Ústav automatizace a měřicí techniky
Browse
Recent Submissions
- ItemEvaluating the Effect of Fatigue on Driver’s Performance(Elsevier, 2024-06-19) Jirgl, Miroslav; Šedivá, Soňa; Bradáč, ZdeněkThe paper presents an experimental cybernetic approach for the evaluation of drivers’ performance in connection with fatigue based on mathematical modelling and description of human behavior during a simple lane-changing task. Data are acquired via the in-house developed car driving simulator under defined conditions. Statistical evaluation of the used parameters for 50 active drivers is presented as an initial data set for the determination of the described drivers’ performance decision-making system built on fuzzy logic. The system was then used for calculating the DPI (Driver performance index) as a self-determined relative measure of drivers’ performance. Then, an example of applying the method with three volunteers influenced by fatigue is presented. All of these participants reached lower DPI in the case of fatigue induced by a 3-hour-long simulated loading drive on a highway. Although the results cannot be statistically significantly interpreted, they indicate a potential for further and deeper research.
- ItemFuzzy Logic-based Techniques for Human Driver Behavior Modelling during a Simple Lane-changing Task(Elsevier, 2024-06-19) Jirgl, Miroslav; Mesárošová, Michaela; Fiedler, Petr; Arm, JakubThe paper presents an experiment with fuzzy-logic-based models for approximating data representing human driver behavior during a simple lane-changing task. Data are acquired via the self-developed car driving simulator under defined conditions. Three fuzzy-logic-based models are considered: fuzzy Hammerstein model, fuzzy PD controller with a fixed structure (analogy to the linear PD controller) and fuzzy PD controller with an optimized structure. A comparison of the effectiveness of the individual approaches was evaluated using a criterion representing the fitness of the model output and the measured data. Although only one dataset was considered in the frame of this paper, the results indicate an investigative potential of the fuzzy models for this kind of task connected with the modelling of human behavior.
- ItemA Simple State-Space Model of Human Driver Applicable to Windy Conditions(Elsevier, 2024-08-14) Čelko, Jakub; Mihálik, Ondrej; Husák, Michal; Bradáč, ZdeněkThe paper is concerned with the design, verification and evaluation of a car-driving test scenario for human driver assessment. The scenario implemented in Unreal Engine adds four different wind characteristics which disturb the motion of a simulated vehicle. Besides, the driver is instructed to change the driving lane at defined intervals. These forcing functions enable the identification of the human-machine loop using state-space models. The parameters characterising the human dynamics are extracted from the model of the whole loop. As opposed to rather obsolete McRuer models, this approach follows the recent trends in the modelling of human-machine systems as multiloop systems or quadratically optimal controllers. Our results suggest that the model relying on a single transfer function with 4 parameters loses prediction capabilities during more realistic scenarios, in which random disturbances, such as wind gusts, affect the vehicle. In such cases, the multiloop model with the same number of parameters is able to capture human behaviour more accurately than McRuer model.
- ItemCNN Architecture for Posture Classification on Small Data(Elsevier, 2024-08-14) Mesárošová, Michaela; Mihálik, Ondrej; Jirgl, MiroslavA convolutional neural network is often mentioned as one of the deep learning methods that requires a large amount of training data. Questioning this belief, this paper explores the applicability of classification based on a shallow net structure trained on a small data set in the~context of patient posture classification based on data from a pressure mattress. Designing a CNN often presents a complex problem, especially without a universally applicable approach, allowing many diverse structural possibilities and training settings. We tested various training options and layer configurations to provide an overview of influential parameters for posture classification. Experiments show encouraging results with the leave-one-out cross-validation accuracy of 93.1% of one of the evaluated CNN structures and its hyperparameter settings.
- ItemDigital twin of heat exchange station(Elsevier, 2024-08-14) Benešl, Tomáš; Husák, Michal; Mihálik, Ondrej; Vancl, Radim; Bradáč, ZdeněkThe article describes the identification of a heat exchanger system using PLC for measurement, followed by system model identification and reconstruction in Matlab Simulink and SIMIT. The root-mean-square error of the identified system was 2.49 % on training data and 5.33 % in the validation scenario. The system model is suitable for predicting and simulating errors and anomalies; it can be used for the optimisation of control processes or system parameters. Therefore, the model can serve as an educational tool, allowing training of students, operators and other experts. They can experiment without fear of damaging components and without having to wait for the long-time transients of a real system.