Showing papers by "Andras Kemeny published in 2012"

Explicit MPC motion cueing algorithm for real-time driving simulator

Abstract: Although Model Predictive Control (MPC) technology can now be found in a wide variety of industrial process control applications, its application in the field of Motion Cueing Algorithm (MCA) is a relatively new issue. Its efficiency has been proven in the driving simulation thanks to the authors' original work (explicit algorithm) and to others (implicit algorithm). The key for applying successfully the MPC algorithm in the field of MCA is to find an appropriate algorithm's stability condition and an efficient in-line or off-line optimization algorithm. The stability condition proposed by this paper reduces significantly the number of partitioned regions for explicit algorithm and thus guarantees the real time tilt-linear optimal motion cues with complicated constraints. This condition can also be applied in the implicit algorithm. The first test in Renault's simulator, ULTIMATE, has shown that the fidelity of driving simulation is effectively improved. It will greatly enhance the MPC algorithm's application in the field of simulator motion cues.

Motion cueing algorithms for a real-time automobile driving simulator

Abstract: The MCA (Motion Cueing Algorithm) for driving simulator takes into account the simulator’s workspace limits and the driver’s motion perception thresholds to reproduce simulated vehicle’s accelerations. For the motion based driving simulators, the most applied MCAs are the classical and the optimal filters. This paper presents a new algorithm, called MPC (Model Predictive Control) explicit algorithm. Compared with the filters’ algorithms, the MPC integrates directly the system constraints into its optimization process, and then gives a real optimal solution and hardly needs the tuning process to check the workspace limits and the driver’s perception thresholds. The reported MCA studies based on MPC implicit algorithm need high computational costs which can destroy the stability properties of optimal MPC in a real-time system. The proposed workspace limit condition improves significantly the MPC optimal stable condition for its application in MCA. The current MPC algorithm can achieve the complicated 2dofs optimization. Key words: Motion cueing algorithm, explicit MPC algorithm, washout filter, tilt technique, real-time simulator.

A LQR washout algorithm for a driving simulator equipped with a hexapod platform : the relationship of neuromuscular dynamics with the sensed illness rating

Abstract: This study proposes a method and an experimental validation to analyze dynamics response of the drivers with respect to the type of the control used in the hexapod driving simulator. In this article, two different forms of motion platform tracking control have been performed: - Classical motion cueing algorithm - LQR motion cueing algorithm For each situation, the EMG (electromyography) data have been registered from arm muscles of the drivers (flexor carpi radialis, brachioradialis). In addition, the acceleration based illness ratings (IR) have been computed. In order to process the data of the EMG and IR, the linear regression with a significance level of 0.05 has been assigned. Three cases have been evaluated: 1) Time exposure neuromuscular dynamics and vestibular–vehicle level conflict illness ratings 2) Time exposure neuromuscular dynamics and vestibular level sensed illness ratings 3) Impulse dynamics effect between the neuromuscular (EMG) and the vestibular dynamics (IR) The results have showed that: a) The vibration exposure condition: When the total RMS acceleration frequency weighted average IR increases, the EMG average total power increases too by driving the classical motion cueing algorithm. However, in contrast to this, the EMG average RMS total power decreases while the IR increases during the LQR motion cueing algorithm. b) Impulse effect condition: As the IR augments; the EMG average RMS total power also increases for the optimal motion cueing algorithm but it decreases for the classical algorithm.

The Influence of the feedback control of the hexapod platform of the SAAM dynamic driving simulator on neuromuscular dynamics of the drivers

Abstract: Multi sensorial cues (visual, auditory, haptic, inertial, vestibular, neuromuscular) [Ang2] play important roles to represent a proper sensation (objectively) and so a perception (subjectively as cognition) in driving simulators. Driving simulator aims at giving the sensation of driving as in a real case. For a similar situation, the driver has to react in the same way as in reality in terms of 'self motion'. To enable this behavior, the driving simulator must enhance the virtual immersion of the subject in the driving situation. The subject has to perceive the motion of his own body in the virtual scene of the virtual car as he will have in a real car. For that reason, restituting the inertial cues on driving simulators is essential to acquire a more proper functioning [Kol20]. Simulation sickness has been one of the main research topics for the driving simulators. It was assessed between dynamic and static simulators [Cur7], [Wat32]). For a braking maneuver, [Sie29] stated that if the motion platform is activated the bias in reaching increased levels of decelerations was reduced strongly comparing to inactivated platform case. However, there has been lack in publications of vehicle-vestibular cue conflict based illness rating approach and its correlation with the neuromuscular dynamics for that kind of research. In order to reduce the simulator sickness, the difference between the accelerations through the visual and the vestibular cues have to be minimized (cost function minimization via model reference adaptive control, in this paper). Because of the fact that, this paper addresses the simulator motion sickness as a correlated function of this deviation for the both cues with the perception questionnaires as well as the EMG analysis results for the subjects who joined in those experiments. Due to the restricted workspace, it is not possible to represent the vehicle dynamics continuously with scale 1 to 1 on the motion platform [Moo22]. Nevertheless, the most desired aim is to minimize the deviation of the sensed accelerations between the represented dynamics as realistic as possible depending on the driving task. The perception of the driving is very difficult to evaluate in that context. This is the reason; the motion sickness is not easy to study as well. This research work has been performed under the dynamic operations of the SAAM driving simulator as an open-loop and a closed-loop controlled tracking of the hexapod platform of the SAAM dynamic driving simulator. It is obvious that inertial restitution addresses a significant role to maintain a developed fidelity of the driver behaviors on diving simulators The dynamic simulators are being used since the mid 1960s (Stewart platform) [Ste1] firstly for the flight simulators, then the use has spread to the automotive applications. The utilization scope diversifies from driver training to research purposes such as; vehicle dynamics control, advanced driving assistance systems (ADAS), motion and simulator sickness, etc. The dynamic driving simulator SAAM (Simulateur Automobile Arts et Metiers) involves a 6 DOF (degree of freedom) motion system. It acts around a RENAULT Twingo 2 cabin with the original control instruments (gas, brake pedals, steering wheel). The visual system is realized by an approximate 150° cylindrical view. Within the cabin for the employment of extensive measuring techniques (XSens motion tracker, and Biopac EMG (electromyography) device [Acq10]) prepared, which have been already used with numerous attempts such as sinus steer test, NATO chicane, etc. The visual accelerations of translations (longitudinal X, lateral Y and vertical Z axes) as well as the visual accelerations of roll and pitch, which correspond to the vehicle dynamics, were taken into account for the control. Then the platform positions, velocities and accelerations were controlled and fed back to minimize the conflict between the vehicle and the platform levels. The research question about this paper explains a comparative study between an open and a closed loop controlled platform in order to determine the spent power by the muscles to maintain the vehicle pursuing among the pylons. For the evaluation and the validation procedure [Kim19], [Wat32], [Rey27], [Kem17], [Che5], [Pic25], [Acq1], the scenario driven on the simulator SAAM with an open and a closed loop controlled platform to describe the impact of the feedback control. Some results from a case study will be illustrated in the scope of this research with real time controls of the platform at a longitudinal velocity of 60 km/h. The results of this study will be discussed also statistically to obtain the distribution of the dynamics behavior for a group of the participants. This research has been undertaken at ENSAM Institut Image, in collaboration with RENAULT.

Motion parallax in immersive cylindrical display systems

Abstract: Motion parallax is a crucial visual cue produced by translations of the observer for the perception of depth and selfmotion. Therefore, tracking the observer viewpoint has become inevitable in immersive virtual (VR) reality systems (cylindrical screens, CAVE, head mounted displays) used e.g. in automotive industry (style reviews, architecture design, ergonomics studies) or in scientific studies of visual perception. The perception of a stable and rigid world requires that this visual cue be coherent with other extra-retinal (e.g. vestibular, kinesthetic) cues signaling ego-motion. Although world stability is never questioned in real world, rendering head coupled viewpoint in VR can lead to the perception of an illusory perception of unstable environments, unless a non-unity scale factor is applied on recorded head movements. Besides, cylindrical screens are usually used with static observers due to image distortions when rendering image for viewpoints different from a sweet spot. We developed a technique to compensate in real-time these non-linear visual distortions, in an industrial VR setup, based on a cylindrical screen projection system. Additionally, to evaluate the amount of discrepancies tolerated without perceptual distortions between visual and extraretinal cues, a "motion parallax gain" between the velocity of the observer's head and that of the virtual camera was introduced in this system. The influence of this artificial gain was measured on the gait stability of free-standing participants. Results indicate that, below unity, gains significantly alter postural control. Conversely, the influence of higher gains remains limited, suggesting a certain tolerance of observers to these conditions. Parallax gain amplification is therefore proposed as a possible solution to provide a wider exploration of space to users of immersive virtual reality systems.

Toward predicting the subjective assessment of ESC in a driving simulator

Abstract: Previous works have sought to develop an evaluation method to describe loss of adherence episodes by means of subjective indicators and to propose a predictive model of the subjective evaluations. This study presents the use of the presented model and predicted evaluations when ESC is triggered. The results confirm the capabilities of the model and the potential of the presented methodology to evaluate and characterize ESC on a driving simulator in the early stages of the engineering design. However further studies will be necessary to improve the robustness of the model and its use in various situations.

Predicting the Subjective Evaluation of Drivers in a Driving Simulator During Loss of Adherence

Abstract: This paper investigates the possibility of using objective indicators to predict the subjective evaluation of a driver in a simulator. Situations of loss of adherence (LOA) were controlled and modulated in intensity and duration on a static and on a dynamic simulator (with and without a motion base). Multiple regression analyses were performed using the subjective evaluation of participants as the dependent variable, and the objective physical variables of the interaction driver/vehicle as the independent variables. The results assigned the most contributive variables to the accuracy of the model’s prediction for each subjective item, lateral acceleration and yaw rate for “control feeling” for instance. They also underlined the consistency of our approach and the influence of motion rendering on the perception of LOA intensity. A similar method could be used to evaluate the perception of various configurations of electronic stability control (ESC) systems.