Vehicle dynamics visualization

Doctoral dissertationTU Dortmund2019

Global Chassis Control

Control of a road vehicle's driving dynamics

A single controller turns the requested vehicle acceleration into all actuator commands in one step, making the coordination of separate vehicle dynamics controllers unnecessary.

Johannes Köppern, 2008–2019

Dissertation as PDFRead the abstract

Abstract

The present global chassis control (GCC) was developed during my work at the German Aerospace Center (DLR) in Oberpfaffenhofen near Munich. It forms the interface between two levels:

  • the motion demand (longitudinal, lateral, and yaw acceleration) and
  • the actuator variables in the vehicle (drive/braking torques and steering angle rates).

The GCC exploits the full vehicle dynamic potential: in a holistic approach, it determines all actuator variables from the requested acceleration in a single step. It uses a precise vehicle dynamics model that during operation can be reconfigured and incorporates additional intervention options such as rear-axle steering or torque vectoring.

This eliminates the coordination effort currently required between individual vehicle dynamics control systems — as well as their negative mutual influence.

I wrote my dissertation on this control system and defended it in 2019 at TU Dortmund under Prof. Torsten Bertram.

Abstract of the dissertation

The Global Chassis Control approach of this thesis determines control signals, by which a car will translate a motion demand as precisely as possible. Inputs to the controller are the motion demand on the level of horizontal acceleration, measurements from the vehicle and parameters of the environment. While determining suitable control signals the controller considers the entire motion demand and the entire set of control possibilities at the same time. If there remains any left choice, the controller uses this to improve the driving stability. This is achieved by minimizing the grip usage of each tire. The control approach is applicable to modern vehicles with today’s control units, sensors and estimators. During the development, the applicability of the control system to road vehicles according to the state of the art with regard to available measured variables and computing power of the available control units was taken into account.

Wording from page II of the dissertation (English original)

The project in pictures

Visualization of the driving simulation with the vehicle model
DLR test vehicle ROboMObil (ROMO)
DLR robotic motion simulator: visitors watching the driving simulation
Simulation model in Modelica/Dymola: ISO 7975 experiment

The Control Concept

Over-actuated means the vehicle has more actuator variables than acceleration components (longitudinal, lateral and yaw acceleration). The ROMO has eight actuator variables, the configuration simulated in the dissertation six. The same motion demand can therefore be implemented in several ways, and the GCC uses this freedom of choice to improve driving stability.

For over-actuated vehicles, the GCC first introduces reduced actuator variables that are mapped to the actual actuator variables via auxiliary variables. The auxiliary variables are chosen to achieve a secondary objective. From these auxiliary variables, the vehicle states, and the requested acceleration change, an inverse model then computes the reduced actuator variables (see dissertation).

In detail: chapters 5 to 7 of the dissertation (PDF, from p. 43) (opens in a new window)

Interactive Block Diagram

Click or tap the blocks for detailed explanations from the dissertation.

varefvâvâ̇*vaevådvårefuvaȳmx̂sγ{β, ψ̇, vay}{f̊d, uΔδ}vålim+−++
Filter
Σ
K(s) Controller
Σ
Inverse Model
Vehicle
Estimator
Auxiliary Variable Selector

Swipe the diagram sideways to see the entire signal flow.

Blocks at a glance (signal flow from left to right)

Symbols in the block diagram

A superscript v in front of a symbol marks quantities in the vehicle-fixed coordinate system, a dot or circle above a symbol a time derivative. The names follow the list of symbols in the dissertation (pages I to III, in German).

  • Reference acceleration, the motion request
  • Filtered acceleration command
  • Rate of the filtered acceleration command without consideration of actuator limits (feedforward)
  • Difference between the filtered command and the measured acceleration
  • Jerk request from the controller K(s)
  • Total jerk request to the inverse model
  • Jerk achievable within the actuator limits, fed back to the filter
  • Actuator variables (drive/brake torques and steering angle rates)
  • Measured vehicle acceleration
  • Vehicle measurements for the estimator
  • Estimated variables from the estimator
  • Total auxiliary variable vector from the auxiliary variable selector
  • Sideslip angle, yaw rate and lateral acceleration
  • Quantities passed from the inverse model to the estimator, including the steering angle rates commanded to set a steering angle difference

Test Vehicle ROboMObil (ROMO)

At DLR, I co-developed the ROboMObil (ROMO) test vehicle. The ROMO has four independent drive/braking torques and four independent steering angles — eight actuator variables in total that affect the vehicle acceleration. However, my global chassis control is not limited to this vehicle but can also be applied to other configurations, such as today's production vehicles.

DLRROboMObil
ROboMObil

Playing the video connects to YouTube.

Lateral acceleration in the ISO 7975 maneuver: setpoint and actual traceAxes and legend in the original German of the dissertation.

Simulation

I developed the controller in the Modelica language using Dymola and identified plant parameters with Matlab among other tools. Via the Functional Mock-Up Interface, I exported it as an AUTOSAR component and subsequently validated it in closed-loop simulations with multi-body vehicle models.

The diagram comes from the ISO 7975 maneuver, braking in a curve: the vehicle drives straight ahead, steers into a circular path, and is then braked while cornering. Black is the demanded lateral acceleration, red dashed the one achieved in the simulation. The two lines nearly coincide.

Show simulation results

Robotic Motion Simulator

Using the robotic motion simulator of the DLR Institute for System Dynamics and Control Engineering, I built an interactive driving simulation. The robot moves the driver according to the simulated vehicle dynamics.

DLR Dynamic Motion Simulator Center
Robotic Motion Simulator

Playing the video connects to YouTube.

My Publications

The following publications are associated with this work.

Scientific Publications

Titles are given in the original German.

  1. 2019
  2. 2017
  3. 2013
    • Köppern, J. (2013). Integrierte Fahrwerkregelung – Konzept und Anwendung. In VDI-Bericht 2196 anlässlich der 6. VDI-/VDE-Fachtagung AUTOREG 2013, Baden-Baden.
  4. 2010
    • Köppern, J. (2010). Integrierte Fahrzeugregelung durch einen hybriden Ansatz aus inversem Modell und modellprädiktiver Optimierung. In GMA‑Fachausschuss 1.40, Theoretische Verfahren der Regelungstechnik, Salzburg, Österreich.

Patents

German patent applications, titles in the original German. Applicant: Deutsches Zentrum für Luft‑ und Raumfahrt e.V., inventor: Johannes Köppern.

  1. 2012
  2. 2011

The doctorate in pictures

Talk with slides on the ROboMObil
Audience in the lecture room
Toy car with a wooden moose figure
Sign of TU Dortmund at Campus Nord
Portrait of Johannes Köppern

About

  1. 2003–2008

    Mechanical engineering, TU Munich

    Specialization in control engineering.

  2. 2008–2013

    Research associate, DLR

    At the German Aerospace Center: co-developed the ROboMObil test vehicle, designed the global chassis control presented here and led DLR's contribution to the European research project Modelisar.

  3. 2014–2019

    Founder and managing director, Ocuplus

    Software that gave visually impaired and blind students barrier-free dictionary access in exams.

  4. 2019

    Doctorate, TU Dortmund

    Defense of the dissertation.

  5. 2020–2023

    Data Scientist and R&D Engineer, Dispendix GmbH

    Medical technology. Before that, a position as Senior Project Manager in the automotive industry.

  6. Since 2023

    Focus on AI and Large Language Models

    AI & Data Science Expert at Education Partners, lecturer at HS Macromedia and founder of NarratiQ.

I hold 24 patents in the fields of vehicle dynamics control, accessibility, and medical technology.

XING Profile

Download

The complete dissertation is available for download as PDF.

DOI: 10.17877/DE290R-20431

Download PDF

PDF · 182 pages · 3.9 MB

Structure of the dissertation

Ten chapters; a click opens the chapter in the PDF. Chapter titles are given in the original German.

Cite this thesis

How to cite the dissertation. Title and details are given in the original.

Köppern, J. (2019). Integrierte Fahrdynamikregelung mittels eines hybriden Ansatzes aus Modellinversion, heuristischer beziehungsweise modellprädiktiver relativer Stellgrößenverteilung und der Integration bekannter Regelungssysteme [Dissertation, Technische Universität Dortmund]. https://doi.org/10.17877/DE290R-20431