Search Results

Photographs and video recordings of vehicle crashes and accident sites are more prevalent than ever, with dash mounted cameras, surveillance footage, and personal cell phones now ubiquitous. The information contained in these pictures and videos provide critical information to understanding how crashes occurred, and  analyze physical evidence. This course teaches the theory and techniques for getting the most out of digital media, including correctly processing raw video and photographs, correcting for lens distortion, and using photogrammetric techniques to convert the information in digital media to usable scaled three-dimensional data.

Convolutional neural networks are the de facto method of processing camera, radar, and lidar data for use in perception in ADAS and L4 vehicles, yet their operation is a black box to many engineers. Unlike traditional rules-based approaches to coding intelligent systems, networks are trained and the internal structure created during the training process is too complex to be understood by humans, yet in operation networks are able to classify objects of interest at error rates better than rates achieved by humans viewing the same input data.

In the evolving landscape of automated driving systems, the critical role of vehicle localization within the autonomous driving stack is increasingly evident. Traditional reliance on Global Navigation Satellite Systems (GNSS) proves to be inadequate, especially in urban areas where signal obstruction and multipath effects degrade accuracy. Addressing this challenge, this paper details the enhancement of a localization system for autonomous public transport vehicles, focusing on mitigating GNSS errors through the integration of a LiDAR sensor. The approach involves creating a 3D map using the factor graph-based LIO-SAM algorithm based on GNSS, vehicle odometry, IMU and LiDAR data. The algorithm is adapted to the use-case by adding a velocity factor and altitude data from a Digital Terrain model. Based on the map a state estimator is proposed, which combines high-frequency LiDAR odometry based on FAST-LIO with low-frequency absolute multiscale ICP-based LiDAR position estimation.

Autonomous Driving is being utilized in various settings, including indoor areas such as industrial halls. Additionally, LIDAR sensors are currently popular due to their superior spatial resolution and accuracy compared to RADAR, as well as their robustness to varying lighting conditions compared to cameras. They enable precise and real-time perception of the surrounding environment. Several datasets for on-road scenarios such as KITTI or Waymo are publicly available. However, there is a notable lack of open-source datasets specifically designed for industrial hall scenarios, particularly for 3D LIDAR data. Furthermore, for industrial areas where vehicle platforms with omnidirectional drive are often used, 360° FOV LIDAR sensors are necessary to monitor all critical objects. Although high-resolution sensors would be optimal, mechanical LIDAR sensors with 360° FOV exhibit a significant price increase with increasing resolution.

Investigating human driver behavior enhances the acceptance of the autonomous driving and increases road safety in heterogeneous environments with human-operated and autonomous vehicles. The previously established driver fingerprint model, focuses on the classification of driving style based on CAN bus signals. However, driving styles are inherently complex and influenced by multiple factors, including changing driving environments and driver states. To comprehensively create a driver profile, an in-car measurement system based on the Driver-Driven vehicle-Driving environment (3D) framework is developed. The measurement system records emotional and physiological signals from the driver, including ECG signal and heart rate. A Raspberry Pi camera is utilized on the dashboard to capture the driver's facial expressions and a trained convolutional neural network (CNN) recognizes emotion. To conduct unobtrusive ECG measurements, an ECG sensor is integrated into the steering wheel.

In pursuit of safety validation of automated driving functions, efforts are being made to accompany real world test drives by test drives in virtual environments. To be able to transfer highly automated driving functions into a simulation, models of the vehicle’s perception sensors such as lidar, radar and camera are required. In addition to the classic pulsed time-of-flight (ToF) lidars, the growing availability of commercial frequency modulated continuous wave (FMCW) lidars sparks interest in the field of environment perception. This is due to advanced capabilities such as directly measuring the target’s relative radial velocity based on the Doppler effect. In this work, an FMCW lidar sensor simulation model is introduced, which is divided into the components of signal propagation and signal processing. The signal propagation is modeled by a ray tracing approach simulating the interaction of light waves with the environment.

As part of the safety validation of advanced driver assistance systems (ADAS) and automated driving (AD) functions, it is necessary to demonstrate that the frequency at which the system exhibits hazardous behavior (HB) in the field is below an acceptable threshold. This is typically tested by observation of the system behavior in a field operational test (FOT). For situations in which the system under test (SUT) actively intervenes in the dynamic driving behavior of the vehicle, it is assessed whether the SUT exhibits HB. Since the accepted threshold values are generally small, the amount of data required for this strategy is usually very large. This publication proposes an approach to reduce the amount of data required for the evaluation of emergency intervention systems with a state machine based intervention logic by including the time periods between intervention events in the validation process.

The optimization and further development of automated driving functions offers great potential to relieve the driver in various driving situations and increase road safety. Simulative testing in particular is an indispensable tool in this process, allowing conclusions to be drawn about the design of automated driving functions at a very early stage of development. In this context, the use of driving simulators provides support so that the driving functions of tomorrow can be experienced in a very safe and reproducible environment. The focus of the acceptance and optimization of automated driving functions is particularly on vehicle lateral control functions. As part of this paper, a test person study was carried out regarding manual vehicle lateral control on the dynamic vehicle road simulator at the Institute of Automotive Engineering.

Traditional CACC systems utilize inter-vehicle wireless communication to maintain minimal yet safe inter-vehicle distances, thereby improving traffic efficiency. However, introducing communication delays generates system uncertainties that jeopardize string stability, a crucial requirement for robust CACC performance. To address these issues, we introduce a decentralized Model Predictive Control (MPC) approach that incorporates Kalman Filters and state predictors to counteract the uncertainties posed by noise and communication delays. We validate our approach through MATLAB Simulink simulations, using stochastic and mathematical models to capture vehicular dynamics, Wi-Fi communication errors, and sensor noises. In addition, we explore the application of a Reinforcement Learning (RL)-based algorithm to compare its merits and limitations against our decentralized MPC controller, considering factors like feasibility and reliability.

Aeroacoustics is important in the automotive industry, as it significantly influences driving comfort. Particularly in the case of battery electric vehicles (BEVs), the flow noise is already crucial at lower driving speeds, since these generate barely any drive noise and the masking effects produced by the engine are eliminated. Due to the increasing importance of drag minimization and elimination of the exhaust system, the underbody of BEVs is typically very streamlined and exhibits a low acoustic interference potential. However, even small geometric modifications to the vehicle can lead to changes in the flow around the vehicle and consequently to significant noise sources. Thus, significant flow resonances in the low frequency range below 30 Hz have been detected on certain vehicle configurations. Initial investigations have shown that the flow around the front wheel spoilers is relevant for the development of the flow phenomenon.

When travelling in an open-jet wind tunnel, the path of an acoustic wave is affected by the flow causing a shift of source positions in acoustical maps of phased arrays outside the flow. The well-known approach of Amiet attempts to correct for this effect by computing travel times between microphones and map points based on the assumption that the boundary layer of the flow, the so-called shear-layer, is infinitely thin and refracts the acoustical ray in a conceptually analogy to optics. However, in reality, the turbulent nature of both the not-so thin shear-layer and the acoustic emission process itself causes an additional smearing of sources in acoustic maps, which in turn causes deconvolution methods based on these maps - the most prominent example being CLEAN-SC - to produce certain ring effects, so-called halos, around sources.

Design verification and quality control of automotive components require the analysis of the source location of ultra-short sound events, for instance the engaging event of an electromechanical clutch or the clicking noise of the aluminium frame of a passenger car seat under vibration. State-of-the-art acoustic cameras allow for a frame rate of about 100 acoustic images per second. Considering that most of the sound events introduced above can be far less than 10ms, an acoustic image generated at this rate resembles an hard-to-interpret overlay of multiple sources on the structure under test along with reflections from the surrounding test environment. This contribution introduces a novel method for visualizing impulse-like sound emissions from automotive components at 10x the frame rate of traditional acoustic cameras.

Broadband active noise control algorithms require high-performance so multi-channel control to ensure high performance, which results in very high computational power and expensive DSP. When the control filter update part need a huge computational power of the algorithm is separated and calculated by the server, it is possible to reduce cost by using a low-cost DSP in a local vehicle, and a performance improvement algorithm requiring a high computational power can be applied to the server. In order to achieve the above goal, this study analyzed the maximum delay time when communication speed is low and studied response measures to ensure data integrity at the receiving location considering situations where communication speed delay and data errors occur.

This course highlights the technologies enabling ADAS and how they integrate with existing passive occupant crash protection systems, how ADAS functions perceive the world, make decisions, and either warn drivers or actively intervene in controlling the vehicle to avoid or mitigate crashes. Examples of current and future ADAS functions, and various sensors utilized in ADAS, including their operation and limitations, and sample algorithms, will be discussed and demonstrated. The course utilizes a combination of hands-on activities, including computer simulations, discussion and lecture.

The Selective Laser Melting (SLM) process is employed in high-precision layer-by-layer Additive Manufacturing (AM) on powder bed and aims to fabricate high-quality structural components. To gain a comprehensive understanding of the process and its optimization, both modeling and simulation in conjunction with extensive experimental studies along with laser calibration studies have been attempted. Multiscale and multi-physics-based simulations have the potential to bring out a new level of insight into the complex interaction of laser melting, solidification, and defect formation in the SLM parts. SLM process encompasses various physical phenomena during the formation of metal parts, starting with laser beam incidence and heat generation, heat transfer, melt/fluid flow, phase transition, and microstructure solidification. To effectively model this Multiphysics problem, it is imperative to consider different scales and compatible boundary conditions in the simulations.

The paper presents a theoretical framework for the detection and first-level preliminary identification of potential defects on aero-structure components while employing ultrasonic guided wave based structural health monitoring strategies, systems and tools. In particular, we focus our study on ground inspection using laser-Doppler scan of surface velocity field, which can also be partly reconstructed or monitored using point sensors and actuators on-board structurally integrated. Using direct wave field data, we first question the detectability of potential defects of unknown location, size, and detailed features. Defects could be manufacturing defects or variations, which may be acceptable from design and qualification standpoint; however, those may cause significant background signal artifacts in differentiating structure progressive damage or sudden failure like impact-induced damage and fracture.