Globally distributed development teams require one common data platform to work in
Elly Yates-Roberts |
Modern vehicles are increasingly reliant on complex software configurations, from today’s advanced driver-assistance systems (ADAS) to tomorrow’s autonomous driving platforms. To build these complex systems, unparalleled volumes of data need to be collected, indexed and made available to globally distributed teams.
Data needed for the testing and validation of advanced automated driving features is captured by a suite of sensors normally comprised of cameras, lidar, radar and occasionally thermal imaging. Today, the average daily output from a test fleet vehicle collecting this data ranges from 10-20 terabytes. With next-generation sensors and 4k cameras, we can expect this to increase by up to 10 times. This means making petabytes or even exabytes of data available to engineers and data scientists. Furthermore, it is making data in these large databases easily searchable and usable that is the critical task.
This is where the EB Assist Test Lab data management platform steps in to streamline workflows and annotate collected data with meta data sourced through several different functions. Once this data is sufficiently indexed, the Test Lab seeks to improve the data socialisation mechanisms of a team. The term ‘data orchestration’ generally refers to the automation of data-driven processes. This includes the preparation of data, making decisions based on that data and taking actions based on those decisions. It is commonly referred to as an end-to-end process that most often spans across multiple departments and borders. Being able to efficiently develop, test, and validate ADAS all the way to autonomous driving is no exception. In fact, team sizes of such areas can be in the hundreds, involved in cross-department, cross-company or cross-continent projects.
To overcome this challenge, Elektrobit has teamed up with Microsoft and its Azure cloud computing platform. The EB Assist Test Lab enables automotive companies to simplify their team workflows and interactions with a tooling that understands how to manage and socialise a big data lake. Plus, by running partly or entirely in Azure, we ensure universal accessibility for developers around the world. The key value lies in data socialisation; the engineer easily defining data needs in a campaign format to the data manager receiving this task and virtually ordering a test drive to be conducted by the test driver. Every developer in your company is empowered to find the data they need, quickly. Furthermore, data scientists can find data with complex requirements in an efficient manner, enabling everyone to easily upload data and ensure that the data is stored and can be used. The EB Assist Test Lab streamlines the process, getting these teams back to writing algorithms instead of wasting time searching for the right data.
Dylan Dawson is business development manager at Elektrobit.
Read more about Elektrobit’s approach to the challenges of validating automated driving.
This article was originally published in the Spring 2020 issue of The Record. Subscribe for FREE here to get the next issues delivered directly to your inbox.