Author: Florian Netter, Audi Electronics Venture GmbH<\/p>\n
Piloted driving \u2013 a technological and functional challenge. This presents current tasks for software development in the areas of embedded development, mobile computing, and cloud-based services, allowing for targeted solutions. Supported by a virtual development framework and adaptive processes, this offers the potential to optimally support software projects and thus overcome the challenges of piloted driving.<\/strong><\/p>\n In recent years, many innovations, including in the field of driver assistance, have been conceived and developed to series production readiness. One example of this is the\u00a0Audi<\/em>\u00a0active lane assist<\/em>When the vehicle leaves its lane, the driver is not only warned, but also actively supported in staying in the lane within the system's limitations. This is made possible by a steadily increasing number of electrical\/electronic systems and software in the automobile. The logical next step leads to piloted driving (and parking), which is already one of the key technologies of the future. At the beginning of the year, AUDI AG launched the\u00a0Audi A7 Sportback piloted driving concept<\/em>\u00a0In the USA, a long-distance test from Stanford to Las Vegas, covering 550 miles, was successfully completed. These new concepts present both a functional and a technological challenge. This results in new challenges for automotive software development, which must also respond to current trends in electrical\/electronic development.<\/p>\n The task of Audi Electronics Venture GmbH, as a wholly owned subsidiary of Audi, is to support the company as a competent partner in overcoming the upcoming challenges.<\/p>\n The current landscape of driver assistance functions is very heterogeneous, distributed across a wide variety of domains, such as parking.\u00a0Parking assist<\/em>\u00a0or in the area of lighting\u00a0High-Beam Assist<\/em>. While information is exchanged between these systems, they are not highly networked and only utilize shared basic components, such as central sensor data fusion, to a very limited extent. To implement novel approaches like a traffic jam assistant, consolidation at the functional level is necessary, and the synergies between all applications must be leveraged. This represents the first step on a development roadmap; in a second stage, assistive applications can be further developed into piloted applications. This milestone, in turn, forms the basis for all further developments.\u00a0Piloted Lane Changing<\/em>\u00a0or\u00a0Piloted City Driving<\/em>\u00a0\u2013 which then gradually find their way into the series.<\/p>\n To implement these concepts, the vehicle requires a complete 360-degree view of its immediate and surrounding environment. Various sensors with different detection ranges are used for this purpose, as shown in Figure 1 (PDF<\/a>The surrounding area must also be recorded redundantly in order to perform plausibility checks, detect disturbances or malfunctions, and, if necessary, determine sensor replacement values.<\/p>\n All measured values are processed in the central sensor data fusion, which calculates a complete model of the environment. Additional information can be collected and processed via Car2x. This model is then used in trajectory planning to design the roadway based on the surrounding environment. Another software component for\u00a0decision-making<\/em>\u00a0It calculates the internal states and activates the corresponding software modules, e.g., the calculation of a new trajectory.\u00a0Actuator control<\/em>\u00a0It now controls the transmission, the electronic accelerator pedal and the brakes, and specifies the calculated steering angle.<\/p>\n For near-series development, the so-called zFAS (central driver assistance control unit) now exists with a very powerful hardware architecture the size of a laptop, see Figure 2 (PDF<\/a>), while as recently as 2012, all the technology was installed in the entire trunk of a prototype.<\/p>\n The aforementioned functional consolidation has a significant impact on the two topics of E\/E architecture and highly integrated control units, with Ethernet serving as the broadband data backbone. Networking is no longer limited to the vehicle itself, but has also developed considerably outside the automobile in recent years, as vehicles increasingly connect and interact with their environment. Car2X communication is one example, but the use of aggregated fleet data via cloud-based services\u2014the key term here being Big Data\u2014should also be mentioned. Connectivity with mobile devices will also play a crucial role, for example, in transmitting the trigger for the piloted parking maneuver to the vehicle. Embedded in a virtual development framework, algorithms, architectures, control units, hardware prototypes, etc., can be tested and verified very early on. Furthermore, all these aspects must be mapped into processes. In summary, the challenges can be extracted as follows:<\/p>\n These points will now be briefly discussed in the following subsections.<\/p>\n The complexity of in-vehicle networking has increased significantly over the last 15 years. This is due to numerous innovations enabled by electrical\/electronic systems and software, which has led to a massive increase in the number of Electronic Control Units (ECUs) in vehicles. Traditionally, these ECUs are interconnected via a heterogeneous bus landscape (CAN, FlexRay, LIN, etc.), with a gateway mediating between the various subnetworks with differing bus physics and configurations.<\/p>\n The trend toward connectivity will continue in the future, with a massive increase in the number of distributed functions and the increasing integration of vehicles with their environment. This also increases the number of external interfaces to the vehicle, which must be considered and secured during the development process from a safety and security perspective. Coupled with increasing bandwidth requirements, this results in entirely new demands that a future-oriented connectivity architecture must address to enable piloted driving functions.<\/p>\n Currently, there is a great deal of activity in the field of Automotive Ethernet, as promising approaches and concepts can be implemented here. This involves a switched network with switched point-to-point connections, as opposed to previous bus systems with a shared transmission medium and, for example, CSMA\/CD access methods (Carrier Sense Multiple Access\/Collision Detection). Figure 3 (see\u00a0PDF<\/a>Figure 1 shows a schematic network architecture as an example with four domain control units. These are connected via an Ethernet backbone; a star topology, for example, could be implemented. The subnets below the domain ECUs can also be implemented using Ethernet, but bus systems such as CAN or LIN will still be relevant here as well.<\/p>\n It is important to mention at this point that the familiar Ethernet from the consumer world cannot be directly transferred to the automotive domain. Many aspects based on automotive requirements must be taken into account, for example, regarding Quality of Service (QoS).<\/p>\n To optimally consider, develop, and utilize all these aspects, central system architectures are required. These must encompass not only the vehicle itself but also innovations from other domains: Car2X, mobile devices, and cloud-based services\/big data. Furthermore, safety, security, and privacy aspects must be designed and implemented during development using the "by-design" methodology. Figure 4 (see\u00a0PDF<\/a>) illustrates this relationship graphically using the example of a communication chain.<\/p>\n The interfaces between the four defined domains in a system architecture are implemented via corresponding platforms and their integrated interface adapters. This ensures the use of defined and standardized interfaces, allowing for seamless integration of subsequent development steps. Furthermore, this approach increases the reusability of software components, thereby reducing development costs. Three platforms from the respective domains are listed as concrete examples:<\/p>\n Virtual development methods, alongside platforms and system architecture, are another important component in automotive software development. They allow vehicle functions to be verified and validated early on in a wide variety of situations and scenarios.<\/p>\n Audi Electronics Venture GmbH uses the so-called\u00a0Functional Engineering Platform<\/em>\u00a0(FEP) developed. The requirement is to be able to integrate not only embedded software modules into a virtual simulation platform, but also components from the entire system architecture, such as those from mobile devices and cloud-based services\/Car2X. Connecting to real ECUs, HIL test benches, virtual ECUs, or combinations thereof results in diverse setup options that can be precisely tailored to the test requirements. The integration of sensor models, such as radar, and the coupling with driver models and traffic simulation provide flexible deployment options during the development phases. A possible test setup is shown in Figure 5 (see\u00a0PDF<\/a>) dar.<\/p>\n With a unified system architecture, the use of platforms, and virtual development, all the technical prerequisites are met to generate software-intensive innovations. However, these three aspects must be used effectively to successfully develop the innovations and use them sustainably within the company.<\/p>\n Audi Electronics Venture GmbH therefore focuses on embedding these aspects in adaptive processes. For example,.\u00a0EnProVe<\/em>\u00a0A project for organizing the continuous improvement process with the goal of providing a unified, modular development framework for embedded software development. It also includes common tools, suitable development methods, high-performance infrastructure, and provides support services for developers. This EnProVe process is certified according to Automotive SPICE\u00ae Level 3 and can be adapted to the specific requirements of a given project.<\/p>\n To overcome the challenges of software development in the field of piloted driving, it is essential to develop and implement common system architectures from the outset. Early adoption of platforms allows for the use of standardized interfaces, enabling the rapid and seamless implementation of subsequent development steps. Embedded in a virtual development framework, software projects can be reviewed and tested very early on. To map all of this into processes, adaptive processes are required to address all aspects of embedded development, mobile computing, and big data.<\/p>\n This approach results in a close integration of vehicle functions with the corresponding software technology. Furthermore, the continuous and consistent application of these aspects from the concept phase to maintenance represents the decisive success factor for Audi Electronics Venture GmbH, see Figure 6 (PDF<\/a>Consequently, it is possible to adopt artifacts from previous process steps in order to shorten development time, reduce costs and optimally support development projects.<\/p>\n Download the article as a PDF<\/strong><\/a><\/p>\n Do you want to bring yourself up to date with the latest technology?<\/strong><\/p>\n Then find out more\u00a0here<\/strong>\u00a0<\/a>MircoConsult offers training courses\/seminars\/workshops and individual coaching on the topics of process, project and product management.<\/p>\n Training & coaching on the other topics in our portfolio can be found here.\u00a0here<\/a>.<\/strong><\/p>\n Valuable expertise in process, project and product management is available.\u00a0here<\/strong><\/a>\u00a0Available for you to download free of charge.<\/p>\nIntroduction<\/h2>\n
Introduction of piloted driving<\/h2>\n
Challenge for software development<\/h2>\n
\n
Future-oriented E\/E architectures and Ethernet<\/h3>\n
Platforms<\/h3>\n
\n
Virtual development methods<\/h3>\n
Development processes<\/h3>\n
Summary<\/h2>\n
\nProcess, product and project management \u2013 our training & coaching<\/h2>\n
\nProcess, product and project management \u2013 expertise<\/h2>\n