Emotion recognition technology have been become a very important factor for humans to communicate with computers or robots. This technology can be used for various interaction fields such as human-computer interaction and human-robot interaction. Recently, researches on emotion recognition using machine learning technology have been actively studied. Among them, we focus a method of emotion feature extraction through learning by using Neural Network to solve the problem that criterion of emotion classification was vague. It stem from that all the features of biosignals emotions can not sufficiently contain emotion feature information. Generally, various biometricsensors can be used to recognize human emotions. In this paper, we measured two vital signs, PPG(Photo plethysmo graphy) and EMG(Eletro myo gram), to recognize emotions from two biometricsensors. The PPG sensor was set on the index finger and the EMG sensor was attached to the Trapezius of the shoulder to measure the vital signs. Because emotion of human is a subjective indicator, it is difficult to represent them to the figures as an objective indicator. In this paper, we devide the emotions into three categories: Happy, Sad, and Neutral states. The classification was conduted by the neural network with multilayer perceptron. The vital signs were measured while watching music videos that could induce emotions in a blocking external factors. Measured vital signs were formulated as a training dataset, testing dataset and used as an input of the multilayer perceptron. As a result, it was confirmed through experiments that the three emotions are classified into good performance.
Yun-Kyu Lee is currently working toward his M.S. degree in Electrical Engineering from Korea University.
The paper presents the results of a theoretical and experimental study on the optimal placement of a welding task in the work area of a 6DOF series welding robot OTC AX-MV6L. The robotic task consists in a complex trajectory required for development of welded corner joints of a metal side frame, used for industrial floor enclosures. Starting from the technical data of the robotic cell for MIG welding, the modeling of studied robot precision by using homogeneous operator method was done and robot precision maps were built. Based on these, high precision areas were identified and optimal side frame placement in the robot workspace was investigated. The results of the study are exploited by optimizing the welding process at ELDON LTD.
Mircea Neagoe has completed his PhD from Transilvania University of Brasov, Romania in 2001. He is professor at Transilvania University of Brasov, Faculty of Product Design and Environment, Department of Product Design, Mechatronics and Environment. He has published more than 200 papers in journals and conference proccedings, and has contributed to more than 10 patents and patent pendings
The paper presents the development of a mechatronic fixing device prototype which allows the fixing adjustment of a planar frame composed by flexible metalic profiles to be welded in a robotic welding cell. The mechatronic device integrates a laser sensor, able to mesure with high precision a large range of lenghts, aiming at delivering useful data for the control system supporting the actuation system based on linear actuators. Different fixing scenarios were tested and the influence of the fixing strategies on the welded frame quality was established using the proposed mechatronic prototype. The obtained results will be used to optimize the robotized welding process at ELDON LTD
Nadia Cretescu has completed her PhD from Transilvania University of Brasov, Romania in 2011 and she graduated M.Sc in Robotics (2003) and B.Sc. in Industrial robots (2002). She is lecturer at Transilvania University of Brasov, Faculty of Product Design and Environment, Department of Product Design, Mechatronics and Environment. She has published more than 30 papers in journals and conference proccedings.
The paper deals with the optimal sequence of MIG welding process used to assemby the side frame components for industrial standing metalic enclosures. The planar metalic frame is composed by two vertical tall profiles and other two horizontal profiles, obtained from small thickness metal sheet, linked throught a small shaped piece in each corner. Exterior and interior welds are required to obtain a rigid joint of the components. The welding sequence on the four corners is analysed aiming at optimising both the total processing time and local temperature. Theoretical studies and experimental reasearch are done to find the optimal solution for the robotized task implemented by the OTC AX-MV6L welding robot. The results of the study are useful for optimizing the welding process at ELDON LTD.
Engineer Radu Saulescu has completed his PhD from Transilvania University of Brasov, Romania. He is the Associate Professor at the Faculty of Product Design and Environment at Trasilvania University of Brasov. He has experience in conceptual design, development and optimization of new mechanisms, being author / coauthor of more than 150 scientific papers in the field, out of which 17 in ISI indexed journals, 20 in ISI proceedings, as well as 8 patents. He is the Winner of Festo young Researcher and Scientist Support Scholarship Award. Austria, Vienna 2008. He is member Romanian Association of Mechanical Transmissions and Mechanisms – subsidiary IFToMM.
In general terms, the three principal duties of an enterprise company are to design new products, manufacture, and sell them. Thus, the first step is critical; the conceptualization stages can serve to evaluate if the design is technically feasible regarding requirements and manufacturability. A reconfigurable manufacturing system (RMS) is a convenient manner to produce customizable goods without compromising the production volume or cost. The Integrated Product, Process, and Manufacturing System Development Processes (IPPMD) is a modeling framework where all the product life stages are mapped and depicted in a concurrent engineering process, and map the activities of cross functional teams during product development. In this research IPPMD is used to design a reconfigurable exoskeleton, hence, giving the opportunity to develop a unified framework for customization and reconfiguration. To accomplish exoskeleton’s functionality, the functional analysis and concept synthesis are driven from different domains such as customers, functions, modules and reconfiguration domain. To this end, the modularity is selected as the fundamental property to construct a reconfigurable product. An evaluation of the feasibility is done in the reconfiguration domain, as well as in RMS metrics. The results show that the unified framework can be useful during the complex conceptualization as for exoskeletons.
Coordinate measuring machines (CMMs) are massively exploited as measuring tools in the modern manufacturing industry. The performance of these machines, in terms of accuracy, have been greatly improved in recent years by using quasi-static errors compensation. However, the demand for shorter measurement cycle times, CMMs are required to be used at high measuring velocity. In such conditions, dynamic errors will have a great impact on the measuring accuracy. Consequently, dynamic errors assessment, modelling and compensation are needed to improve the overall CMM performances. In this paper, a comprehensive predictive modelling strategy developed and applied successfully for dynamic errors compensation is presented. The main measuring parameters that have an influence on the CMM dynamic performance are identified and used in an extensive experimental investigation. The positioning accuracy under different dynamic conditions using a high-precision laser interferometer is designed and evaluated. Using the experimental results, multiple neural networks models are built according to a structured modelling procedure based on Taguchi method. Improved statistical tools and objective criteria are used to extract the most appropriate variables and conditions leading to the best predictive modelling performance. The resulting models are implemented on a moving bridge CMM to compensate for both geometric and dynamic errors. The results demonstrate a reduction of more than 80% of dynamic errors. The achieved results proves that the real-time dynamic errors compensation is achievable in high-speed measurement conditions leading to shorter measurement cycle times while maintaining high measuring accuracy.
The lack of information about the environment suggests probabilistic estimates in the process of elaborating conditioned reflexes. On the basis of the statistical criterion χ2, a computer model of learning is developed for any combinations of the parameters of a probability-organized environment: the probability of accidental performance of the cultivated act, the probability of reinforcement in connection with the conditioned stimulus, obtaining reinforcement outside the connection with the conditioned stimulus. Assigned as an independent parameter, the significance level α (the first kind of error) integrally characterizes the ability to make decisions under uncertainty. Heuristic modeling does not involve the consideration of already known patterns of learning and, at the same time, completely reproduces them. This indicates that the model, firstly, successfully imitates the real process of learning biological objects in a probability-organized environment and, secondly, can be used in works on artificial intelligence.