We have developed many types of Multi-Motor Power Unit (MMPU) that drives an output shaft by using power of multiple motors. In this study, we suggest a new concept of MMPU that consists of different characteristic motors. Also we integrate a reduction gear into a suggested MMPU. The reduction gear is dual ratio reduction, 1st step of reduction gear consists of a star-type planetary gear, and 2nd step of reduction gear consists of mechanical paradox gear mechanism.We developped the reduction gear built-in type MMPU that have two types of motor and applied it for competitiive robot of reactor abolishment technology. In this report, we had suggested imporvement of reduction gear to increse maximum output toruqe of MMPU.

In this paper, we propose a new formula to accurately estimate the output value on the real machine of the proposed mechanism as a part of the development concerning the high power joint mechanism. The previously proposed formula did not consider the energy loss when the slider translated. Considering the mass of the slider, we derived an expression that can estimate the angular velocity of the output link from the energy conservation law and the geometrical relationship. As a result of comparing the theoretical value and measured value, the performance of real machine could be more accurately estimated by using the proposed mathematical formula.

Recently, with aging piping facilities increasing, demand for inspection of pipes is growing. However, conventional manual inspection is difficult for underground and high place pipes. Therefore, development of pipe inspection robots is desired. The pipe inspection robots using fluidic flexible actuators have advantages of simple structure and high safety for pipes. However, depending on the length of fluidic supply tube, movement distance is limited and response is delayed. So we developed a mechanism that has a fluid supply source and moves like inchworm. It has a bellows actuator for moving robot forward in the center, and two balloon actuators that inflate and fix the front and rear of the robot. These actuators are connected with two tube pumps to move the internal fluid. By controlling internal fluid volume at a certain sequence, the robot moves in pipes. Proceeding in the pipe of the robot was confirmed.

There is a demand for an off-road mobile robot having a high mobility and a wide movement range in a complicated environment such as a disaster site in which a person cannot enter. However, even if conventional moving mechanisms such as wheels, legs, crawlers are used, it may be difficult to move such as a steep slope. Therefore, in this research, we propose a method to fix wires directly to the external environment using elastic telescopic arm. And in this paper, we aim to control the tip position by the bending motion of the arm, and we show the results of the experiments on the relationship between winding value of a rope and the tension and the tip position. And we measured accuracy of the tip position control by winding value of a rope.

Snake-like robot is a hyper-redundant mobile robot with an articulated body composed of serially-connected modules driven by actuators. Recently several hyper-redundant mobile robots have been developed in hope of the versatility brought from its simple form. Snake-like robots have been developed aiming at its applications to operations dangerous to man: the in-pipe inspection at chemical or nuclear energy plants and the rescue of victims under collapsed buildings by making use of its flexible and slender shape, and the mine detection by distributing its own weight on the ground and so on. In this paper, the novel mechanical design is proposed to enhance the mobility for various environments, and a prototype model is composed of serially-connected modules with suspensions. The performance of 1st prototype model is described.

In this paper, the control method which can move in the diagonal direction for active omni-directional wheel mechanism with dual-rings is proposed. Active dual-ring wheel with high performance on step traverse had been developed. However, it was difficult for this mechanism to move in the diagonal direction because a rotation direction of the ring wheel is reversed at constant period of the wheel. Thus, the authors proposed a control method for diagonal direction movement. Next, the wheel path was simulated to confirm the validity of the proposed method. Finally, the effectiveness of proposed control method was verified by experiments.

Mechanization and automation of labor works on steep slope are required to reduce its burden. Previous studies use designed wheels and continuous tracks with large a coefficient of friction to improve mobility of robots on steep slope. However, friction force decreases, and attitude of robot becomes unstable when incline of slope angle increases. Not only enlarging a coefficient of friction but also another approach is necessary to realize stable attitude of robot on steep slope like mountainous areas. Therefore, we propose a multi-copter type mobile robot. The robot can keep stable attitude on a steep slope with large friction force because the thrust force from the multi-copter pushes a robot body against the ground surface. In this paper, we simulated multi-copter type mobile robot’s behavior on steep slope by using physics engine to evaluate proposed methods.

PID control decreases gain of high-accelerated input signal against actual signal of actuators due to inertia of its weight. Particularly, in tracking application of galvanometer mirror, its gain decrease causes motion blur as the target speed does not correspond with tracking speed. While in this research we realize to generate high linearity waveform using pulse input into the PID control box. Pulse input generates maximum acceleration of galvanometer, hence we utilized it for generating raster scan. Experimentally, the improvement of waveform and image quality applied for motion-blur compensation was confirmed when the target moves at the speed of 11 km/h.

Personal Mobility Vehicle RT-Mover PType WA (P-WA) can move on rough terrain (including a few step-stairs) and always keep its seat horizontal in spite of the simplicity of mechanism. P-WA has two modes, Wheel mode and Leg mode, which can be used according to the driving environment. In this paper, we have developed a mode change algorithm for P-WA from four-wheel steering mode to two-wheel differential drive mode by using an auxiliary wheel in order to improve the turning performance.

Though leg-wheel mechanisms are expected to have both merits of wheels and legs, proposals so far tend to employ redundant number of actuators and a complicated structure. Thus, this study proposes a new mechanism which employ minimum number of two actuators and a simple structure consisting of two half-circles and a four linkage mechanism to form a wheel-shape and a leg-shape selectively. At first, kinematics is derived between movements of the actuators and the foot. Secondly, a single leg-wheel prototype of is made to confirm the motion. Then, a quadruped prototype is designed and fabricated and experiments are performed to confirm the effectivity of the proposed mechanism.

This research focuses on the dynamic braking of omni-wheel rollers of a dual robot system to improve its mobility. Previously, the authors developed the braking mechanism for the system. In this paper, the modeling and the development of basic motion control of an actual dual robot system is implemented.

In recent years, demand for mobile robots that can move in people coexistence environment is increasing. Many of these are wheels moving type. The footprint of wheels moving type tends to be large in order to adapt to uneven road surface. In a human coexisting environment, compact structure is required. Therefore, in this development, utilizing the compactness of the inverted pendulum wheel robot, we study a robot that can overcome a large step like a staircase by jumping. As a first report, we propose an active suspension mechanism using Chebyshev link and conducted experiments using a prototype robot. Experiments confirmed that the prototype robot can jump to 1.3 times the link stroke. It also confirmed that it has realized the posture stabilizing during jumping and impact absorption at landing time. Therefore, the link mechanism has a function as an active suspension.

In order to move over the concrete wall and inspect it, we developed a wall moving robot using a vacuum pad. This robot walks on the wall while sticking two vacuum pads alternately. In addition, it has 5 degrees of freedom, and it is possible to achieve both movement of the wall surface and movement between the other surfaces. This time, we made a larger improved machine than the smaller prototype we made previously. As a result of experiments, it was possible to perform parallel movement, movement between other faces, and rotational movement. We confirmed that we could do the assumed moving method.

In a facility that stores radioactive waste, it is necessary to inspect the deterioration of drums which filled with waste. In the inspection of drums, a method of visually confirmation whether the surface of the drum is normal or not, The distance between drums is about several centimeters, so people can not get in. Therefore, in this research, we will develop a robot can photograph the surface of a drum can with a camera go into such a narrow gap and even if there is a narrow gap.

The omnidirectional motion can move any direction without changing the direction of its own. The authors have developed Wave-wheel mechanism and Screw-type differential rotating mechanism. Also, the stepping performance were improved by extending the ground point from point to line. In addition, we have evaluated the motion characteristics of one unit and made high rigidity. In this paper, we examined the composition for vehiclization. In the vehicle configuration, the minimum necessary two units are arranged in parallel, and these are actually realized. We conducted experiments using actual mechanisms and confirmed that the inventive principle is also effective for vehicle construction, omnidirectional motion.

In recent years, development and development of a robot wheel chair with many functions capable of overcoming various obstacles in real space has been progressed as expansion and expansion of the range of activities of persons with lower limb disorders. In this paper, we describe various functions and mechanisms of robotics wheelchair under development.

The assistance of an assist walker is effective for movement of elderly people who have anxiety concerning walking. A motor mechanism for assisting acceleration / deceleration has been developed in recent years. However, incorporating a drive mechanism in a swinging caster makes the structure excessively complicated. For this reason, ordinary wheels with fixed axles are generally used for assist. On the other hand, if such a normal wheel is used for a walker, it will impair its turning ability in a narrow place. In order to realize the intermediate characteristic of these two problems, we trialed a walker incorporating motor power to the omni wheel.

In recent years, attention to flying robots typified by drones has been increasing. However, unfortunately, drones often cause accidents. On the other hand, as safer robots, flapping flight robots have attracted attention. However, it is difficult to mount systems other than the flight functions due to the limitation of the weight and size of the airframe. In order to solve this problem, it is necessary to increase the size of the airframe. In this research, we design a large flapping flight robot with a large payload toward practical application of flapping robots.

In this research, we develop an automatic compression and release device for jumping leg mechanism of bird-type flying robot. In our previous papers, we have developed a jumping leg mechanism to takeoff a flapping flight robot and succeeded in takeoff experiment. However, in the experiment, the developed mechanism had to compress and release manually.

In this research, we develop a device that automatically compresses and releases the jumping leg mechanism. The device applies a force from the outside using the string and the motor to the jumping leg mechanism, and performs the compression and release. Using this device, we performed an automatic jumping experiment of the flapping robot and confirmed that the robot can jump with initial velocity close to that when manually compressing and releasing.

We have developed a fire fighting multicopter which utilize lateral rotors to support reaction force of discharging water and tension of hose. These horizontal disturbance to the multicopter are measured by installed sensors and automatically compensated by thrust of the lateral rotors. Thus, the multicopter does not need to incline and the operator is not required additional operation against these disturbances. An experimental multicopter, which has four 28-inch propellers and two 15-inch propellers, performed discharge of as much flow rate as indoor fire hydrants.

We are developing a slung multicopter for fire extinguishing on higher buildings. The multicopter discharges water while controlling its yaw and pitch to aim the fire. First, water discharge locus from the swing up multicopter was calculated for planning of delivering the water to the target point. Experiments are held with a multicopter slung from a connecting corridor at the hose length of 15 m. The results show that the slung multicopter system is able to control water locus to target point with control of yaw and pitch angle of the multicopter.

In this research, we propose UAV guidance method for non-GPS environment. For guidance, laser light of laser range finder (LRF) installed in ground station and UAV are used. When the laser light is oscillated from the LRF, it is reflected by the corner cube installed on the other side and is detected by its own detection unit. PSD is installed in the detector, and the deviation of the laser light from the center of the corner cube is monitored. The position of the UAV can be obtained from the distance measured by the LRF and the angle of the laser beam. This is used to guide UAV.

In recent years, unmanned aerial vehicles have attracted attention around the world. Many of them are expected to carry heavy items, and as more large-sized aircrafts become popular, various advantages can be considered. In this study, we focused on the fact that the freedom of movement of the aircraft is high and we carry heavy objects with multi-copter that can stop in the air. We aim to design all weather type multicopter for transporting goods which can carry about 100 kg at a time, while improving them, and to produce it. This time, we selected 16 motors aircraft, considering the number of motors and propellers to use and the number of pieces necessary for obtaining sufficient thrust and the maintenance of the aircraft. We investigate the problem that this in practical flight and analyze and examine the characteristics of the aircraft.

In this paper, we proposed the long-reach aerial manipulation by multi-rotor UAV. The aerial system consists of a hexarotor with winch mechanism and wire-suspended robotic hand equipped on swing-suppression device.The system allowed robotic hand act as a long-reach manipulator. The swing motion of robotic hand is suspended from swing-suppression device. To verify maximum length of two wires, we analyzed yaw axis swing motion of wire-suspended robotic hand. In experiment, we succeed to perform the pick and place manipulation task.

Multicopters are being used to inspect the wall surfaces of tunnels and dam embankments. However, because the number of GPS satellites received can not be obtained sufficiently in tunnels or dams, the precision of position control of multicopters declines. Therefore, in this study, Laser range finder (LRF) was used as an alternative to GPS. Measure the distance and angle between the wall surface and the multicopter by irradiating the wall surface with LRF. We verified the Remote Control Support System to make multicopter flying in parallel at a fixed distance from the wall surface using them.

The goal of this research is to realize stabilized control of a drone under serious conditions, especially in strong wind area. We use DDPG (Deep Deterministic Policy Gradient) method, one of the deep-reinforcement learning algorisms which can be used for control in continuous action space. Reinforcement learning is useful for an autonomous tuning of control rules, and deep learning is suitable for obtaining a control mechanism in complex states. In this paper, we explain a proposed control method and simulation environment for reinforcement learning of a drone.

In Japan, when disasters such as typhoons and earthquakes happen, the road may be broken. Therefore, in this research, we propose manned multicopter operation as a new transportation method which can cope with such a situation. By using manned multicopter, it is possible to move to a place that is difficult to access, or to move to a destination at a straight line distance.

In this research, we first create a dedicated simulator for the purpose of verifying the operation of a manned multicopter flight controller and investigating the behavior of the aircraft, and conduct simulation.

In disasters, quick information gathering through drones is beneficial. However, it is not suitable for long-term investigation because operation time is short. Also, it is unsuitable for collecting information near the surface on the ground. Therefore, we developed a sensor module installation system by falling method using drones. In this system, it is possible to install fixed point observation sensors quickly. Moreover, we verified that the sensors can be installed by experiment.

In this paper, we examine low-speed flight characteristics of a tilt-rotor UAV by estimating the aerodynamic coefficients based on the results of flight experiments and CFD analysis. The longitudinal motion in the rotary-wing mode of the vehicle is modeled by using the aerodynamic forces that have a component proportional to the airspeed. The experimental results show that the component is especially useful for representing the drag force acting on the vehicle in the rotary-wing mode. Although the aerodynamic forces obtained from the experimental data are quite noisy, the magnitude of the forces is consistent with the results of CFD analysis.

We describe a stereo camera and HMD system for teleoperation of aerial manipulation robot. The aerial robot in this study consists of a hexarotor platform, a top-mounted gripper and a manipulator mounted at the bottom of the body. After fixing the robot by grasping a bar-like object using the top-mounted gripper, the bottom-mounted manipulator is used to perform manipulation tasks.The proposed system consists of an on-board stereo camera that can rotate and slide according to the movement of operator’s head. During the manipulation process, the movable stereo camera eliminate an occlusion and provides stereo perception to operator.

In this paper, we construct a method of detecting safe landing sites for a tilt-rotor UAV from monocular images. We divide an image into small patches whose size corresponds to the area necessary for the UAV ’s landing, and calculate a safety score for each patch. The safety score is obtained by averaging two scores, the score from ground surface classification and the one from estimated flatness of ground surface. The ground surface in each patch is classified into 9 classes by using Convolutional Neural Network, and the flatness of ground surface in each patch is estimated by optical flows obtained from consecutive two images.

The raerial manipulator is the UAV (Unmanned Aerial Vehicle) which is equipped the robot arm used for working during flight. In this paper, we describe that the development of an aerial manipulator which based on the multi-copter UAV with tilt propeller mechanism. In our research, the aerial manipulator combining the UAV which can control both position and attitude independently and one robot hand has been developing. In order to reduce weight, the robot hand is also serve as landing gear during landing. Additionally, the flight experiment result is showed in this paper.

Many researches have tried to imitate the flight of birds, but there are not many successful non-running takeoffs. In this study, in order to achieve the non-running takeoff, we prototype several kinds of wings that have different shape and structure. The flapping experiments are carried out and the forces are measured and evaluated. The appropriate wing was clarified.

The tandem wing type aircraft transforming mechanism for VTOL has many examples which equip each wing tip with a tilt type rotor. In this research, It transforms it into VTOL/STOVL by equipping the barycentric position of the airframe with two tilt type rotors. The proposal method can be used though interests of both are counterbalanced as for the comparative result by the scene for which a steady flying characteristic is necessary.

In this research, we propose two methods to change the flapping angle of flapping flight robot using crank mechanism. Currently, as a flapping mechanism of bird-like flight robot, there are two types of methods using servo motors and using a crank mechanism. The servo motor type machines can freely change the flapping angle during flight by motion program. However, when the flapping speed is increased, it becomes hard to achieve a sufficient flapping angle. On the other hand, although the crank type mechanism can keep the flapping angle even if the flapping speed is increased, the crank type mechanism can not change the flapping angle during flight. Actual birds fly while changing the flapping angle during flight.

In this research, we propose two methods to change the flapping angle of flapping flight robot using crank mechanism and verify it with the developed machines.

We developed a quad tilt rotor type VTOL with fixed wing, which aims to transit to a horizontal flight state, behaving as a quad rotor helicopter during vertical take-off and landing, and as a thrust device by tilting four rotors during level flight. In this experiment, hovering experiment and horizontal flight experiment by rotor inclination were carried out. As a result, it was found that horizontal flight is possible without affecting the pitch axis when the aircraft stabilizes at hovering and tilting the rotor.

Since safety, reliability, and durability are required when the flight robot becomes large, we have studied the use of jet engines for both thrust and power generation, and have carried out technical development including manufacturing. In order to install a jet engine on a flying robot, it is necessary to reduce the noise to an acceptable level in society, and report on this noise reduction technology this time.

Laser scanners are widely used for obstacle detection of a mobile robot. For wide-area scanning with fewer laser scanners, the measurement system consisting of rotating mirrors and a 2D laser scanner was developed by authors[6]. In this paper, we discuss the mirror width and the adjustment of the scanning surface in consideration of the laser beam. The experimental results show that the proposed system can measure the square area of 1.6 m × 1.6 m on the road surface.

It aims at autonomous movement of the robot in a crowded environment such as a station or a downtown area. In this paper, we propose a method of extracting moving objects and visualizing flow of moving objects. In the proposed method, a moving object is first extracted by a background subtraction method by image processing using a depth sensor mounted on a robot. Then, label the moving object. Find the centroid position of each segment. Find the path of the moving object from the depth information corresponding to the centroid position. In this paper, we describe the outline of the algorithm and the results of the evaluation test on the proposed method.

This research aims to develop a method of estimating road boundaries using a 3D LIDAR. Our previous methods estimate road boundaries on a flat terrain with multiple sensory feature and particle filter. However, they are not suitable for unpaved roads on a complicated terrain such as mountain roads. In this research, we extend the road model used in these methods to three dimensions and change the state transition method accordingly. We tested the proposeed method using a simulated road including a gradient change and confirmed that the road boundaries can properly be estimated with respect to the gradient change.

In recent years, development of mobile service robots has been actively conducted. It is necessary for such a robot to recognize persons reliably. This paper describes a person detection using 3D LIDAR for person tracking with a small robot. Since our LIDAR has a vertical viewing angle of 30 degrees and is put at a low viewpoint, the whole body of a person is not observable if the distance to the person is less than 1.5 m. By adding data of partially observed persons for training, we improved the person detection rate. We evaluated the precision and the recall for the discriminator.

Recently, as many public institutions are promoting barrier free facilities, an accessibility map is very effective for using them. The research proposes a method of analyzing a 3D geometrical map for generating an accessibility map. A 3D geometrical map is generated by SLAM with NDT scan matching, and is then analyzed to detect walls and ground planes as key objects for accesibility map generaton. Ground detection is done by finding ground segments in divided local regions and connecting them. Wall detection is done by finding line segments in a 2D map obtaining by projecting 3D points on the ground plane, and verifying suggested vertical planes. We are planning to extend the method so that it can detect other objects, such as slopes and stairs, and to generate an accessibility map using such detected objects.

Autonomous driving is one of the hottest field in these days. Some researches suggests automated driving method using GNSS and 3D LRF, but it takes too much costs to apply large area. In this paper, we suggest a image-based control of a autonomous driving robot using omni-directional camera, in urban district which needs less costs compare to methods using 3D LRF. We suggest a course detection with SegNet, and Grabcut algorithm. We varified its validity with experiments excuted at outdoor.

In this paper, we propose a tracking method of moving objects considering received intensity of 3D-LIDAR. Also assume that there are multiple objects similar to the tracked object. In the case where it deviates from the view of LIDAR, there is a possibility that it can not be correctly determined whether it is an object to be tracked or not. By using the attendance strength together, we try to rediscover and track the object once lost. We describe the effectiveness of this method as a verification experiment by correctly tracking an object in an environment that can not be tracked with position information alone.

In this paper, an indoor acoustic localization method which uses asynchronous acoustic beacons and two-element microphone array is proposed. The location likelihood is calculated by projecting the cross-correlation functions of two elements of microphones to the state space. This projection enables likelihood calculation without unique direction-of-arrival detection and relieves constraint of microphone number. The proposed method was evaluated by an experiment with four acoustic beacons and mobile robot. The results showed that the proposed method with two microphones achieves 90% error of approximately 0.4 m, which is reasonable comparing to the result with four microphones.

This paper presents an object recognition system using deep learning for a mobile robot with LRF (Laser Range Finder). First, the quadratic distance data obtained from the LRF is flattened and classified by point groups using the developed algorithm. Next, the point groups were imaged one by one from the obtained point cloud information. These images were rotated according to the orientation of the robot. Finally, a CNN (Convolutional Neural Network) was used to recognize what these point cloud images represent. The LRF data was gathered by moving the mobile robot around many passengers and vehicles in an outdoor environment. And it was identified whether the point cloud images made from the data represent a human, a wall, a bicycle, a car, other objects.

This paper describes a method for estimating the position of a transfer robot using AR markers and an algorithm for moving the robot to a specified position. In particular, we aim to improve the accuracy of the stop position of a robot moved using SLAM. We proposed a robot position estimation method and an algorithm for moving to the target position based on it, and carried out an experiment using the experimental omnidirectional rover to confirm the positioning accuracy. As a result of the experiment, it was within 3 centimeters in the direction perpendicular to the marker and about 10 centimeters in the parallel direction with respect to the target point 1 meter away from the marker.

In general, the autonomous robots charge their batteries by using exclusive docking stations. However, these robots’ operating range are limited around charging station’s area and it takes cost each time to set up a new charging point. In previous researches, to solve those problems, we suggested a new charging method called “Plugin-docking system” that inserts charging plug into outlet directly with the image measurement of outlet’s position and angle. In this paper, we propose new measurement methods of the angle of outlet in order to do the Plugin-docking more successfully for the outlet that another plug is inserted.

It is very important for the autonomous mobile robot to recognize the environment. We have developed Wide FOV 3D LiDAR aiming for better environmental recognition than general 3D LiDAR. Wide FOV 3D LiDAR is far superior to the vertical viewing angle of the conventional one. Furthermore, its rotational scanning speed is controllable. Therefore, if the rotational scanning of Wide FOV 3D LiDAR is slowed down, its point cloud becomes high density. However, the slow rotational scanning will lower real-time performance of 3-dimensional scan. This paper presents the density control method of the point cloud with Wide FOV 3D LiDAR for compatibility between generation of high density point cloud and maintaining of real time performance of 3-dimensional scan. Our proposed method extended object detectable distance without reducing real-time performance.

Obstacle avoidance functions are necessary for mobile robots. However, there are few methods to generate a path considering the height of obstacles. In this paper, we describe a verification experiment of a method of generating a avoidance over a small obstacle using the minimum ground clearance of a wheel type robot. A robot equipped with a three-dimensional distance sensor was used for a differential two-wheel robot. We were able to detect small obstacles by a three-dimensional distance sensor and avoid avoiding the minimum ground clearance.