelligent algorithms pre-select data during the collection process and on runtime directly on MAX, the Test Demo Car of b-plus. A perfect basis of a high-quality dataset.
The speakers from b-plus, Bertrandt, Deutronic, IBM, Incenda.ai and Zukunft Mobility (ZF) are experts when it comes to pioneering future mobility. The single presentations give interesting and inspiring insights in the toolchain for ECU development. In total, they show us a holistic approach on pushing autonomous driving. After each talk, you’ll get the chance to meet the speakers and our product experts.
If the Data Sushi Lessons did not satisfy your thirst for knowledge or rather hunger for information, feel free to contact us!
We are looking forward to examine your use case and to work out an individual solution with you.
You are also welcome to contact us for further information material and a personal appointment with our product experts.
