Advanced Microsystems for Automotive Applications 2016

The objective of this paper is to propose a smartphone based outdoor dead reckoning localization solution, show its experimental performance and classify this performance into the context of Vehicle-to-X (V2X) based pedestrian protection systems for vehicle safety applications. The proposed approach estimates the position, velocity and orientation with inertial measurement unit (IMU) sensors, a global navigation satellite system (GNSS) receiver and an air pressure sensor without any restriction to pedestrians, like step length models or a relationship between smartphone orientation and walking direction. Thus, an application makes sense in handbags, trouser pockets or school bags. The dead reckoning localization filter was evaluated in measurements and compared with a highly accurate reference system. Furthermore, the requirements of measurement and modelling uncertainties in a pedestrian protection system with a low false-positive rate were derived and compared with the reference measurements. The results demonstrated that an appropriate use of the proposed system approach is only possible with more accurate positioning solutions from the GNSS receiver. According to this the necessity of differential GNSS methods was predicted.

The NAVENTIK project will develop an automotive platform for computational demanding applications in the field of sensor data fusion and software defined radio. Based on this platform, the first component launched will be an automotive-grade GNSS (Global Navigation Satellite System) receiver that integrates state-of-the-art signal processing for lane level accurate navigation and that guarantees bounded false alarm rates. This is possible, thanks to a software-defined approach and the probabilistic integration of GNSS signal tracking algorithms on radio level. The explicit modelling of GNSS error sources and local signal degradation provide the basis for the proper Bayesian integration. The project will enable the first mass-market GNSS receiver based on a software-defined approach that is able to meet safety-critical requirements as it copes with false alarm specifications and safety related requirements.

Years after publication of the IEEE 802.11p and IEEE 1609 standards for Intelligent Transportation Systems (ITSs), first production vehicles equipped with conforming communication hardware are about to become available. The standard’s suitability for the hard real-time automotive environment has been debated intensively in recent years. Most publications use synthetic message sizes, while the comprehensive ITS-G5 standard allows for a performance evaluation of IEEE 802.11p in real scenarios. Realistic performance and scalability assessments of current automotive communication hardware can be derived from such an evaluation. Based on these we examine the suitability of the available standards for demanding hard real-time control tasks as cooperative adaptive cruise control (CACC).

Innovation by semiconductor manufacturers is a key requirement for future cars. In communication architectures of today’s high-end cars we see more than one hundred control units, a trend that continues to increase. When we introduce Automated Driving, communications will also have to pass the boundary of the car. A closer look unveils the need for steadily increasing bandwidth, higher fault tolerance, and real-time performance. Data security is an urgent issue, given the prominent car hacks in the US. System costs also come into the equation, as not all solutions are affordable. The following paper shows the state-of-the art car communications and roadmaps its expected evolution showing Vehicle-to-Everything (V2X) as the pre-requisite technology for Automated Driving. It also outlines an approach to categorize the levels of data security in the car and ends with an overview of requirements for secured car communication systems.

What once started as Advanced Driver Assistance Systems (ADAS) has evolved into vehicle control systems that partly or completely take over the driver’s task. In doing this, many assumptions are made on the design of the infrastructure that the car will have to deal with. Infrastructure, users and cars should not be looked at separately but in combination. Road operators are faced with these new developments, a larger variety of cars, different user behaviour and are restricted in budget. The areas where infrastructure providers and the car industry should work together more closely are explored in this paper.

In this paper we describe the concept of the data fusion and system architecture to be implemented in the collaborative research project Smart Adaptive Data Aggregation (SADA). The objective of SADA is to develop technologies that enable linking data from distributed mobile on-board sensors (on vehicles) with data from previously unknown stationary (e.g., infrastructure) or mobile sensors (e.g., other vehicles, smart devices). Data not only can be processed locally in the car, but also can be collected in a central backend, to allow machine learning based inference of additional information (enabling so-called crowd sensing). Ideally, crowd sensing might provide virtual sensors that could be used in the SADA fusion process.

This paper introduces a novel system for facial landmark detection using a modified Active Appearance Model (AAM). Traditional AAMs operate directly on the pixel values of the image, leading to problems with inhomogeneously illuminated scenes. Instead of using the gray-level image to detect the facial landmark directly, the Modified Census Transformation (MCT) is performed on the region of interest (ROI) being analyzed, making the system invariant to illumination variations and nonlinear camera characteristics. To achieve efficient and robust fitting with regard to occluded or invisible parts of the face, parameter constraints, coarse to fine fitting and occlusion handling are introduced. The result shows that the new system yields good results even if some areas of the face are occluded or unrecognizable in the image.

Driven by the vision of automated driving (AD), future advanced driver assistant systems (ADAS) will require considerably stronger capability of environment perception compared to the current ones. Among the automotive sensors, camera(s) can deliver the richest environmental information and will be indispensable in AD-scenarios. However, it is practically not easy to match in real time the objects of a camera picture to the original ones on the road. This paper suggests a solution to this problem consisting of simple calculations based on optical and mounting parameters of the camera and road topology data provided by the electronic horizon (eHorizon). Given the fact that the eHorizon is increasingly deployed due to its great potential for optimized fuel and energy consumption, the proposed solution does not require additional budget of material (BOM) and provides large benefit in enabling and enhancing a lot of ADAS and AD applications.

Most countries around the world present regulation and rules, applying on public roads, by putting up traffic signs. Therefore it is useful for driver assistance systems and important for autonomous vehicles to understand the meaning and consequences of those signs. One class of traffic signs that present important information is speed limit signs, which underlie strict norms. In this paper, we will introduce performance enhancing methods for the detection of rectangular traffic signs on the example of speed limit signs in the United States of America (USA). We will show that with a small and acceptable loss of accuracy the number of calculations needed and their complexity can be greatly reduced. Due to that, the energy consumption of the embedded hardware and the processing time per frame are reduced.

Supporting drivers by Advanced Driver Assistance Systems (ADAS) significantly increases road safety. Driver’s emotions recognition is a building block of advanced systems for monitoring the driver’s comfort and driving ergonomics additionally to driver’s fatigue and drowsiness forecasting. This paper presents an approach for driver emotions recognition involving a set of three physiological signals (Electrodermal Activity, Skin Temperature and the Electrocardiogram). Additionally, we propose a CNN (cellular neural network) based classifier to classify each signal into four emotional states. Moreover, the subject-independent classification results of all signals are fused using Dempster-Shafer evidence theory in order to obtain a more robust detection of the true emotional state. The new system is tested using the benchmarked MAHNOB HCI dataset and the results show a relatively high performance compared to existing competing algorithms from the recent relevant literature.

The paper aims to give a compact overview of the changes towards the complete system “vehicle” imposed by the introduction of high-level-automation. At first, the changes are focused on three disruptive elements. Each of these elements is analyzed with respect to its requirements which are caused. Finally four main systemic changes are derived from the requirements. The idea of this breakdown is to propose a path aiming to structure the future developments in the area of highly automated driving (HAD). Highlighted are cross-over effects of requirements, the impact of framework conditions and derived consequences mainly from a technological point of view. Within the conclusions, some approaches are proposed in order to cope with these challenges—especially with focus on market deployment.

The continuous development and integration of Automated Driving Systems (ADS) leads to complex systems. The safety and reliability of such systems must be validated for all possible traffic situations that ADS may encounter on the road, before these systems can be taken into production. Test-driving with ADS functions requires millions of driving kilometers to acquire a sufficiently representative data set for validation. Modern cars produce huge amounts of sensor data. TNO analyses such data to distinguish typical patterns, called scenarios. The scenarios form the key input for validating ADS without the need of driving millions of kilometers. In this paper we present a newly developed technique for automatic extraction and classification of scenarios from real-life microscopic traffic data. This technique combines ‘simple’ deterministic models and data analytics to detect events hidden within terabytes of data.

For the safety of highly automated driving it is essential to identify critical situations. The possible changes over time of a given situation have to be taken into account when dealing with criticality. In this paper, a method to generate trajectories with polynomials is considered. Thus, the trajectories can be tested and characterized analytically. With this approach it is possible to calculate a huge amount of feasible outcomes of a driving situation efficiently and therefore the criticality of the situation can be evaluated.

Safety estimation for a given driving style is of utmost importance in the perspective of fully automated vehicles. Recent progress on accident estimation measurement made by insurance companies has revealed that correlation approaches based on “severe braking” events are not satisfactory. Here we propose a new solution. Taking into account the dynamics of the vehicle and inputs of different nature (visibility, grip, shape of the road, …) a dimensionless quantity is calculated which detects “near miss” events. This solution, based on deep knowledge of causality relationships, can be used for making data-based services richer, for a more relevant estimation of safety level and a better accident anticipation.

An innovative optimal predictive control strategy is proposed in this paper for connected energy management purposes applied to hybrid vehicles, for minimization of energy usage and CO₂ emissions during a given trip, according to the driving conditions that can be predicted by intelligent navigation systems with real-time connectivity to the cloud. The theory proposed for such real-time optimal predictive algorithms is based on the mathematical Pontryagin’s maximum principle that provides general solutions for optimization of dynamic systems with integral criteria, under given constraints. Several technical approaches are presented to get feasible real-time computational effort for this dynamic optimization. The calculation of a “trip planning” becomes then possible in embedded controllers synchronized to more powerful servers and computers connected to the vehicle. Significant gains of more than -10 % of CO₂ emissions are demonstrated, maintaining acceptable performances and drivability.

For the first time in history, the majority of people live now in urban areas. What is more, in the next four decades, the number of people living in the world’s urban areas is expected to grow from 3.5 billion to 5.2 billion. At the same time, populations around the world are rapidly ageing. By 2050, the global population of people aged 60 years and over is expected to reach almost 2 billion, with the proportion of older people doubling between 2006 and 2050. This growth and ageing of the population will pose great challenges for urban mobility, which will be addressed within the SilverStream project. In particular, it will develop and demonstrate a radically new light and affordable Light Electric Vehicle concept for the ageing population in congested European cities with scarce parking space.

With the introduction of electrified drive-trains and autonomous driving features, the requirements for electronic systems in vehicles have rapidly become more and more demanding. In order to face these complex requirement scenarios, a paradigm shift from closed-loop-controlled to truly self-deciding and self-learning automata is strongly needed. In this paper, a novel concept for drive-train platforms enabling self-learning capabilities based on sensor integration, micro and power electronics and secure cloud communication directly integrated into the electric motor will be introduced.

Improving the range of electric vehicles is a key requirement to achieve market acceptance—one important aspect is the efficient management of the electric energy stored into the cells. Within the INCOBAT (INnovative COst efficient management system for next generation high voltage BATteries, http://​www.​incobat-project.​eu/​. The research leading to these results has received funding from the European Union’s Seventh Framework Programme (FP7/2007–2013) under grant agreement no. 608988.) project, we propose innovative and cost efficient battery management systems for next generation high voltage batteries targeting a more accurate estimation of the battery state over the lifetime. Target of this paper is to present the project’s achievements with respect to (a) deployment of safe and secure multicore computing platforms, (b) embedding of electrochemical impedance spectroscopy algorithms, (c) combined thermo-mechanical robustness tests, and finally (d) outlook towards integration of INCOBAT technology into a vehicle demonstrator.

This work proposes a method for improving performances of the energy storage system of an electric car with high autonomy, and analyzes the influences of a system with unbalanced batteries versus different types of traction control of the vehicle. The experimental procedure consisted in conducting a series of test vehicle driving while modifying the traction control. The aim is to analyze the power system behavior versus the use of different strategies vehicle traction, and to study the behavior of each battery. Furthermore, it has been developed a protocol for selective charging of the unbalanced batteries in order to optimize the charging of the whole energy storage system.

This paper discusses major and effective aspects that in future will be important for R&D management in the automotive industry and that will have an enormous impact on the entire value creation chain. Firstly, there are efficiency potentials for reducing development time in design and management of the entire product development process (for example, implementation of an IT system with real-time information, predictive analytics and total networking of all value creation partners). Secondly, the complexity involved in the new technologies and the associated networking being set up between business environment and automotive industry. For example car-to-customer, car-to-car or cooperative intelligent transport systems increasingly demand horizontal and agile cooperation among various industry branches (for example, car industry, telecommunications industry, over-the-top players) and the legislatures of the different markets. Moreover, it becomes clear that the particular public or private highway operators either support or hinder the market success of the new car technologies. Here, the urgent need for close cooperation has been recognized and already exists. However, the process with its milestones is still in a very early phase of product development. Nevertheless, the companies leading the market and the government agencies will cooperate intensely in future and thus also make application of the new technologies possible. For this reason, a form of business architecture was chosen that permits competitively neutral and non-discriminating cooperation with the stakeholders. The theory of the Institutional Role Model (IRM) created a multifunctional approach that permits a holistic type of cooperation and creates very good prerequisites for improving efficiency. The concept of the Institutional Role Model was used successfully for three national research projects (e.g. CONVERGE; Market Design for C-ITS). Within these projects the IRM was used for market phase introduction and penetration. This paper integrates the IRM early into the agile digital product development process. This process redesign will enable managers in the automotive industry to generate unique optimization opportunities in future.