This study aimed to design a deep neural network for electroencephalography (EEG)-based drowsiness detection in multiple consciousness states, i.e., “awake,” “sleep,” and “drowsiness.” Few studies have seriously considered the optimal input vector size or labeling method in classifying multiple consciousness states, which may affect classification performance. To determine the optimal input vector length, i.e., window length, three neural network models (long short-term memory [LSTM], convolutional neural network [CNN], and combined LSTM and CNN) and four feature-based models were tested with six different levels of window length. The EEG dataset was acquired from 19 participants with randomly assigned auditory stimuli and button responses. The EEG data were labeled into three classes (awake, sleep, and drowsiness) based on the defined button response pattern corresponding to the stimuli. The results demonstrated that when the input vector size exceeded 8 sec, the performance of the neural network models dropped rapidly; however, when the window size was less than 8 sec, the performance change according to the window size was small. In contrast, the performance of feature-based models increased continuously as the window size increased. The LSTM model yielded the best accuracy (86%) for a 1 sec window length, and the LSTM-CNN model yielded the best kappa index (0.77) for a 4 sec window length. In addition, the proposed model was applied to the binary classification of normal consciousness (awake) and low consciousness (drowsiness and sleep) states to determine whether this model works appropriately in actual applications such as drowsiness detection in a driving environment. For binary classification, the LSTM-CNN model resulted in 0.95 F1 scores in 4000-ms. When a short input data (500 msec) is used, the LSTM-CNN model resulted in an average accuracy of 85.6% and a kappa index of 0.77 for the three-class classification problem and 0.94 F1 scores for the binary classification problem. In conclusion, we demonstrated that the proposed model could effectively detect drowsiness. Furthermore, a significant correlation was found between reaction time and drowsiness. However, using the reaction time as an index for labeling drowsiness was challenging because of the high false-negative ratio.

Near-Infrared spectroscopy (NIRS) is an emerging non-invasive brain computer interface (BCI) modality that measures changes in haemoglobin concentrations in the cortical tissue. To date most NIRS studies have used standard multiple subject/session dependent classifiers for neural signal decoding. Such approach is preferable to avoid large degree of variabilities in the acquired data that affects classifier generalization. This study presents a classification algorithm that maintains a good performance under the presence of variability in the NIRS data. It is based on ν- support vector machines and its extensions to a multiple kernel learning framework. Empirical evaluations have shown that through the proposed method one can improve the overall BCI decoding accuracy, and its robustness against the variability in neural data.

Various robot platforms have been designed and developed to perform given tasks in a hazardous environment for surveillance, reconnaissance, search and rescue, etc. We considered a terrain-adaptive and transformable hybrid robot platform that is equipped for rapid navigation on flat floors and good performance in overcoming stairs or obstacles. The mode transition is determined and implemented by adaptive driving mode control of the mobile robot. The terrain-adaptive and user-friendly remote control was verified through its navigation performance experiments in real and test-bed environments.

Robot applications with rescue operations have been widely developed for disaster prevention purposes such as fire-fighting robot, rescue robot, and surveillance robot. This paper proposes two novel wireless remote controllers which offer not only user friendly operating interface but also easy robot controls. The handheld type controller is able to be easily carried by fire fighters and the joystick based controller enables more accurate and rapid controls for specified tasks in emergence environment. Both controllers are designed to be capable of gathering environment information such as video data, temperature, gas concentration and etc. Through the performance evaluation and questionnaire survey with a mobile rescue robot system, we have analyzed their efficiency considering fire fighter's usability. Then by accepting the opinions of fire fighters, we suggest applicable scenarios for both controllers in expected fire sites.

Robot technology is emerging for applications in disaster prevention with devices such as fire-fighting robots, rescue robots, and surveillance robots. In this paper, we suggest an portable fire evacuation guide robot system that can be thrown into a fire site to gather environmental information, search displaced people, and evacuate them from the fire site. This spool-like small and light mobile robot can be easily carried and remotely controlled by means of a laptop-sized tele-operator. It contains the following functional units: a camera to capture the fire site; sensors to gather temperature data, CO gas, and O2concentrations; and a microphone with speaker for emergency voice communications between firefighter and victims. The robot's design gives its high-temperature protection, excellent waterproofing, and high impact resistance. Laboratory tests were performed for evaluating the performance of the proposed evacuation guide robot system.

Fuzzy algorithms provide intuitive method for robot obstacle avoidance. Fuzzy controllers incorporating a design based on lookup tables (LUT) enable faster obstacle avoidance in environment with multiple obstacles. In an earlier study, we introduced a full LUT-based architecture for an 18-rule positive/negative (P/N) fuzzy controller. In this study, the number of fuzzy rules is expanded to 50. Because of the extra rules, the controller apparently needs more LUT(s) buffers. In other words, the buffer size increases with the complexity of the fuzzy controller. Therefore, we propose a LUT sharing method to reduce the buffer size without significantly degrading the performance of the controller. The final objective of this work is to design a LUT-based fuzzy controller whose buffer size is independent of the complexity of the fuzzy system. The proposed method is evaluated by simulating a 50-rule P/N fuzzy controller using Microsoft Robotics Developer Studio (MSRDS). The simulation results show that in comparison with the method not using LUT(s), full LUT-based method and the LUT sharing method reduce the operational time by nearly 80% and 70%, respectively. Although the LUT sharing method needs 1.5 times more operational time than the full LUT methods, it reduces the buffer size by more than 90%.

This paper is aimed at developing the control method of multiple degree-of-freedom (DOF) hybrid haptic interfaces with active/passive actuators. After the mathematical description of the kinematic requirements of multiple DOF haptic interfaces with passive actuators, we outline the concept of the control method based on the passivity. The fundamental consideration of the passivity is examined by following the "power observer" (PO) approach. The "switching control" (SC) is outlined briefly and then proposed the "bridging control" (BC) comparing with SC. To evaluate the feasibility of the proposed BC/PO, the experiments were conducted via 2-DOF hybrid haptic interface which is a five-bar-linkage mechanism with two motors and two MR brakes.

Physical damping is essential to achieve passivity in haptic display. To improve stability, an additional physical damper was used with motor. This fixed damping approach, however, has limitation to display free motion, so that the motor actually has to work for damping cancellation. This paper suggests that the controllable damping approach without necessity for damping cancellation to improve both stability and performance. Firstly, we analyze the effect of the controllable physical damping on stability and also propose a controllable physical damping approach applying controllable magnetorheological brake to haptic display. Finally, we experimentally show the contribution of the proposed controllable physical damping approach to the performance in terms of Z-width.

This paper describes the design and implementation of a hybrid active/passive force feedback device. It has a single degree of freedom that is actuated by a motor and a brake pair. The use of a motor and a brake allows various objects to be simulated without a stability problem and related safety issues involved with using high torque motors only. The device performance is measured by its ability to simulate various test objects. Simple test objects are modeled as a benchmark test of the system's performance. The force feedback device is capable of simulating forces in a variety of virtual environments. This device demonstrates the effectiveness of a hybrid active/passive haptic actuator. An analysis of the use of hybrid active/passive actuators in the 2 degrees of freedom case finds that they are capable of simulating a rigid body in a broad range.