Search Results

The introduction of autonomous vehicles has gained significant attention due to its potential to revolutionize mobility and safety. A critical aspect underpinning the functionality of these autonomous vehicles is their sensor perception system. Demonstrating the reliability of the environment perception sensors and sensor fusion algorithms is, therefore, a necessary step in the development of automated vehicles. Field tests offer testing conditions that come closest to the environment of an automated vehicle in the future. However, a significant challenge in field tests is to obtain a reference truth of the surrounding environment. Here, we propose a pipeline to assess the sensor reliabilities without the need for a reference truth. The pipeline uses a model to estimate the reliability of redundant sensors. To do this, it relies on a binary representation of the surrounding area, which indicates either the presence or absence of an object.

Bicycle-drawn cargo trailers with an electric drive to enable the transportation of high cargo loads are used as part of the last-mile logistics. Depending on the load, the total mass of a trailer can vary between approx. 50 and 250 kg, potentially more than the mass of the towing bicycle. This can result in major changes in acceleration and braking behavior of the overall system. While existing systems are designed primarily to provide sufficient power, improvements are needed in the powertrain control system in terms of driver safety and comfort. Hence, we propose a novel prototype that allows measurement of the tensile force in the drawbar which can subsequently be used to design a superior control system. In this context, a sinusoidal force input from the cyclist to the trailer according to the cadence of the cyclist is observed. The novelty of this research is to analyze whether torque impulses of the cyclist can be reduced with the help of Model Predictive Control (MPC).

Deep neural network models have been widely used for environment perception of intelligent vehicles. However, due to models’ innate probabilistic property, the lack of transparency, and sensitivity to data, perception results have inevitable uncertainties. To compensate for the weakness of probabilistic models, many pieces of research have been proposed to analyze and quantify such uncertainties. For safety-critical intelligent vehicles, the uncertainty analysis of data and models for environment perception is especially important. Uncertainty estimation can be a way to quantify the risk of environment perception. In this regard, it is essential to deliver a comprehensive survey. This work presents a comprehensive overview of uncertainty estimation in deep neural networks for environment perception of intelligent vehicles.

In order to predict reality as accurately as possible leads to the fact that numerical models in automotive vibroacoustic problems become increasingly high dimensional. This makes applications with a large number of model evaluations, e.g. optimization tasks or uncertainty quantification hard to solve, as they become computationally very expensive. Engineers are thus faced with the challenge of making decisions based on a limited number of model evaluations, which increases the need for data-efficient methods and reduced order models. In this contribution, variational autoencoders (VAEs) are used to reduce the dimensionality of the vibroacoustic model of a vehicle body and to find a low-dimensional latent representation of the system.

A reliable environment perception is a requirement for safe automated driving. For evaluating and demonstrating the reliability of the vehicle’s environment perception, field tests offer testing conditions that come closest to the vehicle’s driving environment. However, establishing a reference ground truth in field tests is time-consuming. This motivates the development of a procedure for learning the vehicle’s perception reliability from fleet data without the need for a ground truth, which would allow learning the perception reliability from fleet data. In Berk et al. (2019), a method based on Bayesian inference to determine the perception reliability of individual sensors without the need for a ground truth was proposed. The model utilizes the redundancy of sensors to learn the sensor’s perception reliability. The method was tested with simulated data.

The recent growth of available computational resources has enabled the automotive industry to utilize unsteady Computational Fluid Dynamics (CFD) for their product development on a regular basis. Over the past years, it has been confirmed that unsteady CFD can accurately simulate the transient flow field around complex geometries. Concerning the aerodynamic properties of road vehicles, the detailed analysis of the transient flow field can help to improve the driving stability. Until now, however, there haven’t been many investigations that successfully identified a specific transient phenomenon from a simulated flow field corresponding to driving stability. This is because the unsteady flow field around a vehicle consists of various time and length scales and is therefore too complex to be analyzed with the same strategies as for steady state results.

The Deep Orange program is a concept vehicle development program focused on providing hands-on experience in design, engineering, prototyping and production planning as part of students’ two-year MS graduate education. Throughout this project, the team was challenged to create innovative concepts during the ideation phase as part of building the running vehicle. This paper describes the usability studies performed on two of the vehicle concepts that require driver interaction. One concept is a human machine interface (HMI) that uses a holographic companion that can act as a concierge for all functions of the vehicle. After creating a prototype using existing technologies and developing a user interface controlled by hand gestures, a usability study was completed with older adults. The results suggest the input method was not intuitive. Participants demonstrated better performance with tasks using discrete hand motions in comparison to those that required continuous motions.

A key criterion for launching autonomous vehicles on real roads is the knowledge of their capability to ensure traffic safety. In contrast to ADAS, deriving this measure of safety is difficult to achieve as the functional scope of an autonomous driving function exceeds by far the one of ADAS. As a consequence, real-world testing solely is not sufficient enough to cover the required test volume. This assessment problem imposes new requirements on a valid test concept for automated driving. A possible solution represents simulation by enabling it to generate reliable test kilometers. As a first step, we discuss in this paper the feasibility of simulation frameworks to re-simulate a real-world test in certain scenarios. We will demonstrate that even with ground truth information of the vehicle odometry and corresponding environment model an acceptable accordance of functional behavior is not guaranteed.

Over the past years, the question as to what may be the powertrain of the future has become ever more apparent. Aiming to improve upon a given technology, the internal combustion engine still offers a number of development paths in order to maintain its position in public and private mobility. In this study, an innovative combustion process is investigated with the goal to further approximate the ideal Otto cycle. Thus far, similar approaches such as Homogeneous Charge Compression Ignition (HCCI) shared the same objective yet were unable to be operated under high load conditions. Highly increased control efforts and excessive mechanical stress on the components are but a few examples of the drawbacks associated with HCCI. The approach employed in this work is the so-called Spark Assisted Compression Ignition (SACI) in combination with a pre-chamber spark plug, enabling short combustion durations even at high dilution levels.

The analysis of the acoustic behavior of flow fields has gained importance in recent years, especially in the automotive industry. The comfort of the driver is heavily influenced by the noise levels and characteristics, especially during long distance drives. Simulation tools can help to analyze the acoustic properties of a car at an early stage of the development process. This work focuses on the low-frequency sound effects, which can be a significant noise component under certain operating conditions. As a first step in the fluid-structure interaction workflow, the flow around a series-production vehicle is simulated, including passenger cabin and underhood flow. The complexity of this model poses extensive demands on the simulation software, concerning meshing, turbulence modeling and level of parallelism. We conducted a transient simulation of the compressible fluid flow, using a hybrid RANS/LES approach.

The effectiveness of ADAS addressing property damage has an increasing impact on car manufacturers, insurers and customers, as accident avoidance or mitigation can lead to loss reduction. In order to obtain benefits, it is essential that ADAS primarily address monetarily relevant accident scenarios. Furthermore, sensor technologies and algorithms have to be configured in a way that relevant accident situations can be sufficiently avoided at reasonable system costs. A new methodology is developed to identify and configure monetarily effective parameters for ADAS during parking and maneuvering. ADAS parameters e.g. relevant accident scenarios, required crash avoidance speeds and different sensor layouts are analyzed and evaluated using a real-world in-depth accident database of insurance claims provided by Allianz Center for Technology and Allianz Automotive Innovation Center. For this purpose, a sensitivity analysis is conducted to identify most monetarily effective accident scenarios.

In times when automated driving is becoming increasingly relevant, dynamic simulators present an appropriate simulation environment to faithfully reproduce driving scenarios. A realistic replication of driving dynamics is an important criterion to immerse persons in the virtual environments provided by the simulator. Motion Cueing Algorithms (MCAs) compute the simulator’s control input, based on the motions of the simulated vehicle. The technical restrictions of the simulator’s actuators form the main limitation in the execution of these input commands. Typical dynamic simulators consist of a hexapod with six degrees of freedom (DoF) to reproduce the vehicle motion in all dimensions. Since its workspace dimensions are limited, significant improvements in motion capabilities can be achieved by expanding the simulator with redundant DoF by means of additional actuators.

The Deep Orange program immerses automotive engineering students into the world of an OEM as part of their 2-year graduate education. In support of developing the program’s seventh vehicle concept, the students studied the sponsoring brand essence, conducted market research, and made a heuristic assessment of competitor vehicles. The upfront research lead to the definition of target customers and setting vehicle level targets that were broken down into requirements to develop various vehicle sub-systems. The powertrain team was challenged to develop a scalable propulsion concept enabled by a common vehicle architecture that allowed future customers to select (at the point of purchase) among various levels of electrification best suiting their needs and personal desires. Four different configurations were identified and developed: all-electric, two plug-in hybrid electric configurations, and an internal combustion engine only.

A main challenge when developing next generation architectures for automated driving ECUs is to guarantee reliable functionality. Today’s fail safe systems will not be able to handle electronic failures due to the missing “mechanical” fallback or the intervening driver. This means, fail operational based on redundancy is an essential part for improving the functional safety, especially in safety-related braking and steering systems. The 2-out-of-2 Diagnostic Fail Safe (2oo2DFS) system is a promising approach to realize redundancy with manageable costs. In this contribution, we evaluate the reliability of this concept for a symmetric and an asymmetric Electronic Power Steering (EPS) ECU. For this, we use a Markov chain model as a typical method for analyzing the reliability and Mean Time To Failure (MTTF) in majority redundancy approaches. As a basis, the failure rates of the used components and the microcontroller are considered.

Improving fuel efficiency while meeting relevant emission limits set by emissions legislation is among the main objectives of engine development. Simultaneously the development costs and development time have to be steadily reduced. For these reasons, the high demands in terms of quality and validity of measurements at the engine test bench are continuously rising. This paper will present a new methodology for efficient testing of an industrial combustion engine in order to improve the process of decision making for combustion-relevant component setups. The methodology includes various modules for increasing measurement quality and validity. Modules like stationary point detection to determine steady state engine behavior, signal quality checks to monitor the signal quality of chosen measurement signals and plausibility checks to evaluate physical relations between several measurement signals ensure a high measurement quality over all measurements.

Highly Automated Driving (HAD) opens up new middle-term perspectives in mobility and is currently one of the main goals in the development of future vehicles. The focus is the implementation of automated driving functions for structured environments, such as on the motorway. To achieve this goal, vehicles are equipped with additional technology. This technology should not only be used for a limited number of use cases. It should also be used to improve Active Safety Systems during normal non-automated driving. In the first approach we investigate the usage of machine learning for an autonomous emergency braking system (AEB) for the active pedestrian protection safety. The idea is to use knowledge of accidents directly for the function design. Future vehicles could be able to record detailed information about an accident. If enough data from critical situations recorded by vehicles is available, it is conceivable to use it to learn the function design.

The increasing complexity of automotive functions which are necessary for improved driving assistance systems and automated driving require a change of common vehicle architectures. This includes new concepts for E/E architectures such as a domain-oriented vehicle network based on powerful Domain Control Units (DCUs). These highly integrated controllers consolidate several applications on different safety levels on the same ECU. Hence, the functions depend on a strictly separated and isolated implementation to guarantee a correct behavior. This requires middleware layers which guarantee task isolation and Quality of Service (QoS) communication have to provide several new features, depending on the domain the corresponding control unit is used for. In a first step we identify requirements for a middleware in automotive DCUs. Our goal is to reuse legacy AUTOSAR based code in a multicore domain controller.

The quiet nature of hybrid and electric vehicles has triggered developments in research, vehicle manufacturing and legal requirements. Currently, three countries require fitting an Approaching Vehicle Alerting System (AVAS) to every new car capable of driving without a combustion engine. Various other geographical areas and groups are in the process of specifying new legal requirements. In this paper, the design challenges in the on-going process of designing the sound for quiet cars are discussed. A proposal is issued on how to achieve the optimum combination of safety, environmental noise, subjective sound character and technical realisation in an iterative sound design process. The proposed sound consists of two layers: the first layer contains tonal components with their pitch rising along with vehicle speed in order to ensure recognisability and an indication of speed.

Among the issues affecting the design process of a vehicle, there is the lack of multidisciplinary knowledge among the different teams involved. This often leads to the risk of loosing important key points from the initial concept idea of designers to the final vehicle package definition made by engineers. Therefore, this study builds up a method based on parameters defining exterior aesthetic priorities according to car segment to support engineers involved in the automotive design process. In particular, during the early design-engineering phase, this method should help them to understand better vehicle proportions defined by designers. This work is currently used in the course of design and simulation of road vehicles (Chair of Automotive Technology, TUM) to explain fundamentals of automotive design.

As computational methodologies become more integrated into industrial vehicle pre-development processes the potential for high transient vehicle thermal simulations is evident. This can also been seen in conjunction with the strong rise in computing power, which ultimately has supported many automotive manufactures in attempting non-steady simulation conditions. The following investigation aims at exploring an efficient means of utilizing the new rise in computing resources by resolving high time-dependent boundary conditions through a series of averaging methodologies. Through understanding the sensitivities associated with dynamic component temperature changes, optimised boundary conditions can be implemented to dampen irrelevant input frequencies whilst maintaining thermally critical velocity gradients.