direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Inhalt des Dokuments

 Masterarbeiten (Details und Kontaktinformationen: siehe unten):

  • An Integrated Estimator for Online Vehicle Mass and Road Slope based on IMU
  • Real-time Estimation the Position of Vehicle Center of Gravity based on IMU
  • Steering Feedback Torque Computation for Steer-by-Wire System Based on Artificial Neural Network
  • Entwicklung einer Fahrzeugsteuerung für automatisiertes Fahren


  • auf Anfrage

Bitte auch eigene Vorschläge mit den wissenschaftlichen Mitarbeitern des Fachgebiets absprechen 

Wenn Sie Interesse haben, informieren Sie sich bitte hier über das Vorgehen.

An Integrated Estimator for Online Vehicle Mass and Road Slope based on IMU pdf

A significant number of mass estimation algorithms have been developed with longitudinal dynamics. However, most of these approaches are based on the method of constant vehicle mass and time-varying road grade. Although both vehicle mass and road grad could be identified, the nature of time-varying road grade could lead to significant disturbance for the precision of vehicle mass estimation. In addition, parameters including rolling resistance, drag coefficient and wind velocity are necessary, which also should be estimated. Consideration of the limitations about estimation approaches, a novel approach will be proposed in this task.

In order to decouple the coupled influence of road grade on vehicle mass estimation, this task proposed a novel method for vehicle mass estimation based on frequency-information-extraction. Figure 1 shows that the principle of the MTWFFT method. Normally, the dynamic signals are directly obtained from measurements in the time domain. This task adopts vertical acceleration and angular velocity from IMU measurement which treats the vertical acceleration of the body mass as inputs in the dynamic equations.

The estimation approach is based on the observation that the frequency spectrum of the both the vertical accelerometer and the motions angular rate significantly varies as a function of the vehicle loading mass and its distribution. This can be indicated by the instruction in Fig. 2, which shows the ratio index and the accelerometer spectra obtained where the same vehicle is facing the same road profile, but with different loading mass, located in the same position inside the vehicle.
  • Simulink modelling of vehicle dynamics and kinematics system based on different road profile, then vehicle body movement signals were collected, processed and filtered respectively, and body mass acc. and angular rates which as the inputs of the ratio index.

  • Simulation based on Simulink model with ANN estimators which will be developed, mainly focus on the novel ANN algorithm.

    • Developed ANN estimator for the vehicle body system to identify the loading mass with the ratio index.
    • For training the ANN Estimator, we choose variable loading mass under different road profile conditions, until the estimated results match with the real one well
    • After finished training, then we test any loading mass under random road profile.

  • Evaluation for the simulation results of the ANN algorithm.
  • Validation about the estimation algorithm in CarMaker-Simulink co-simulation.

  • Good academic background
  • Experience in MATALB / Simulink
  • Fluent in writing & speaking English
  • Good at vehicle dynamics and control algorithm
  • Familiar with artificial neural network control algorithm.
    Xiongshi Wang, M. Sc.

    Real-time Estimation the Position of Vehicle Center of Gravity based on IMU pdf
    The position of a vehicle’s center of gravity (CoG) are used as an important parameter for vehicle safety control systems for improving handling stability, while it can be changed considerably according to various driving conditions. Therefore, in order to make vehicle safety control systems to have the better performance, it is essential to obtain the accurate CoG position. However, it is generally difficult to acquire the value of this parameter directly through sensors due to cost reasons. In this task, a practical algorithm for estimating vehicle’s CoG position in real time will be proposed.

    This algorithm is derived only based on pitch and roll movements of the vehicle. Figure 1, Vehicle dynamics model with roll and pitch movements. Moreover, the main differences in the proposed algorithm compared to previous studies is that it does not require information such as vehicle mass, vehicle moments of inertia, road grade or tire-road surface friction, which are difficult to acquire.

    In the proposed algorithm, the relationship between the tire vertical force and the corresponding Pitch&Roll angles are used to determine the CoG position. To demonstrate a practical use of the proposed algorithm, the tire vertical force distribution will be tested under variable loading position and payloads. The proposed CoG estimation algorithm and its practical use will be verified via simulations and experiments with using a test vehicle equipped with Inertial Measurement Unit (IMU).
    • Literature research on the vehicle inertial parameters - the CoG position.
    • Mathematical model of road-vehicle kinematics system based on Pitch&Roll motions in forms of differential equation and state space equation.
    • Simulink model of the sprung mass kinematic model system based on different road profile, then the pitch and roll angle rates, longitudinal and lateral accelerations from IMU of the sprung mass body were collected, processed and filtered respectively, and these measurements which as the input of the estimator.
    • Simulation based on Simulink model with estimators which will be developed.
    • Evaluation for the simulation results.
    • Validation about the estimation algorithm in HiL system based on Vertical Test Bench.
      • Basic knowledge of vehicle vertical dynamics and Roll&Pitch kinematics.
      • Familiar with control algorithms.
      • Ability of mechanic analyzing and mathematical modelling.
      • Theory of matrix analysis for converting differential equations to state space.
      • Use Matlab/Simulink to make simulation models for simulation.
      • Ability of doing experiment in HiL system.
        Xiongshi Wang, M. Sc.
        Steering Feedback Torque Computation for Steer-by-Wire System Based on Artificial Neural Network PDF
        Steer-by-Wire (SbW) is a highly prospective steering technology for intelligent vehicles. As the elimination of the mechanical connection between the steering wheel and the steered wheels, it becomes free to adjust steering characteristics for an SbW system, which contributes to improving steering sensitivity, steering stability, and handling performance. On the other hand, how to generate a proper artificial steering feedback torque comes to be a vital and challenging issue.
        The target of this research is to design a new method to calculate the desired steering feedback torque which provides the driver with a realistic steering feel, the same as in an electrical power steering (EPS) system. For this purpose, an artificial neural network (ANN, shown in Fig. 1) is adopted to estimate the steering feedback torque, as ANN is capable of learning complex non-linear correlations without requiring specific mathematical models. The development of this ANN requires certain steps. Firstly, piles of steering data are recorded by imposing various steering maneuvers in IPG CarMaker to the vehicle simulator coupled with the well-performed EPS steering testbench (Fig. 2) which is available in our Department Automotive Engineering. Secondly, the inputs of ANN are selected by parameter sensitivity analysis of the test data. Besides, the training and validation of this ANN are conducted based on the steering database. Overall, a proper approximation of the steering feedback torque for SbW system can be developed, which provides the driver with a similar feeling as an EPS system.
        • Knowledge acquisition on EPS testbench and IPG CarMaker
        • Parametric configuration for the EPS testbench and its operating data collection under various steering maneuvers implemented in CarMaker
        • Parameter sensitivity analysis of the test data for the selection of the ANN inputs. And, pre-processing of data to train the ANN
        • ANN training and validation based on the steering database
        • Good knowledge of Matlab/Simulink
        • Familiar with test data analysis and processing
        • Experience with real-time or hardware-in-the-loop systems
        • Basic understanding of artificial neural network algorithm
        Qiao Zhang, M.Sc.
        Entwicklung einer Fahrzeugsteuerung für automatisiertes Fahren
        Die Entwicklung von automatisiertem Fahren ist sehr komplex und mit hohem Aufwand verbunden. Um zukünftig automatisiertes Fahren sowie deren Entwicklungsprozess untersuchen zu können, wird am Fachgebiet Kraftfahrzeuge ein Versuchsfahrzeug aufgebaut, das auf einem Serienfahrzeug basiert. Der Versuchsträger soll mit einer Fahrzeugsteuerung, Umfeldsensorik ausgestattet werden um letztlich automatisiert Fahren zu können. Im Rahmen der Arbeit soll ein existierendes Steuerungssystem analysiert und auf eine Echtzeit Entwicklungsumgebung übertragen werden. Bei der Analyse sollen die verschiedenen Funktionen, wie bspw. „Operator override“ dargestellt und überprüft werden, ob und wie diese mittels der Echtzeitentwicklungsumgebung umgesetzt werden können.
        • Analyse des existierenden Steuerungssystems
        • Funktionsübertragung auf eine Echtzeitentwicklungsumgebung
        • Dokumentation der Software
        • Ein/Aufbau im Fahrzeug
        • Test der Funktionalität im Fahrzeug
        • Gute Kenntnisse in Matlab/Simulink
        André Hartwecker, M. Sc.

        Zusatzinformationen / Extras


        Schnellnavigation zur Seite über Nummerneingabe

        Diese Seite verwendet Matomo für anonymisierte Webanalysen. Mehr Informationen und Opt-Out-Möglichkeiten unter Datenschutz.