Automatisiertes Fahren 2022

Autonomous Driving (AD) is expected to heavily change the landscape of future mobility solutions, with machines carrying out the driving task to larger and larger extents. A wide range of applications for this disruptive technology exist, such as driving in urban environments or in highway scenarios. Within this challenging field, Continental is developing driverless solutions, relying on a vehicle fleet covering various AD applications. In this paper, we present our approach of a “Common Core” software architecture for AD systems, allowing to effectively share AD key elements between different AD platforms. This includes the description of applied design principles, resulting advantages, limitations as well as impressions of real-life demonstrations and future development directions.

We are developing a visual SLAM approach for highly automated vehicles, mainly using monocular cameras, which makes it independent of other sensors and allows for redundant solutions. Furthermore, it simplifies extrinsic calibration and time synchronization issues. Therefore, we target a single-shot and real-time capable algorithm, which is deterministic, robust to environment structures, and complemented by confidence levels. Our cost-effective solution generates a 3D map of the environment, in which the vehicle can precisely determine its position. In the future, this can be used to supplement LiDAR systems in highly automated vehicles or even replace them in parts. In this paper, we present our recent progress in developing a robust single-shot monocular SLAM system.

Solving a nonlinear optimization problem to plan an optimal trajectory is a complex task. Moreover, no convergence guarantee can be given. On the other hand, nonlinear vehicle dynamics and nonconvex environmental conditions can be considered. This paper proposes an optimization problem for lane change trajectory planning. To reduce the computational complexity, the output of the solution algorithm is approximated by an explicit function. This is applied to a kinematic model to reproduce the optimal trajectories. The interpretability of the learned function is improved by choosing a grid-based model. To fully exploit the parameters of the model, a global optimization algorithm finds an optimal input partitioning. The performance of the approach is compared to that of a grid with regular partitioning of the input space and a decision tree on a set of optimal lane change trajectories. Furthermore, the output of an optimal grid model is visualized and analyzed for plausibility.

We present a novel approach for using industrial grade ultrasonic sensors to perform echolocation by detecting ultrasonic echoes in a noisy environment using machine learning. Autonomous driving is expected to become a huge market and among other technical challenges, environmental perception will be the most critical one. For high automation level, classical technologies are limited. On the other hand automotive is cost sensitive. The main part of the lecture starts with state of the art technology and then explains how we have used machine learning approaches to train a net for several classifications tasks: Distinguish whether the ultrasonic echo comes from the sensor or another noise source, distinguish whether the echo is relevant or not and finally a height classification. Results are presented in the form of F1-Score. In addition to this, a method will be presented to use CNN for noise suppression in real time. We demonstrate the potential of using the “bat principle” for perception and prove that by that we also achieve the low cost targets.

New mobility services such as ride-pooling depend on new vehicle concepts in order to make their service available to as many potential customers as possible and to be economically profitable at the same time. One possible approach would be purpose-built vehicles. With their purpose-built designs, significant cost reductions would be possible, but most importantly, the requirements and needs of service operators and users could be met. When a new type of vehicle is designed, many questions arise initially, including what existing knowledge can be built upon and how the market potential of the new development can be assessed. This is where the study starts and tries to provide answers. A survey is designed to record the existing expert knowledge based on comprehensive literature research. After all, there are not yet many publications on the topic, and much information tends to be disseminated internally. In the course of a current market review and determination of demand, the sales potential of such a vehicle class is explained. For this purpose, current fleet data, mileage, and expert opinions are obtained. Finally, the paper shows the current status quo of purpose-built vehicles and points out the next necessary development steps.

We motivate the development of a shadow testing mode for the ACC function and realize a recurrent neural network that recognizes certain standard scenarios and predicts a critical event in each of these scenarios. This is a first step in setting up the entire AI-based shadow system which, in addition to the perception and prediction stage for edgy scenarios, comprises an automated comparison between simulated predictions of the actuators for a small temporal neighborhood of the critical event in actual driving and in shadow mode.

In Germany and the European Union, the legal framework for automated lane keeping systems (ALKS) is now in place. They are the first Level 3 systems to be ready for approval and no longer require monitoring by the driver on highways. Since such systems require an enormous amount of testing to ensure safety, simulation is increasingly being used.

This requires a simulation platform which contains the vehicle including integrated driving functions and sensor technology, dynamic objects, the static environment as well as environmental conditions. The 3D environment must be available in equal detail. The basis to set up the 3D environment is a mobile survey of the relevant road section using measurement vehicles equipped with high-resolution scanners and cameras. The result of the surveying as well as the processing of the measurement data is an HD map in Road5 or OpenDRIVE format, optionally including a high-resolution image of the road surface in OpenCRG format. The HD map enables the automatic generation and visualization of the corresponding road model in the simulation environment.

Based on this, the rest of the static environment can be built. After the model of the static environment is completed, simulation scenarios can be created by integrating the dynamic objects plus their behavior. Objects from various sources are suitable for this purpose. In addition, there is the possibility to increase the size of the test space under investigation by varying the behavior of the dynamic objects, the environment, etc.

It is accepted that ADAS/AD validation mainly relies on virtual tests, but the core problem is that the number of tests explodes when all scenarios need to be tested in all possible variants.

We present an approach for vehicle verification and validation that allows to quantify the degree of safety for a vehicle and optimizes the overall test execution process to maintain cost-effectiveness.

Our solution starts with the declaration of scenarios based on the Operational Design Domain (ODD) [1] [2]. The tests are executed in a scalable environment and yield reliable statements about the coverage of the ODD and the safety degree of the vehicle. We identify ways to reduce the number of tests required without compromising the validity of the overall results. Furthermore, we will present how the tests can be executed in a hybrid approach to maximize the cost-effectiveness.

As in many other industries, the automotive sector is facing challenges related to the development, validation, and homologation of the growing number of digital and networked products. Previously based on standards and the classic V-model, software-based automated systems now require data-driven as well as iterative and agile development methods. More than ever before, selling a traditionally discrete product also involves rendering regular services for the operation, maintenance, and continued development of vehicle systems that have already been delivered.

Automated driving functions require a validation effort that can hardly be handled effectively anymore by means of traditional methods. Moreover, it is desirable to provide software updates over-the-air (OTA) during a vehicle’s lifecycle. This drastically shortens software development cycles and obviously results in trends toward agile software development processes using continuous integration, testing and deployment (CI/CT/CD).

By means of scenario-based testing, virtual test driving offers the opportunity to test the driving functions of automated vehicles and their OTA updates in a wide variety of relevant and critical situations and to generate findings for the development and validation of functions within a short period of time. So-called functional decomposition assists in this area as well. It enables isolated investigations of individual modules in testing and validation contexts, which makes it possible to reduce the relevant parameters in the scenario description. Virtual prototypes – i.e., digital twins of real-world vehicle prototypes – provide the basis for firmly anchoring these modern test methods in the development process and assist in the area of OTA updates. For this purpose, the system under test is integrated into a virtual prototype and tested in realistic scenarios as a total system – consistently in each phase of the development process.

A combination of agile software development and virtual vehicle development can decisively contribute to successfully meeting the demands made on the validation of automated driving functions and to providing tested OTA updates within a short period of time.

While the market penetration of partially automated vehicles is steadily progressing, the gain in safety is not yet explicitly assessed and the risk of misuse remains a major concern. Driver monitoring systems are used to improve engagement but these are reactionary to driver behavior. Proactive ADAS functions, which relies on elaborated human-machine interaction, are introduced to encourage driving collaboration and reduce the risk of disengagement. Admittance control scheme for haptic shared control is proposed as an evolution from the blended control scheme as it enables the implementation of the rules of engagement without performance tradeoffs and compliance with the regulation at minimum safety oncost. Furthermore, admittance control contributes to long term development goal as it is compatible with all levels of automation.

Highly and fully automated vehicles allow a multitude of new interior ideas and interactions between humans and vehicles. In this paper, relevant aspects that are affected by the introduction of such high levels of automated driving are identified and the implications on different human-machine-interfaces against the background of four different vehicle concepts will be discussed. The paper is structured according to a disintegrated version of the driver-vehicle-control- loop, which in its original version can be used to analyse driver-vehicle interaction for manual and assisted driving. Based upon that, the paper presents potential activities as requested by users, as well as required functions for exploiting the full potential of AVs. As discussed in the paper, one of the major impact on interior and interaction design could result from alternative seating concepts. These would allow an improved user experience on the one hand, but would incorporate various challenges on the other hand.

The topic of automated driving has been finding its way from scientific and industrial research to specific (pre-)development projects in the vehicle industry for around ten years. Whereas the academic dialogue as well as in the industrial implementation focus on situation recognition and automated driving decisions, the driver task as the monitor of the technical condition of the vehicle and the road has to be automized additionally. Even though all relevant vehicle systems have suitable monitoring functions, several central monitoring tasks remain with the driver in conventional vehicles. This paper covers several monitoring tasks, that are usually covered by the driver and provides solutions to ensure safe operation of a fully automated vehicle.

This paper approaches the development of Software-Defined Vehicles from the angle of efficient software development. We analyze the requirements for Software-Defined Vehicles, analyze state-of-the-art and discuss current challenges and illustrate these using the automotive development V-cycle as an example. We propose a common cross-domain as SDK as the solution to improve the efficiency of software development and as the enabler for a common mobility digital ecosystem.

Testing and validation is currently the limiting factor in the deployment of automated driving. This bottleneck can only be overcome with simulation that allows to significantly decrease the required test miles in the real world. However, results from simulators can only be trusted if they reflect the full complexity and diversity of reality. It is therefore indispensable to bring real- world scenarios into simulation to bridge the gap between development and deployment much faster. For this purpose, this paper leverages the unique advantages of aerial data to reconstruct traffic scenarios in a highly flexible and cost-effective way. A novel deep learning-based 3D computer vision approach is introduced that overcomes the severe limitations of existing 2D methods in terms of accuracy and quality of the reconstruction. The robustness and generalizability of this 3D approach is demonstrated in four traffic datasets that were collected in highly interactive urban and suburban environments. All four datasets are several hours long and contain a map of the road network. Real-world scenarios are extracted from these datasets for testing and validation of automated vehicles. Test results provide insights about which environments are most challenging and where coverage needs to be increased. This scenariobased validation process allows to significantly reduce the number of test miles in the real world enabling the deployment of automated driving at scale.

The vehicle’s sensor and camera systems are exposed to various environmental influences. However, even under challenging conditions, fault-free functionality must be guaranteed at all times. When driving in the rain, water accumulations and dirt particles contained therein can lead to reduced visibility of the camera systems. In this context, IAV is developing a simulation model to reproduce the physical processes occurring during such weather scenarios. Thus, it is possible to predict functional limitations at an early stage of the development process and to take appropriate precautions.

The digitalization of the automotive industry presents companies with legal challenges. These are typically to be found in the areas of cybersecurity and data protection. In addition, against the background of data-driven business models, the economic exploitation of data and the associated legal regulations on access to and use of data are becoming increasingly important. The article presents the central legal requirements of the above-mentioned areas, which can arise both from laws and from contract law, and their future development. In addition to the content-related requirements, the focus is also on the question of who the addressee of the individual requirements is and in which constellations they must be taken into account. Following on from this, the practical challenges in implementing the legal requirements by companies are presented and a compliance management system for the strategic implementation of legal requirements in the context of digitalization is proposed.

The city of Hanau is located in the Main-Kinzig district and is part of the FrankfurtRhineMain regional association. In 2021, the city of Hanau will have exceeded 100,000 inhabitants and will be now Hessens "smallest" large city. This is another reason why the city of Hanau, in cooperation with the Main-Kinzig district, is striving to become a district-free city in the medium term.

The development of (inner-city) transport of the future is a very important and challenging task for all cities, which cannot be solved alone, but only in cooperation with the respective region.

In March 2019, the city of Hanau discussed various aspects of future urban life with over 7,000 visitors at a citizens' participation weekend. One aspect was mobility. The basic message, as it was often presented at this weekend, was: "Green instead of grey". In other words, the transformation of traffic areas into green spaces: More plants, less asphalt, more quality of life and stay.

In the summer of 2020, the city councillors decided to draw up the mobility model as the first module for future transport development.

The process for the "Vision" hold up almost two years and is the results of the cooperation of experts from research and science, urban and mobility planning, municipal transport development and administration (Büro für Mobilitätslösungen, Technische Hochschule Nürnberg, ADFC, Frankfurt University of Applied Sciences, Hanauer Parkhaus GmbH, Lokale Nahverkehrsorganisation, Hanauer Straßenbahn GmbH, KVG Main-Kinzig mbH, Benz+Walter, dmo - digital mobilities consultants, StetePlanung, Hanau Infrastruktur Service, Stadtwerke Hanau, Standortmarketing, Straßenverkehrsbehörde, Nachhaltige Strategien, Stadtplanung), the political bodies and, last but not least, the citizens of Hanau.

On 13.12.2021, the Mobility Model Hanau - Towards Future was decided unanimously.

In the following paper the reader gets an overview how Hanau reaches the goals of Mobility Model.

Automated vehicles are leading to a significant revolution in road transportation systems, with extensive challenges to mobility through several factors, including economic, legal, social, psychological, and technological aspects. Various countries around Europe have tested and piloted autonomous vehicles on their roads and subsequently published research on their approach towards reaching such goals and also on findings from their various projects. However smaller peripheral countries tend to fall behind, whenever innovative technologies are being researched and tested. Research on Malta’s level of preparedness for the introduction of autonomous vehicles is limited. This research project deals with an analysis of the expected impacts of shared autonomous vehicles on the physical and digital road infrastructures and assesses the current infrastructure preparedness on a local scale. The project aims to provide Malta with a new mobility solution, that is sustainable and technologically advanced, that will at the same time increase the users’ choices of alternative transport. Through various meetings and consultations with local stakeholders, it emerges that Malta is still in the very initial planning stages for the introduction of Autonomous Vehicles and that an immediate intervention on Malta’s current physical and digital infrastructure is required to enable the actual implementation of shared autonomous mobility field testing. The outcomes of this research project include a roadmap and recommendations to local authorities, which shall pave the way for the first shared autonomous shuttle pilot project to take place on Maltese roads.

The trend in automotive electronics goes clearly towards service-oriented E&E-Architectures. Classical or flat partitioning using huge number of ECUs is past. Autonomous and Electric Vehicles need different approaches that support both high bandwidth in-vehicle communication, but also wireless connectivity to remote Edge and Cloud services. 5G is a big enabler, but not the only one. It’s also important how the E&E-Architectures within the vehicles evolve. However, is this all really an evolution? Or is it maybe more a revolution that is ahead of us?

Service oriented solutions for all kinds of connectivity need to support safe operation of the car as well as innovative business models for OEMs and 3rd parties. The combination of highest reliability and highest flexibility is a real stretch goal for all stakeholders. How can this be achieved? What is the role of Connectivity, beside of classical Telematics?

This session builds a bridge from user requirements over business opportunities to technology outlooks of future Connected Autonomous / Electric Cars.