In 2020, the autonomous underwater vehicle (AUV) has already become a vital part of deep-sea research. There is a long history of R&D of AUVs that dive into the deep sea, where radio waves cannot reach, thus making remote control difficult so that no help can be provided, which implies that careful and adequate preparation is necessary. Their successful development has been based on the accumulation of experience and achievements contributing to the remarkable results that no other system can produce. The aggressive R&D of Japanese AUVs started approximately 40 years ago. This paper looks back at this history and introduces various Japanese AUVs.

Brain machine/computer interface (BMI/BCI) technologies are based on analyzing brain activity to control machines and support the communication of commands and messages. To sense brain activities, a functional NIRS and electroencephalogram (EEG) that has been developed for that purpose is often employed. Analysis techniques and algorithms for the NIRS and EEG signals have also been created, and human support systems in the form of BMI/BCI applications have been developed. In the field of rehabilitation, BMI/BCI is used to control environment control systems and electric wheelchairs. In medicine, BMI/BCI is used to assist in communications for patient support. In industry, BMI/BCI is used to analyze sensibility and develop novel games.

This special issue on Brain Machine/Computer Interface and its Application includes six interesting papers that cover the following topics: an EEG analysis method for human-wants detection, cognitive function using EEG analysis, auditory P300 detection, a wheelchair control BCI using SSVEP, a drone control BMI based on SSVEP that uses deep learning, and an improved CMAC model.

We thank all authors and reviewers of the papers and the Editorial Board of Journal of Robotics and Mechatronics for its help with this special issue.

We propose a method to detect human wants by using an electroencephalogram (EEG) test and specifying brain activity sensing positions. EEG signals can be analyzed by using various techniques. Recently, convolutional neural networks (CNNs) have been employed to analyze EEG signals, and these analyses have produced excellent results. Therefore, this paper employs CNN to extract EEG features. Also, support vector machines (SVMs) have shown good results for EEG pattern classification. This paper employs SVMs to classify the human cognition into “wants,” “not wants,” and “other feelings.” In EEG measurements, the electrical activity of the brain is recorded using electrodes placed on the scalp. The sensing positions are related to the frontal cortex and/or temporal cortex activities although the mechanism to create wants is not clear. To specify the sensing positions and detect human wants, we conducted experiments using real EEG data. We confirmed that the mean and standard deviation values of the detection accuracy rate were 99.4% and 0.58%, respectively, when the target sensing positions were related to the frontal and temporal cortex activities. These results prove that both the frontal and temporal cortex activities are relevant for creating wants in the human brain, and that CNN and SVM are effective for the detection of human wants.

Brain-computer interface (BCI) enables us to interact with the external world via electroencephalography (EEG) signals. Recently, deep learning methods have been applied to the BCI to reduce the time required for recording training data. However, more evidence is required due to lack of comparison. To reveal more evidence, this study proposed a deep learning method named time-wise convolutional neural network (TWCNN), which was applied to a BCI dataset. In the evaluation, EEG data from a subject was classified utilizing previously recorded EEG data from other subjects. As a result, TWCNN showed the highest accuracy, which was significantly higher than the typically used classifier. The results suggest that the deep learning method may be useful to reduce the recording time of training data.

This study considers a brain-machine interface (BMI) system based on the steady state visually evoked potential (SSVEP) for controlling quadcopters using electroencephalography (EEG) signals. An EEG channel with a single dry electrode, i.e., without conductive gel or paste, was utilized to minimize the load on users. Convolutional neural network (CNN) and long short-term memory (LSTM) models, both of which have received significant research attention, were used to classify the EEG data obtained for flickers from multi-flicker screens at five different frequencies, with each flicker corresponding to a drone movement, viz., takeoff, forward and sideways movements, and landing. The subjects of the experiment were seven healthy men. Results indicate a high accuracy of 97% with the LSTM model for a 2 s segment used as the unit of processing. High accuracy of 93% for 0.5 s segment as a unit of processing can remain in the LSTM classification, consequently decreasing the delay of the system that may be required for safety reasons in real-time applications. A system demonstration was undertaken with 2 out of 7 subjects controlling the quadcopter and monitoring movements such as takeoff, forward motion, and landing, which showed a success rate of 90% on average.

Neural networks (NNs) are effective for the learning of nonlinear systems, and thus they achieve satisfactory results in various fields. However, they require significant amount of training data and learning time. Notably, the cerebellar model articulation controller (CMAC), which is modeled after the cerebellar neural transmission system, proposed by Albus can effectively reduce learning time, compared with NNs. The CMAC model is often used to learn nonlinear systems that have continuously changing outputs, i.e., regression problems. However, the structure of the CMAC model must be expanded to apply it to classification problems as well. Additionally, the CMAC model finds it difficult to simultaneously classify categories and estimate their proportional linear measure because designated learning algorithms are required for both regression and classification problems. Therefore, we aim to build a composite-type CMAC model that combines classification and regression algorithms to simultaneously classify categories and estimate their proportional linear measures.

The occurrence of serious traffic accidents due to driving error has recently become a social issue. The decline in cognitive functions is considered a factor particularly among elderly drivers, for which some corrective measures are urgently needed. Currently, as a measure dealing with elderly drivers, drivers aged 75 years are required by law to examine their cognitive functions when they renew their driver’s license. This examination is conducted to measure memory and power of judgment. In the present study, we used a device that allows simple measurements and attached electrodes to the frontal pole, where it is easy to attach; measure electroencephalograms for the cognitive functions of memory, mental calculations (thinking), and memory recall; and examine the changes in the power spectra to determine how they vary in young and elderly individuals.

Having the capability to control a wheelchair using brain signals would be a major benefit to patients suffering from motor disabling diseases. However, one major challenge such systems are facing is the amount of input needed over time by the patient for control. Such a navigation control system results in a significant mental burden for the patient. The objective of this study is to develop a BCI system that requires a low number of inputs from a subject to operate. We propose an autonomous wheelchair that uses steady-state visual evoked potential based brain computer interfaces to achieve the objective. A dual mode system was implemented in this study to allow the autonomous wheelchair to work in both unknown and known environments. Robot operating system is used as the middleware in this study for the development of the algorithm to operate the wheelchair. The mental task for the subject using this wheelchair is reduced by relegating the responsibility of navigation control from the subject to the navigation software.

Educational support robots have been the focus of study in recent years. Studies have reported that robots providing educational support, based on cognitive apprenticeship theory, provided learners with effective collaborative learning. However, the robots were remote controlled, so no behavioral model was constructed of robots operating autonomously to provide educational support. Therefore, in this paper, we construct a behavioral model in which robots autonomously provide educational support based on cognitive apprenticeship theory. In addition, through a comparative experiment with a behavioral model providing educational support in accordance with learner requests, which is a conventional technique, we verify the learning effects of this behavioral model on university students.

This paper describes about improvements in the mobility of a stair climbable mobile robot on a step-field. A step-field, standardized by the National Institute of Standards and Technology (NIST), is a simulated artificial rough terrain used for testing rescue robots. Its use was popularized worldwide in the RoboCup Rescue contest, but a method for evaluating test results from it has yet to be established sufficiently. The author of this paper attempts to evaluate the mobility of stair climbable mobile robots on step-fields. A novel method using the 2 parameters of yaw-angle error and lateral position error is proposed. Two actual robots, SMART-IV type A and SMART-IV type B, are constructed and evaluated on small step-fields. The test results indicate that both robots can traverse the step-field under some appropriate conditions, but a comparison of the results indicates that the proposed design changes may be declared an improvement of the robot.

In Japan, the deterioration of industrial plants built during the period of high economic growth in the middle of the 20th century has recently become a social concern. Corrosion under insulation (CUI) of piping in such plants is a pressing problem. X-ray and ultrasound inspections are conventional methods for detecting CUI; however, these methods are time-consuming and expensive. Therefore, rapid and low-cost screening techniques for CUI are required. We develop a hammering-type inspection robot system that moves inside the piping and records hammering sounds. Furthermore, we propose an acoustic analysis method to identify anomalous parts from the hammering sound using machine learning techniques. Using three testing pipes, we can successfully identify anomalous parts through acoustic analysis using a deep neural network as a supervised learning method. However, in practical piping inspections, the detection of anomalies without training data is required for further applications. Therefore, we investigate unsupervised learning anomaly detection using an autoencoder and a variational autoencoder and report the results.

Parkinson’s disease (PD) is a common progressive neurodegenerative disease that affects a wide range of motor and non-motor symptoms. Freezing of gait (FOG) is such a motor symptom of PD that frequently results in falling, and almost half of PD patients suffer from FOG. In this study, we investigated the effectiveness of a robotic assistance system called UPS-PD, which was developed to suppress FOG. The double limb support phase (DLS) in a 10-m straight-line walking task, the gait time and step counts were measured in five PD subjects. In addition, the safety of the UPS-PD in a healthy person was investigated using OpenSim, and the DLS parameters in four healthy elderly subjects were evaluated. In the experiment with the PD patients, the DLS parameters of two subjects showed an improvement. Furthermore, the step length of one subject and the step length and walking speed of the other subject were improved. Moreover, there were no problems in terms of instability of gait in both the PD patients. The UPS-PD did not adversely affect the gait of healthy elderly subjects and the walking of a healthy subject model in the simulation. Therefore, the UPS-PD is considered to be a useful device for improving walking in PD patients.

In general, legged robots are designed to walk with a fixed rhythmic pattern. However, most animals can adapt their limb movements while walking. It is necessary to understand the mechanism of adaptability during locomotion when designing bio-inspired legged robots. In this paper, we propose an approach to analyze the flexible locomotion pattern of animals using a polar histogram. Field crickets were used to investigate variations in leg movement of insects depending on the environment. Crickets have a tripod gait; however, their leg movement changes depending on the texture of the ground. There was a significant difference between the leg movement when walking and when swimming. Our approach can explain how animals move their legs during locomotion. This study is useful for evaluating the movements of legged robots.

This paper presents a method for analyzing the throwing motion of a yo-yo based on an integrated model of a yo-yo and a manipulator. Our previous integrated model was developed by constraining a model of a white painted commercial yo-yo and a model of a plain single-link manipulator with certain constraining conditions placed between two models. However, for the yo-yo model, the collisions between the string and the axle of the yo-yo were not taken into account. To avoid this problem, we estimate some of the yo-yo parameters from the experiments, thereby preserving the functionality of the model. By applying the new integrated model with the identified parameters, we analyze the throwing motion of the yo-yo through numerical simulations. The results of which show the ranges of the release angle and the angular velocity of the joint of the manipulator during a successful throw. In conclusion, the proposed analysis method is effective in analyzing the throwing motion of a manipulator.

Small and medium-sized hydroelectric power plants are scattered in mountainous areas of Japan. Many tunnels that have been constructed for the purpose of introducing water have been in operation for decades, and inspections to aging deterioration are indispensable, however checks and maintenance work within the tunnels where the ceiling is low and water is flowing is very burdensome. This research aims at labor saving of visual check work of an inspector who moves through a narrow tunnel and searches for a deformed portion, and a simple imaging support device with a camera on a walking aid and autonomous operation at a constant speed in the tunnel. In this article, a prototype of a walking assist type inspection device and a self-propelled monitoring robot that creates a developed image of the wall surface are described. The prototype device is tested in a free-flow tunnel at the Tsukuyone Hydroelectric Power Station in Tottori Prefecture where water intake into the tunnel is stopped due to renewal work from 2018 to 2019, and its practicality is evaluated.