Paper published as PhD thesis. Authored by Newton, A.
This thesis discusses the optimisation of motion platform simulators and was motivated by Loughborough University’s acquisition of a low cost six strut moving platform vehicle simulator. Historically, we see that automotive vehicle simulators are more generally used for human factors experiments that examine driver behaviour during low severity manoeuvres or short events e.g. obstacle avoidance. The purpose of this thesis is to examine the potential for the simulator to be used for vehicle handling experiments where the vehicle is free to explore the limits of the vehicle for sustained periods of time. This research has a significant emphasis on vehicle handling models. In particular, we examine data acquisition systems and testing methods before investigating potential optimisation and identification techniques for estimating vehicle model parameters that have the potential to be implemented on the simulator. Here we examine the possibility of producing high quality vehicle models within a short space of time with a view to rapid identification of different types of vehicle directly from vehicle testing. This includes the data acquisition process and addresses the significance of the sensors and equipment used to measure the vehicle states and the importance of the recorded vehicle manoeuvres and test track characteristics. The second phase was carried out once the simulator was installed and functional. Clearly, the simulator is a piece of experimental equipment and as with any engineering experiment, the equipment should be well understood. Consequently, the accuracy to which it adheres to the real world, i.e. its fidelity, is assessed by investigating the simulators capabilities and limitations and is achieved by analysing the raw performance of the motion platform and conducting driver-in-the-Ioop experiments; this work proves valuable as it is used to optimise how the motion platform responds to vehicle dynamics and provides the motivation behind conducting a driver-in-the-Ioop handling experiment for the final section of this thesis. Here, the simulators potential to be used as a tool to assess race car driver skill is investigated. After conducting various tests in the simulated and real world, the correlation between the subjects simulated and real world performances are used to critically assess the simulators performance and draw conclusions concerning its future potential for handling based research. This thesis shows it possible to use an Inertial GPS Navigation System for capturing vehicle data to good effect and describes how a comprehensive set of new vehicle dynamics measurements can be collected and used for model tuning and optimisation within a relatively short space of time (approximately one day). The work presents substantial evidence that shows how dominant the influence of steer ratio and toe compliance is on the accuracy of the handling models and that they are a likely source of modelling errors. The importance of vehicle slip angle measurement is a particular point if of interest and is examined concurrently with the driving manoeuvres, where some guidelines for test methodology and data collection are established. A novel identification process is also presented with the Identifying Extended KaIman Filter. It has been shown possible to identify separate front and rear tyre models as well as a single tyre model. The thesis also describes the relative importance of motion for vehicle simulators that are to be used for handling based experiments. It appears more valuable to emulate only those vehicle motions that are within the platforms capabilities and limitations in a quest for quality over quantity. Finally, this work demonstrates the simulators potential to be used as tool to evaluate race car driver skill, which also fundamentally assesses the fidelity of the simulator. This is achieved by examining the correlation between a simulated and real world experiment, where we see a positive correlation which indicates high fidelity. Further analysis shows the importance that adequate driver training is being administered before beginning experimentation.