Generation of Multiple Types of Driving Scenarios with Variational Autoencoders for Autonomous Driving

In today's cars, more than 50 electronic control units are used to provide safety and to care about the occupants comfort. The development of advanced driver assistance systems is a key role in the automotive domain. It is essential to validate and verify results and to ensure faultless interoperability of the embedded systems. Not uncommonly, the dimensioning of parameters affects safety aspects and changes need to pass expensive test drives. By increasing the density of integration, future control units will be able to handle more functionality associated with less space requirements. Highly integrated systems will open up new possibilities, but challenges, such as security aspects in the final approval process, need to be faced. In example of advanced driver assistance systems, we introduce a methodology, based on the classical ‘V’-diagram, for the development in an overall virtual environment. By combining rapid prototyping technologies with commercials off-the-shelf, a safe but realistic instrument is given. The concept is shown with the implementation of cruise-control systems and a drowsiness detector.

The development of intelligent transport systems such as driver assistance systems currently faces increasing system complexity and a growing number of requirements, e.g. from safety standards. Thus, requirements engineering activities gain an increasingly important role when it comes to the development of safety-critical transport systems. As an example, the introduction of ISO 26262 in the automotive domain requires the adaptation of existing industrial development processes to comply with the upcoming standard. The proposed methods address the new demand in requirements management for a safer development of driver assistance systems. By means of ontologies, domain requirements such as from ISO 26262 are formalized providing a reference model to support semi-automated requirements discovery. Based on this formalized knowledge and an automatically identified selection criterion such as the system's Automotive Safety Integrity Level (ASIL), the large set of process and product related requirements of ISO 26262 is tailored to a reduced set of applicable steps and methods for a project specific development process. The proposed methods have been implemented in a prototype toolchain. In terms of driver assistance systems, a lane departure warning system is used as an application example to illustrate the proceeding.

This paper deals with the analysis of the closed–loop system consisting of driver, vehicle and the surroundings in critical driving situations. This study was geared towards analysing typical driver errors based upon driving situations registered frequently in accident statistics; this analysis should then be used for the development of assistance systems to optimize the driver's actions. The selected driving manoeuvres were simulated on the Daimler–Benz driving simulator with real drivers in order to reveal typical errors. The results of these tests formed the basis for the development of driver assistance systems. The results of the efficiency and acceptance studies concerning the braking and steering assistance developed by Daimler–Benz are presented here.

This paper describes an embedded architecture designed for the Navigation Aided Intelligent cruise Control (NAICC) project and more generally for the development of driver assistance systems. An original CAN bus based hardware architecture is presented. This approach allows the easy integration of numerous heterogeneous components and the management of a great part of the complexity inherited from this diversity. On the other hand, the corresponding software architecture is proposed. It is based on computer aided control system design and high level object oriented concepts in order to integrate and make all the conception stages of an embedded system easier. An original localisation-based bending light system is presented in order to illustrate the implementation facilities obtained with the proposed approach.

There has been remarkable development of driver assistance systems in recent times, and it is expected that these assistance systems will significantly contribute to reducing traffic accidents caused by human attention errors. To improve the functionality of such assistance systems, several studies have been conducted on computational models to compute dynamic saliency maps for predicting the gaze points of drivers. Against this background, this study introduces a prototype simulator that can produce vehicle-mounted camera videos in various driving situations using 3D computer graphics. This simulator easily enables us to validate the effectiveness of models for computing dynamic saliency maps in various driving situations, even dangerous driving situations that are difficult to capture by actual vehicle-mounted cameras. As an initial validation of this simulator, we reproduced vehicle-mounted camera videos in situations with other cars and pedestrians. Subsequently, we applied the models to compute dynamic saliency maps, which were proposed in our previous work to reproduce vehicle-mounted camera videos. The obtained results proved the usefulness of the proposed simulator. Improvements to our models have also been discussed.

Communication is a crucial process in ensuring safety in road transportation. This paper explores the constructs of communication from a cognitive ergonomics point-of-view in order to explain the need and relevance of communication in ensuring on-road safety. An attempt to establish a communication error model has been made with the help of communication and information processing tasks, and delving into the cognitive aspects of traffic behavior. Direct observation on errors in various communication tasks was used to understand the relevance of the proposed model. From an information ergonomics perspective, workflows of major driving maneuvers are proposed that could be used in the development of driver assistance systems, as well as in driver training, to make novice drivers understand the importance of accurate communication in ensuring safety of self and other road users.

Problem statement: The rapid development of Driver Assistance System (DAS) provides drivers with radically enhanced information and functionality. The nature of the current DAS requires a complex human-machine interaction which is distracting and may increase the risk of road accidents. The interaction between the driver and DAS should aid the driving process without interfering with safety and ease of vehicle operation. Speech based interaction mechanisms employed are not sufficiently robust to deal with the distraction and noise present in the interior environment of the vehicle. Approach: Thus, suitable hybrid earcon/auditory icon design principles for DAS are developed. These interfaces are investigated in driving simulators in-order to test their durability and robustness. Several evaluation parameters will be applied. This will ensure the driving-related information from the DAS was delivered to the driver without affecting the overall driving process. Results: This study produces auditory design principles for information mapping (visual into nonspeech based interaction) and presentation framework was produced. It outlines representation architecture that enables concurrent auditory driving related information to be transmitted from four different sources in the vehicle’s interior environment. It outlines a set of hybrid design principles that integrates auditory icons with earcons to map and present real-time driving related data from a visual to a non-speech auditory interface. Conclusion/Recommendations: The major contribution of this research project takes a genuine approach by considering the entire DAS (safety, navigation and entertainment subsystem). It proposes hybrid representation strategy based on the cognitive availability of the driver and cognitive workload. It was a significant discovery that aid future DAS auditory interaction design.

Driver's visual fixation attention prediction in dynamic scenes using hybrid neural networks

The development of driver assistance systems has shown various innovation waves in the last few decades. While such systems can contribute to increasing driving safety and comfort on highways and rural roads, future systems will be enhanced to include urban traffic. Such trends also face challenges, especially in the design of human-machine-interfaces (HMIs). In the UR:BAN project, one aim was to design user-oriented, integrative HMI concepts of current and future assistance systems by considering the challenges of urban driving. Therefore, the cross-functional “HMI tool kit” was developed comprising a strategy for the systematic derivation of action-oriented HMI concepts. The structure of the HMI tool kit differentiates between cases of applications for safe, comfortable, and energy-efficient driving, which takes requirements of different time horizon and user-actions into account. The HMI tool kit provides insight into 1) how information, warnings, and system interventions should be filtered and prioritized to the driver concerning the current traffic situation and 2) how the preferred driver behaviour can be achieved in the situation by the selection of suitable HMI components (e.g. display in the instrument cluster, warnings or sounds). The aim is to warn the driver adequately in safety-critical situations and allow him to develop a sufficient understanding for continuous system interventions or advanced navigation recommendations. The UR:BAN HMI concepts can contribute to an anticipatory driving style, to mitigate safety-critical situations, and to improve stress-free and low-emission driving in urban traffic.

In the automotive sector, the importance of autonomous driving will continue to increase in the future, in addition to further development in the areas of vehicle safety and driving comfort. In order to be able to develop the systems required for this, the automotive industry is dependent on appropriate sensor technology that is able to reliably detect the vehicle's environment in real time and provide the system with additional relevant information. Radar, Lidar and camera systems are among the environmental sensors used in a modern vehicle. These sensors will play a decisive role in the future development of driver assistance systems and autonomous driving. For the development and testing of radar systems, a suitable test environment is needed to verify the reliable function of radar sensors. In this publication a new method developed at Fraunhofer FHR (Fraunhofer Institute for High Frequency Physics and Radar Techniques), is presented, which was implemented with the ID radar target simulator ATRIUM.

This paper presents the investigations of RF ingress automotive immunity measurement on STP cable in the vehicle up to 6 GHz. Some most important goals, which the automotive industry is following, are autonomous driving (artificial intelligence), new 5G communication standard, Car2-Car and Car-2-X, HD video data transfer, and further development of driver assistance systems in future vehicle projects. The Prerequisite for implementation of the mentioned technologies is the use of a high-speed data transmission network, namely the Multi-Gig Ethernet bus system. The electromagnetic compatibility (EMC) of these networks play a significant role in the design phase, as the reliability and safety of the whole system depend on EMC investigations. One of the essential EMC issues, which should be studied, is the EM noise immunity of the Multi-Gig channel against the surrounding noise. To the author's knowledge, this is the first time that vehicle immunity test of the Ethernet Multi-Gig network channel for frequencies from 150 KHz to 6 GHz is presented. The results and the suggested specification lines can help the semiconductor manufacturers to calculate the possible coupled noise to the channel during the design of the corresponding PHYs. This study presents the result of a worse case situation of the STP cable routing in the vehicle, which is measured in absorber hall.

Research and development of driver assistance systems is currently a very active field. One building block of future systems will be an accurate and reliable positioning, which can be realized by relative measurement, using on-board sensors and maps of the environment. However, a prerequisite will be that such maps can be produced fully automatically.

Radar sensors are already employed in production model vehicles e.g. for adaptive cruise control (ACC) systems. Further development of driver assistance systems has also led to the use of radar sensors in active safety systems (active brake assistance, collision warning, emergency braking, etc). However, the costs of manufacturing such radar-based systems, capable of gathering reliable information from surroundings, for vehicles across the market spectrum or for compact executive cars are still too high. Thus, despite the improved reliability characteristics, detection properties and safety required for these sensors, the aim is to manufacture such systems more cost-effectively. The German national “KRAFAS (Cost-optimized Radar Sensor for Active Driver Assistance Systems)” project is aiming at integrating 77 GHz components (esp. SiGe MMICs) into a printed circuit board, combining driver and 77 GHz RF circuitry and integrating antenna elements. This will significantly reduce current costs of the 77 GHz RF module by 20–30%. A sketch of such a module with an adapted cylindrical radar lens is depicted in Figure 1. In this paper, design, simulation, technological development, demonstrator realization and subsequent measurement of interconnects of embedded active 77 GHz chips to a high frequency substrate using microvia technology is described. The used molded embedding technology offers great opportunities for a very broad range of frequencies and applications as well as large potential for cost reduction.

Naturalistic Driving Studies (NDS) are becoming an integral tool for development of driver assistance systems. Because of its large volume, one challenge with working with NDS data is identifying driving scenarios of interest automatically. This study introduces a methodology for identifying situations where the driver of the instrumented vehicle applied the brakes while following another vehicle. These car following events are of interest for designers of Forward Collision Warning (FCW) systems. This algorithm could be used in conjunction with a large scale NDS, such as the Virginia Tech Transportation Research Institute's 100-Car database, to generate population distributions of braking behavior during car following. These population distributions could be used to inform the design of warning thresholds for FCW. The heuristic algorithm developed in this study identifies car following events using forward looking radar (object range and range rate) and vehicle dynamics (speed, vehicle yaw rate). The proposed algorithm identified the same car following scenario as a visual inspection of the data in 91.8% of brake applications, suggesting it can automatically identify car following events.

With the advent of the information age, the development direction of automobiles has gradually changed, both from the domestic and foreign policy support attitude, or from the actual actions of the automotive industry and scientific research institutes’ continuous efforts, it is not difficult to see that driverless vehicle. At this time, the testing and evaluation of the intelligent behavior of driverless vehicles is particularly important. It is particularly important not only to regulate the intelligent behavior of unmanned vehicles, but also to promote the key It can not only regulate the intelligent behavior of unmanned vehicles, but also promote the improvement of key technologies of unmanned vehicles and the research and development of driver assistance systems. The evaluation of comprehensive obstacle-avoiding behavior for unmanned vehicles is often considered as a multi-attribute group decision making (MAGDM) problem. In this paper, the EDAS method is extended to the interval neutrosophic sets (INSs) setting to deal with MAGDM and the computational steps for all designs are listed. Then, the criteria importance through intercriteria correlation (CRITIC) is defined to obtain the attribute’s weight. Finally, the evaluation of comprehensive obstacle-avoiding behavior for unmanned vehicles is given to demonstrate the interval neutrosophic number EDAS (INN-EDAS) model and some good comparative analysis is done to demonstrate the advantages of INN-EDAS.