Search

This paper presents solutions for methodological issues that can occur when obtaining face recognition and object detection for a ROS-based (Robot Operating System) open-source platform. Ubuntu 18.04, ROS Melodic and Google TensorFlow 1.14 are used in programming the software environment. TurtleBot2 (Kobuki) mobile robot with additional onboard sensors are used to conduct the experiments. Entire system configurations and specific hardware modifications that were proved mandatory to make out the system functionality are also clarified. Coding (e.g., Python) and sensors installations are detailed both in onboard and remote laptop computers. In experiments, TensorFlow face recognition and object detection are examined by using the TurtleBot2 robot. Results show how objects and faces were detected when the robot is navigating in the previously 2D mapped indoor environment. acceptedVersion

Safety training has been identified as one of the most important safety management practices. Previous research has suggested several models for evaluating the effectiveness of safety training. Evaluation is crucial to implement evidence-based training interventions. However, the outcomes of safety training are difficult to demonstrate. Studies better specifying and measuring the outcomes of training are needed. The aim of this study was to design and test a model for evaluating the outcomes of safety training. The model addressed 1) safety knowledge, 2) safety attitudes, beliefs and motivation, 3) safety behavior and 4) safety performance. A survey (n = 25) was used to measure the employees’ perceptions after the safety training. Based on the results, some improvements in safety knowledge, attitudes, behavior and safety performance can be found. The model was found as suitable for its purpose. The results can be utilized in identifying factors related to the effectiveness of safety training. acceptedVersion

In the safety-critical process industry, safety competence is emphasized. It is essential to ensure that process operator students adopt good safety competence during their studies. Due to Finnish vocational education and training (VET) reform, on-the-job learning has increased, inducing a need for new methods. In this study, an operational model for developing process operator students’ safety competence in on-the-job learning was constructed and evaluated. The model was compiled in cooperation with six process industry companies, five VET organizations, and eight expert organizations. The main data collection methods were interviews and workshops. The model consists of eight phases and provides instructions, tools, and good practices for implementing on-the-job learning. Companies and VET organizations considered the model useful and identified several purposes of use for the model. acceptedVersion

Organizations have several indicators for their safety performance. However, the use of performance indicators often fails to create overall insights on the level of safety and the various factors affecting it. Performance indicators could be better utilized in safety-related decision-making. Maturity models have been presented in many different managerial fields, but no such models for safety performance measurement can be identified. Maturity analysis can provide information on why performance measurement utilization is flawed and how can it be improved. The aim of this paper is to design and test a maturity model for safety performance measurement. The study presents an approach for evaluating maturity which combines written descriptions of best practices, the overall satisfaction of employees in the evaluated aspects, and the experienced level of safety performance. acceptedVersion

acceptedVersion

As research into autonomous vehicles gets mainstream, automobile manufacturers are trying to reinvent the ways we operate and interact with our vehicles. This is now more evident in the central instrument clusters than ever before. While some manufacturers are focusing on adding touchscreens to replace most in-vehicular infotainment (IVI) controls, others are trying to make the IVIS smarter and multimodal. Unfortunately, the lack of reliable tactile feedback for touchscreen interaction in IVI systems can be a major issue. Although this may not be a critical flaw in mobile device interaction, it can be a dangerous limitation for in-car systems where visual distraction can be fatal. For this reason, our research* is focused on exploring and developing new methods of providing tactile actuation using both vibrotactile and pneumatic feedback techniques.