Development & Simulation of a Test Environment for Vehicle Dynamics a Virtual Test Track Layout.

Transcription

1 Development & Simulation of a Test Environment for Vehicle Dynamics a Virtual Test Track Layout. PhD.C. -Eng. Kmeid Saad 1 1 Introduction Vehicle Dynamic Libraries Virtual Driver ROAD Test Track Development Development & Simulation of a Test Environment Conclusion References Abstract: As computers have gotten faster, and user interfaces have drastically improved, virtual development and simulation platforms have become widely used in the automotive industry for efficiently evaluating hundreds of test conditions. Configurable vehicle dynamics (Engine, Transmission, Brake Hydraulics, Steering, Driveshaft, Differential, suspension, Brakes, Tires) and simulation packages already exists in many integration platforms. These platforms usually also enables the user to create virtual test tracks for vehicles with various dynamic and static parameters to run on. Considering some of the main advantages of virtual test tracks, i.e. high level of repeatability, direct control over road parameters and possibilities in identifying dangerous maneuvers, this paper will concentrates on the development of a virtual test track that aids not only in the development and simulation of a test environment for vehicle dynamics, but also in the Actual construction (on-site) of the test track. The paper also provides an insight on some of the possible benefits of virtual test tracks, such as ease of adjustment and modification, ease of use, flexibility in road configuration and trajectory/path definition, as well as the possibility to run Software In the Loop (SIL) and Hardware In the Loop (HIL) simulations. Keywords: Vehicle dynamic libraries, virtual simulation, integration platforms, virtual test track layout. 1 MSc. Eng. Kmeid Saad, PhD Candidate, Technical University of Graz-Austria/ University Assistant, Faculty of Electrical Engineering, ADAS Masters Course/ University of Applied Sciences Kempten- Germany, kmeid.saad@hs-kempten.de 1

2 1 Introduction Looking at the automotive industry one can see that engineers have achieved dramatic advancements in the technology employed in automobiles. More than ever vehicle dynamics and the ability to test and fine tune them is playing an important role in vehicle design and development. A knowledge of the forces and moments generated by pneumatic (rubber) tires at the ground is essential, especially when this knowledge is further used to generate configurable vehicle dynamics models for virtual integration platforms. Vehicle dynamics in its wide-ranging sense includes all forms of conveyance, i.e. ships, airplanes, railroad trains, track-laying vehicles, as well as rubber-tired vehicles (the automobile) [1], which we will concentrate on. The dominant forces acting on the vehicle to control performance are developed by the tire against the road, so one should also understand the behavior of the tires, over the broad range of operating conditions. Much vehicle dynamics study involve the study of the motions accomplished in accelerating, braking, cornering and ride, how and why these forces are produced. Understanding vehicle dynamics and various associated use cases, can be accomplished at two levels: 1 Empirical: The empirical understanding derives from trial and error where we can learn which factors influence the vehicle performance. However extrapolating, adapting past experience to new condition may produce new, if not, unexpected results. 2 Analytical: The analytical approach attempts to describe the mechanics of interest based on the known laws of physics so that an analytical model can be established. The analytical model is usually represented by algebraic or differential equations. The model provides a predictive and repeatable capability, so that changes necessary to reach a given performance goal can be identified. Favoring the Analytical method for its repeatability and reliability overtime gave birth to many new mathematical models and sub models of the vehicle. Soon enough the ability to simulate these models became a necessity. 2

3 Today with the computational power available, it is now possible to assemble models (equations) for the behavior of individual components of the vehicle that can be integrated to comprehensive models of the overall vehicle. This leads to better simulation and evaluation of the car s behavior. In the following we will continue on elaborating on the process of development and simulation of a test environment for vehicle dynamics, especially on the process and on the advantages of developing a virtual test track, i.e. proofing ground. 2 Vehicle Dynamic Libraries Vehicle dynamic Libraries provide a foundation for model-based vehicle dynamics analysis, in particular related to road-vehicle handling. Vehicle dynamics libraries can be featured as a user extensible and object-oriented architecture that can be accessed to optimize and verify the design of vehicle s systems/subsystems from the early design phases through control design and implementation. The possibility of real-time simulation performance makes vehicle dynamics libraries suitable for HIL/SIL simulations [2]. Vehicle dynamics libraries also provide the possibility to study models in detail for indepth understanding of vehicle components/subcomponents as well as updating or adding new components at different levels [2]. A hierarchical structure can be created or even directly provided, templates and predefined components, configuration of different vehicles is convenient and straightforward for different test scenarios. It is unique in that it provides true multi-body, multi-domain simulation with real-time performance, and model export capabilities [3]. So all in all one can say that vehicle dynamic libraries allow for a very flexible design. 3 Virtual Driver After the construction of the automobile from various vehicle dynamic models, one should also think about the possibility to control the vehicle and drive it through different test cases with different maneuvers. For this purpose the so called Virtual Driver is created and configured. The Virtual Driver enables the user to add the control actions of a human driver to the complete vehicle simulation. These actions include steering, braking, gas pedal position, gear shifting and clutch operation. Sure they can vary from one virtual driver to another, i.e. from one software to another. 3

4 Virtual Driver actions could include the following: Driving within the lane boundaries (corner cutting) Driving speed adaption according to track and vehicle behavior Influence on the speed by other vehicle parameters (gear select, gas, etc...) Orientation and steering Figure 1 illustrates one possible structure of a virtual driver, where the driver is also capable of learning abilities. The importance of the virtual driver also lies in the ability to automatically and consistently adapt to the vehicle to handle it by identifying its behavior. These features are mainly essential in order to exclude any possible human driving errors. This also enables us to execute various maneuvers that are considered to be dangerous if done by human drivers. 4 ROAD Fig.1 Virtual Driver Model The road is one of the most important features that should be modeled or integrated in any simulation platform. It provides the foundation of all use cases since without it the virtual driver will have nowhere to run/drive. Different integration platforms enables you to build open and closed tracks, add obstacles and markers to the road, manipulate specific road conditions and introduce many crucial environmental and external factors to your simulations. The width of the entire track can be defined with specific friction coefficients and different driving lanes can be set to be used: the left or the right one or maybe even in 4

5 the middle of the road. Traffic signs road bumps and others can also be configured and adapted to the road parameters to insure a more realistic test scenarios. Environmental conditions like the temperature, the time of day or the wind velocity for the simulation can also be defined. If the model takes these parameters into account, they will influence the results of your simulation. For specific sensor models also the sun s position, respectively the shadows on the ground can be altered and thus the behavior of the sensor models, this could be of great importance for image based sensors. 5 Test Track Development Until the 1920s, automotive testing was done in the same place as most automobile driving -- on city streets and country roads. But as the automobile became an increasingly important mode of transportation and the roads filled with cars, this ceased to be feasible. It was too dangerous to test cars in public places. In 1924, General Motors opened the Milford Proving Grounds in what was then a fairly isolated portion of Michigan. It was the world's first dedicated automotive proving ground. A typical automotive proving ground looks like a combination of a military base and an amusement park. From the air, Proving Ground consists of loops and whorls, straight lines and circles [4]. Contidrom [5] VW Testgelände Ehra-Lessien [6] Opel Testzentrum Dudenhofen [7] Porsche Testgelände Weissach [8] Audi Neustadt / Donau [8] Daimler Prüfgelände Immendingen [9] BMW Aschheim bei München [10] Fig.2 Proving Grounds in Germany 5

6 This is the place where developers and engineers can test and learn how to get an outof-control skid back under control or how to deal with an unexpected tire blowout and evaluate various vehicle dynamic tunings. Further tests are illustrated bellow: 1. Active Safety-Tests: Braking on μ-split according to ECE R13H. Change test according to ISO Progressive cornering according to ISO Constant cornering according to ISO Braking in a turn according to ISO Car-Tests: Cornering. Braking in a turn. Load change. Steering evaluation. Driving performance measurement. 3. Tire-Tests: Dry handling. Dry braking. Wet brakes. Driving stability (dry). 4. Connected Cars-Tests: Adaptive Cruse Control (ACC). Car-to -car communication. Car-Infrastructure communication. Intersection assistant. 5. City Park: Crossings. Roundabouts. Traffic Signals. Road Signs. Building. 6

7 As our final goal is to integrate all of our virtual components (vehicle, driver, environment, road etc.) in a development and simulation environment a virtual test track layout should also be modeled. The following example roughly shows the process of developing such a model based on a given area/physical developing ground. 1. Airfield (former military airfield) Google Maps: 2. Airfield Testing Ground AutoCAD: 3. Testing Ground Digital Track (CarMaker): 7

8 8

9 Now that the virtual test track is generated, as shown in the previous images (using IPG CarMaker), the next step would be to configure the simulation to the user s needs. 6 Development & Simulation of a Test Environment At this stage all of the predefined components come together to create a configurable and repeatable testing environment. Adding Environmental factors (Fog). 9

10 Conducting and comparing specific use cases, (Emergency brake assist). 7 Conclusion As a conclusion and in order to get to benefit from all of the simulation capabilities, we should be able to answer the following questions: Are the two approaches equivalent?? If yes, then how to prove it? If no, then by what % the virtual simulation approach does reflects the real life testing approach? Table 1 represents an example of a look up table that could be populated with data based on test results and comparison between virtual and real testing. 10

11 7 References [1] Fundamentals of Vehicle Dynamic, Thomas D.Gillespie [2] [3] [4] [5] [6] [7] [8] [9] html [10] 11