In order to move in an environment with limited space, it is desirable to have an omnidirectional mobile mechanism. As an example, an active caster type omnidirectional mobile mechanism has been proposed. In this research, we have developed an omnidirectional mobile robot using an active dual-wheel caster capable of unlimitedly turning and implemented running control. In this paper, the performance of the developed mobile robot is evaluated through running experiments.

Mobile robots have been applied to transporting cargos, and we are studying our crawler robots for that purpose. It is difficult to control traction of a crawler robot accurately because the amount of slips during a turn strongly depends on the running environment. Therefore, we have introduced the simulation taking into account road surfaces and cargo’s weight and compared the simulated results with the actual trajectories of a robot. The outcomes show that the simulated trajectories and the actual trajectories had similar tendency, but we found that the directions of casters on a cart effected the differences. We will improve the accuracy of the simulation by including this casters direction effect in the future.

Shared control, in which teleoperation and autonomous control are mixed to move the robot, is expected to improve the efficiency of the user's teleoperation operation. However, there is a problem that the user's acceptability decreases due to the conflict of intention between teleoperation and autonomous control. In this paper, we address this problem by providing an illusion to humans. We proposed a teleoperation method called "Illusory Control" that can achieve both mobility efficiency and user acceptance by implementing a cyber-physical system that controls a robot in a real space through robot operations in a virtual space. Illusory Control has two functions: one is "Illusion of Intension", which gives the illusion that the robot is operating according to human intention, and the other is "Illusion of Time", which gives the illusion of time to fill the gap by changing human behavior when the positions of the robot in the virtual space and the robot in the real space diverge.

This paper develops an IMU-type scope of human manipulability sharing (SoHMS) which is composed of an IMU-type Mocap device and a head-mounted-type VR system. The IMU-type SoHMS can steadily analyze a human manipulability ellipsoid in real time, even when the limbs are contacted to the body and/or external objects in motion. A user can monitor manipulability ellipsoids attached at the hands and/or feet of a computer mannequin through VR head-mounted-display. A set of basic experiments was carried out for the task of push-up motion according to three different distances between both hands in order to validate the developed SoHMS. The results demonstrated the effectiveness of the new SoHMS which can successfully improve the problem of the previous one.

Recently, the driving assistance system of the welfare vehicle has been proposed in which the virtual vehicle is projected on Head Mounted Display and controlled by a third person viewpoint so that the vehicles can be easily operated by the aged. By sharing the vertual space between the biginner and the instructor in the same way as in the the training at the phisical space, we believe that the biginner ’s operating skills can be improved in a short period of time if the instructor can provide guidance and effective training. In this study, we propose a novel training system for the driving support system by sharing the virtual space in which automatically displaies AR objects. The effectiveness of the proposed system is evaluated by quantifying the work load and evaluating it quantitatively using NASA-TLX.

In recent years, there has been a remarkable increase in the popularity of augmented reality. This is due to the fact that AR technology is being used in applications for smartphones, as well as in events and marketing in the industrial field. The market for services using AR technology in Japan is estimated to be 387.8 billion yen in 2020, and is expected to grow 47 times faster than in 2020 to 16 trillion yen by 2030. This AR technology uses media such as smartphones and tablets to superimpose and display location data that indicates where you are currently viewing the camera image in the real space, images that should not exist in the real space, and virtual objects. In each frame of the image, the position where the virtual object is superimposed has to be determined according to the position and posture of the camera at that time. This problem is called the geometric positioning problem in AR.

In this study, we utilize high-precision maps to solve this geometric positioning problem. By using high-precision maps, it is possible to place virtual objects to be displayed in advance. In addition, by using the camera information and the high-precision map to estimate the position, we can match the coordinate axes of the real space and the virtual space.

While immersive VR is one of the useful tools for analyzing human behavior, it has been concerned that human behavior may be different between a VR and a real environment due to the differences in cognitive characteristics. Therefore, to clarify a condition when using VR in analyzing human behavior is reasonable, we investigated the differences between the VR and the real environment in an object searching and manipulating task. The differences were evaluated by the required time to complete the task and questionnaire for the task difficulty. From the results, we realized that using VR is not suitable when the required time for the task completion should be used as an alternative value for the real environment. However, we found there is no problem to use the required time to compare two conditions, such as scoring the performance in robot competitions.

In recent years, with the development of origami, a lot of complex origami works have been published. However, it is difficult to support folding complex origami works by the traditional origami instructions. To assist users who fold origami, a number of Augmented Reality (AR) based methods have been studied. Although AR has been shown that it is useful in assisting origami, we have not found any work that supports folding complex origami works. In addition, the low performance of the origami simulator for creating the model to be displayed is another factor that makes it difficult to support folding complex origami works. In this paper, we propose a system that assists users to fold complex origami works with an origami simulator that can deal with papers of nonzero thickness. It can present paper folding processes with AR.

The purpose of this study is to investigate the effects on human fall sensation when a VR fall sensation presentation device is used to decelerate the device during descent. A questionnaire survey using VAS was conducted to evaluate the effects of acceleration and deceleration on the sensation of falling. A questionnaire using VAS was used to evaluate the sensation of falling and the sense of reality depending on the amount of acceleration and the amount of deceleration. The results suggest that, within the range of acceleration used in this experiment, the deceleration of the device does not significantly impair the sense of falling and the sense of reality.

In this paper, we construct a projector-camera system for presenting electronic gimmick picture books using the optical illusion effect. Our conventional study realized the gimmick by combination of a projection mapping and a stereoscopic vision by an optical illusion effect. In this paper, we propose a projector-camera system for solving projection gap for the system. Our approach estimates homography between the projector and the camera from images taken with an external camera. The image of the projector is controlled and calibrated using the obtained homography. By using this system, it will be easy to install these kinds of projection systems in a general household.

In recent years, cell spheroids have been attracting large attention for their applications in drug efficacy evaluation or tissue engineering for organ and blood vessel regeneration. In our study, we proposed a method to fabricate cell spheroids using vibration-induced flow for easy recovery and mass production. However, due to insufficient design of the generated flow pattern, the produced cell spheroids varied in size and shape. In this paper, we discuss designs of micropillars for vibration-induced flow to improve the reproducibility of cell spheroid production. By using a new design of pillars, we succeeded in formation of cell spheroid. And then, we considered effect on formation of cell spheroid by a vortex flow at the center of micropillar array with vibration-induced flow.

This paper presents a two-dimension motion control system that using a magnetic field to drive a magnetically driven microneedle for trajectory-controlled motion. In our system, two vertical electromagnets are used to directly drive a microneedle that containing ferromagnetic beads to perform two-dimension motion. The experimental results show the microneedle tip can follow complex trajectories that have been planned in two-dimension space. Moreover, the position deviation was less than 0.2 μm during the microneedle motion, which demonstrates the proposed system has a high-precision capability. The proposed magnetic system realizes trajectory-controlled and high-precision two-dimension motion, which will broaden its industrial and biomedical applications.

In this paper, we made a piezoelectric actuated XY stage which can be driven in high frequency and evaluated a vibration-induced flow when a high frequency vibration was applied to a chip. The vibration-induced flow is a local flow that is induced around microstructures by applying vibration to a microchip with micropillar array patterns. We use piezoelectric actuated XY stage to apply circular vibration. However, this stage that we use is not able to apply high frequency vibration because the resonance frequency of the stage is low as a few kHz. Thus, the vibration-induced flow induced by applying high frequency was not evaluated. In this study, we evaluated a vibration-induced flow induced by applying high frequency vibration using the stage that we made.

In the field of soft robotics, which focuses on the physical flexibility of biological systems, hydrogels with high water content have been applied as a material for soft robotics because of their elasticity and light weight. Conventionally, lightcuring resins used for three-dimensional modeling of hydrogels have been considered problematic due to their high toxicity to living bodies. We fabricate three-dimensional objects by aggregating microgels made of N-isopropylacrylamide (NIPAm), which has high biocompatibility, by relative diamagnetic assembly and electrostatic attraction. This paper reports Young's moduli of microgel aggregation measured by performing indentation tests on the aggregates using a micro force sensor. When the microgel aggregation were subjected to repeated heating and cooling, the aggregation shrank and its Young's moduli increased. In addition, we will simulate the deformation of the aggregate using Young's modulus. The deformation of the aggregation by laser heating is applied to actuators for grasping micro objects in microdevices.

Bio-actuators are attracting interest as they are expected to enhance the performance of the robot by introducing the advantages of the living organism to the robot system. Along with the progress in bio-actuators, the microfiber structure is considered to be one of the optimal candidates. However, it has been difficult to achieve the manipulations to the microfiber without the use of jigs with anchors inserted into the tissue, which will cause damage to tissue structure due to the stress concentration. To overcome this difficulty, we fabricated magnetic ends on both sides of the microfiber, therefore the magnetic force could be used for balance shrinkage, fixation, transportation, and evaluation of muscle microfibers.

In recent years, with the development of the life science field, there has been an increasing demand for the measurement of cell characteristics. In our laboratory, we developed a micro hand system that uses two end-effectors like chopsticks to grasp and manipulate objects. We also succeeded in measuring the stiffness of cells by installing a force sensor on the end-effector tip to measure the reaction force during grasping. However, the conventional microhand system has a problem of stable grasping for measuring the characteristics of intracellular structure. Furthermore, in order to measure the stiffness accurately, it is necessary to perform precise grasping movements manually, which places a heavy burden on the operator. In this study, to solve these problems, we constructed automated microhand system using optical tweezers and an optical system combining two microscopes with different magnifications to enhance the working area.

The accuracy of scan matching-based SLAM strongly depends on the result of the initial alignments. In this paper, we improve the accuracy of scan matching-based SLAM by applying accurate initial alignments calculated by global registration using measurements from LiDAR intensity and water puddles as features, which are often found in damaged nuclear power plants. Experiment results in the real environment show that the proposed method improved the accuracy of the map and the trajectory of the robot by taking advantage of these features observed from the environment into account.

Decommissioning of Fukushima Daiichi Nuclear Power Station is a national urgent problem in Japan. To analyze and characterize the fuel debris distributed in Reactor Pressure Vessel (RPV), the optical fiber LIBS analysis method can be a promising solution. To deploy the LIBS sensor to the RPV, I have developed (1) a lightweight telescopic boom that can carry the sensor to the RPV, (2) a mechanism that stores the cable according to boom ’s expansion and contraction. The boom expands/contracts by winding/unwinding the synthetic fiber ropes. The cable storage mechanism stores the cable in a plane that is perpendicular to the extension direction in synchronization with the extension/contraction. The proposed mechanical design achieved a compact and lightweight mechanism of the boom. The mass, maximum length and payload were 3.18 kg, 6.03 m, and 2.48kg, respectively. The basic operation by a prototype model was successfully demonstrated.

This paper describes an investigation of the position accuracy of a long-reach coupled tendon-driven robot arm. An operational test of the arm was conducted in a full-scale field. Tip positions estimated using joint angles were compared with the tip position measured by the laser scanner. As a result, the tip position error in the yaw bending posture was larger than that in the horizontal posture. The reason is that the torsional deformation was caused by gravity. In the future, we will conduct calibration of joint sensors and modeling of deformation error caused by gravity.

To support remotely controlled robots in decommissioning, an observation system in the Primary Containment Vessel (PCV) is anticipated. This study develops a bendable modularized rail structure and moving mechanism of a monitoring robot that moves on the rail structure. The rail structure has active joints to avoid obstacles in the PCV during construction period. The bendable rail is suitable for assembling because each module has common connecting parts. The monitoring robot is designed to hang from a side plate of the rail. A side plate interpolation mechanism is attached on the joint of the rail and it enables monitoring robots to pass on the joint smoothly. Driving experiment of the monitoring robot revealed that the interpolation mechanism worked effectively. Additionally, an assembling experiment simulating actual work shows that modules can be connected in an acceptable short time.

In this paper, we fabricate a prototype to wind a monitoring arm that uses a low-melting point alloy (LMPA) in each joint compactly and verify the performance of the non-contact temperature sensing using the prototype arm. For winding compactly, we design the mechanism that winds the arm in a spiral shape. The experiment is conducted to confirm the winding and unwinding operation of the prototype of the proposed mechanism. In addition, the system needs to recognize the free and locked state of the joint for winding and unwinding automatically. Therefore, we verify the non-contact temperature sensing proposed in the previous study using the prototype of the arm. Finally, we confirm the influence of each component of the arm for short time heating through the experiment.

In the decommissioning of nuclear facilities, it is important to investigate the temperature distribution inside. In this study, we propose a method to generate 3D map information including temperature distribution using a mobile robot equipped with a depth camera based on NIR (near-infrared) and a thermography based on FIR (far-infrared). To this end, a novel calibration scheme for the depth camera and thermography line information is proposed given that the NIR image and the FIR image have the similar mechanism of projecting infrared rays into the image, the line distributions in both images are considered to be similar. The experimental results demonstrate that our mapping framework can generate reliable 3D temperature.

This paper reported an experimental trial and the results of a subject experiment using a remote maneuvering skill test system based on a robot simulator. By utilizing developed robot simulator with data collection function, we conducted the experiments with several subjects, including both skilled and unskilled operators, on the task of passing through a narrow passage in a dark. We showed that it was possible to extract the remote operation characteristics and tendency of each subject based on collected data.

At the disaster site, it is necessary to operate the robot remotely based on limited information. In such an environment, it is important to take into account high temperature areas in the environment where progress is impossible. In this paper, we propose a method for generating semi-autonomous control motions that reflects the operator’s intention. To this end, the proposed method extends the evaluation function of the dynamic window approach (DWA), a method for generating autonomous motions considering obstacle avoidance. The experimental results show that the robot is able to avoid collisions with obstacles and to avoid the high temperature zone without delicate operation by the operator.

This paper reported a simulator for verifying performance to image acquisition conditions, which was constructed to promote the development of a 3D reconstruction method from images to grasp internal state of nuclear reactor. We constructed a simulator to acquire the images under arbitrary conditions using virtual camera model running on a straight rail. Feasibility study was conducted by capturing the images in the simulator and constructing 3D model, and its result was described.

In this paper, we propose a semiautonomous motion control method for an underwater vehicle using only image information. LED marker which mounted on the vehicle is captured by the camera, then we can automatically recognized yaw angle and position of the vehicle. Using this information, we can control the position and posture of the vehicle. In addition, the attitude of the underwater vehicle is measured from the IMU sensor information, and a GUI system for maneuvering assistance is constructed. To show the effectiveness of our proposed method, experimental results are given.

We have developed a 3D reconstruction method based on image sequences recorded by remotely operated robots for decommissioning. This paper describes the quantitative method for evaluating reconstruction accuracy of a 3D model generated by Structure from Motion. We verify the performance of the 3D reconstructed model based on an image sequence. As a result, we confirmed that the reconstruction accuracy of 3D models can be quantitatively evaluated by comparing with a 3D point cloud measured by a laser scanner as ground truth.

There are problems such as a shortage of workers and construction equipment, waste, and the inability to secure accommodation facilities required by workers when building temporary housing. Therefore, if it is possible to build a building using a 3D printer, we thought that it would not only solve these problems but could also be used at construction sites. In recent years, 3D printer has attracted worldwide attention, and 3D printer for home use have become so popular that they are sold. ABS, PLA , and acrylic resin are used as the main materials for 3D printer. Also, many cases have already been reported, mainly abroad countries, in which construction materials such as cement-based materials are used to construct members that make up full-scale pedestrian bridges and simple dwellings. The purpose of this research is to develop a super-large binder jetting concrete 3D printer.

In this study, we have been investigating a cart equipped with an omni-wheel type moving mechanism for ground-penetrating radar surveys. In order to investigate a power-assist control method for a measurement cart in an outdoor environment, an experimental cart was constructed and the behavior of the cart was measured using load cells and acceleration sensors. As a result, it was found that the measurement cart was exert to cyclic forces when the operator pushed it by hands.

The purpose of the research is visual human detection approaching the crane’s suspended load in order to avoid serious accidents near cranes. There are two approaches to human detection. One is frame-wise detection using machine learning. The other is motion-based one. In this research, we take the first approach and test various machine learning algorithms, we also test YOLO, which is capable of real-time processing.

Teleoperated excavators have a risk of overturning or falling when digging because it is more difficult to receive information regarding the machine posture and the work object condition than in the case of actual machine boarding. In this study, the machine instability derived from the attachment posture and the digging reaction force has been proposed. In addition, for feedback regarding the machine instability to the operator, an intuitive visual presentation method using a meter is proposed. To verify the effect of the proposed machine instability feedback, an experiment with the actual teleoperated excavator was conducted. As a result, it was confirmed that the work could be conducted without generating a greater tilting of the machine body while maintaining productivity. Therefore, the possibility of working safely in the teleoperated excavator using machine instability feedback was clarified.

The conventional remote operations of an excavator have low work efficiency compared with direct operations. To overcome this problem, we have developed a semi-autonomous control system that consists of autonomy (dynamical system with attractor) and human action (admittance control). Furthermore, we have designed task selection, digging position selection, and frequency separation of operation in the semi-autonomous control system. This paper proposes a leader-follower system with a low DOF leader robot that replaces the turning DOF with video information. The follower robot’s turning angle is displayed as an animation on the LCD monitor on the underside of the leader robot. We implement the semi-autonomous control system that has been proposed so far with the proposed leader-follower robot and verify the operation.

This paper proposes a position control method for hydraulic excavators with reduced effects of a phenomenon known as pressure interference. The pressure interference takes place when two cylinders sharing one pump are driven simultaneously. The proposed position controller avoids the pressure interference by limiting the velocity of the bucket in the task space. The proposed method is validated with our realtime simulation environment.

Conventional plastered wall is made by workers that have special skill. However, the number of workers is decreasing, and additional work force is needed, because of the hard work environment and lack of younger human resources. In this paper, robot arm supposing plaster work is introduced. To achieve this, mathematical model is proposed. This plastering robot aims to do plaster work on behalf of human workers and helps construction industry.

Recently, as buildings have become taller, base structures on construction sites have to be excavated[中村 1] wider and deeper. Generally, excavated soil is transported vertically using clamshells. However, clamshells are large construction machines, and their operation is subject to physical and time limitations. In this study, we focused on the intestinal tract. We used a peristaltic conveyance system that simulates a peristaltic movement pump to vertically transfer excavated soil. In a previous study, we successfully lifted low-water content sediments using a watering system that increases the water content ratio of the transported sediment. From the experiments, the lifting conditions were determined according to the type of sediment. However, the relationship between the driving conditions of the device and vertical transport is not known. For practical use, the rate of lifted soil should show an improvement. In this study, we investigated the driving conditions. The lifting rate was compared for various patterns under different conditions.

There is an urgent need to automate earthmoving works for large-scale six-wheeled dump trucks. At actual sites, a dump truck stops at a loading position, informed by the backhoe's backet, with its bed facing the backhoe. Therefore, it is necessary to plan a path including a turning back and rapidly replan according to the goal position change. However, there is no suitable path-planning method for earthmoving works of a large-scale six-wheeled dump truck. Here we propose a path-planning method that moves the dump truck forward and then backward to the working position. According to the goal position change, it also rapidly generates a path from the turning position to the new goal position. We confirmed the proposed path resembled the trajectory of human-operated dump trucks, and the time required for replanning was shortened. We also confirmed the autonomous dump truck could do earthmoving works in cooperation with a human-operated backhoe.

In this research, we aim to automate the peeling of the coating film using IH by running on the wall surface with a magnet and a mecanum wheel. The goal of this research is to be able to automatically adsorb and run on the wall surface with the coating film peeling device loaded. We are conducting experiments with a testing machine and have actually succeeded in running on a wall surface with neodymium magnets. Currently, we are manufacturing lifting magnets for self-sustaining operation and wall surface adsorption by detecting the steel frame on the wall surface.

In determining the bolted state of a bridge, the visual judgment varies depending on the operator. Therefore, in this research, we will use machine learning using images to improve the accuracy of discrimination. This report describes a basic study of the identification method using markers.

Inspection of concrete structures is necessary to ensure the safety of the buildings and structures. The demand for inspection robots is increasing in order to inspect a huge number of structures. The hammering test is one of the inspection methods for grasping the internal state of concrete. A hammering mechanism that can be mounted on a robot is necessary to realize an inspection robot for the hammering test. In this research, we propose an inspection robot system equipped with a compact and lightweight hammering mechanism using a snap motor.

As being eco-friendly has become a major keyword in many industries, the construction industry has put much effort toward making its products eco-friendly. Excavators being the representative machine of the construction industry, it has tuned in to the new trend as well. Hybrid excavators have taken over the market with their high fuel efficiency and low emission. To further push the boundaries of eco-friendliness, this paper suggests a model of a fully electric excavator and evaluates its effectiveness based of the comparison manipulatability with existing hydraulic excavators.

In a geothermal power plant, the soundness of steam-pipeline is an essential factor that could influence power generation efficiency. However, it is laborious and dangerous for staff to inspect the state of pipes. Therefore, authors have developed a system that automatically conducts inspections using a small unmanned aerial vehicle (UAV) equipped with Laser Imaging Detection and Ranging (LiDAR) and RGB/Infrared Rays camera. In this research, authors focus on misalignment, deformation, surface colour change, temperature change detection on piping. The authors propose a piping model in the 3D-map generated by the UAV to detect these abnormal sections in pipes. The experimental results show that our system could detect misalignment less than 25 mm in horizontal view on the straight pipeline and deformation larger than 2 cm. Recall of colour-changing detection on the test was 96.2%, and the result of temperature-changing detection on dummy data also show the effectiveness of our methods.

In recent years, teleoperated robots are expected to be used in the inspection of infrastructure and plant facilities and at disaster sites. With the increasing sophistication of inspections and surveys in recent years, high-capacity and stable communication quality is required for remote control of robots. In this study, we aim to solve this problem by constructing a wired communication network using multiple robots in the field. For cable laying using robots, we are developing an auto-reel system that uses drum rotation control by cable tension. In this paper, we describe the modeling of tension dynamics in cable auto-reel.

The ACC (Adaptive Cruise Control) have been applied for many passenger vehicles generally. Even the conventional ACC has an enough performance to follow a preceding vehicle. On the other hand, it has a possibility to cause excessive acceleration and jerk. To avoid that excessive behavior, in this research, the RL (Reinforcement Leaning) which can consider some objectives to determine the control input is proposed for ACC. In addition, to balance performances of following to a preceding vehicle and reducing jerk, the RL’s reward is designed by using unique thresholds to this research. This method is evaluated by being compared with ACC by LQR (Linear Quadratic Regulator). As a conclusion, the balance of performances and the robustness to a dead time are evaluated for RL method.

In this paper, we propose an alternative method to the Ratio-test for determining the reliability of RTK positioning solutions considering the vehicle trajectory. Ratio-test is known to have a problem that it does not work effectively when the initial position of the search for integer indeterminacy is not reasonable due to multipath and other effects. In addition, because the reliability of the RTK positioning solution is judged, there is a problem that the output solution is reduced by the judgment. While, if we focus on the first solution of the LAMBDA method, it is known that there are many places where the location accuracy is sufficient even before the implementation of the Ratio-test. Therefore, in this study, we focus on the use in automobiles and aim to improve the usability of high-precision positioning even in a multipath environment by reviving the results before the Ratio-test by considering the vehicle trajectory.

In this research, we aim to develop a mobile robot that can navigate through on a university campus using end-to-end learning. We can easily obtain data frames that consist of sensory data (e.g., RGB camera) and control signals (e.g., steering) from a driving simulator. Although many prior works have proposed end-to-end driving methods trained on simulated data, they are not applicable to real-world scenarios. Therefore, we propose a domain adaptation method to apply a driving policy learned on simulation data to a similar real-world use case.

This paper proposes a method for a small truck with four cameras installed to turn at an intersection to transport harvested products on a farm road. Semantic segmentation is used to recognize the road surface and the images are divided into straight road, branch road, and non-drivable area for more accurate intersection recognition. To calculate the amount of vehicle movement, visual odometry by tracking feature points on the road surface is executed. A grid map around the vehicle is created by integrating these results, and the reference path is generated on the map using a clothoid curve. The vehicle is controlled according to the generated path. The proposed method is verified by conducting experiments on a real vehicle at an uneven intersection.

As an interface between human and automobile, we are developing a system that enables advanced information input to automobiles via a steering wheel using string vibration sensor. However, there is a problem that it is difficult to distinguish force to grip the steering wheel for driving and input actions. In this paper, we focus on differential sensor and investigate a method to determine input information according to its characteristics. Specifically, we set two threshold values each for time axis and voltage axis for sensor output, and implement four input actions. In experiments using a differential sensor and a proportional sensor, the differential sensor showed higher average input success rate than the proportional sensor for the input actions while gripping the steering wheel for driving.

Although ultra-compact mobility was capable of traveling the limited region, the Ministry of Land Infrastructure and Transport has announced that the newly revised the security standards to ensure safety on the basis of the Road Trucking Vehicle Law to the prospect of the 2019 fiscal year. If ultra-compact mobility becomes widespread, measures to prevent accidents for elderly drivers and expected, but the equipment is not provided to ease the driver's steering burden due to the space inside the vehicle. In our research group, the biological information and the vehicle information are fed back to a steer-by-wire system that can assist the driver's steering operation. In this study, we investigated the effect of steering reaction torque to the steering burden on the upper limbs by measuring the surface myoelectricity.

In this paper, we focus on training systems for aircraft pilots through a flight simulator. In order to quantitatively evaluate the flight skills, we provide an autopilot system based on CNN. In contrast to autonomous driving vehicles, aircrafts have 6 DOF in the dynamics model. Therefore, control outputs from the autopilot are divided into vertical and horizontal components. In order for the autopilot to determine the respective control outputs, we propose a CNN with branched architecture. Furthermore, the CNN estimates a state of the aircraft, such as altitude and goal direction, in addition to the control outputs. For this purpose, multi-task imitation learning is applied for network training. In the simulation experiments, we show that the aircraft based on the autopilot system is enabled to imitate the instructed flight and land on a desired runway.

In this research, we proposed an active management method to monitor more autonomous vehicles (AVs) remotely with few observers. A management system is created to get the status from the AVs and separate the monitoring requirement to the observers optimally. When the requirements might be intensive, the management system can adjust the movements of the AVs actively to distribute the monitoring time, which can make the observers monitor more vehicles. Based on the data acquired from the driving simulator, we create a numerical simulator and test our method with different pairs of a large number of AVs and observers. The result showed that our method can reduce the utilization degree of observers and make them monitor more AVs.