The Record - Issue 15: Winter 2019

63 T o sufficiently test and validate automated driving functions, development teams of car manufacturers and automotive suppliers are facing a seemingly impossible task; managing peta- bytes of test data while having to rely on multiple, often unaligned, tools to handle the complete val- idation process of advanced driver assistance sys- tems and automated driving automotive systems. With more sophisticated technology and extended functionality, such systems become increasingly more complex and data intensive as the underlying process becomes correspondingly more challenging and convoluted. Scaling such complex systems is especially problematic. The employees who need to be able to access and use data for developing, testing, and validating range from test drivers and software engineers to data scientists and project managers, who are often located in different parts of the world, making data orchestration an even more challenging task. To manage this complexity, Elektrobit recently launched EB Assist Test Lab, a cloud-based val- idation toolchain that efficiently manages such data, allowing automotive original equipment manufacturers and suppliers to bring automated driving to series production faster. It’s designed to help solve all the logistic challenges engineers within automated driving development face when interacting with big data. Making use of the Microsoft Azure cloud to store and make available the huge amount of data allows for near infinite flexibility and scalability, while also meeting peak demands during parallel execu- tion of tests and ensuring universal accessibility for developers around the world. Elektrobit is working on providing a holistic toolchain for recording, simulating, analysing, labelling, managing and processing data, lev- eraging strong relationships with simulation providers like IPG Automotive or data centre operators like Equinix. Through a user-friendly and intuitive web portal, it’s possible to easily ingest, track, and discover data. Furthermore, this approach includes smart pri- oritisation around hot, cold and frozen data to bring costs down and easily accommodate for the Automotive Data and Time Triggered Framework (ADTF), Robot Operating System (ROS) and other custom data formats, while also providing data quality indication to easily rate driving scenes and their quality. As the platform is designed in an open way, partners and customers can easily add their own, proven in-use tools to the toolchain. EB Assist Test Lab will enable development teams to cooperate and successfully test and val- idate algorithms, functions, and systems, bring- ing automated driving to the road faster and in a more efficient manner. Sebastian Weik is product marketing manager of automated driving at Elektrobit The validation challenge Making huge volumes of test drive data accessible and usable is critical to bringing automated driving to the road S E BA S T I AN WE I K : E L E K T ROB I T V I EWPO I NT  AUTOMOT I V E