In this paper, we describe the soil sampler developed for robotization of sample collection for soil diagnosis. Our goal is to develop the sampler which can be mounted on a small robot running in paddy field (about 300 [mm] in length and width). Sampler is able to collect 10 [g] of soil with depth of 5 [cm] at nine or more sampling points. In order to achieve this specifications, we made a prototype of sampler which connect and rotate containers to take and store soil in chains. As a result of the test, it was confirmed that it is possible to collect soil by the robot equipped with this mechanism.

The average age of agricultural workers is aging. We focused on under-trellis work done on crops on the shelf with raising their arms. In the upward work, it is necessary to keep the arm with its own weight of approximately 2 to 4 kg. It is considered to be a cause of fatigue of arms and back. Therefore, in this research, we attempted to reduce the fatigue of the upward work by the weight compensation mechanism that balances the elastic force of the spring and their arms weight. It can compensate for their arms weight by the force of the spring in various postures of the arm at the time of work by using lightweight and highly accurate weight compensation mechanism authors developed. Furthermore, a horizontal multi-linkage mechanism that follows the movement of the shoulder and arm of a human is used. In this way, it is completely pure mechanical and simple structure without a motor and battery, and it can be expected effective in all upward works.

In this paper, we introduce research on robots that can harvest tomatoes. It is now done by farmers to harvest agricultural crops such as tomatoes. There are problems that many farmers take too much cost and time. Therefore, in this study, we discussed the behavior decision of robot using image processing aiming at autonomous tomato harvesting robot. The tomato harvesting robot aims to harvest after moving to a position where tomato can be recognized and harvested. Also we will redesign the end effector part which was a problem from last year. We also participate in the tomato harvesting robot competition to be held at Kyushu Institute of Technology in December 2016.

We proposed a mechanical system capable of continuing degassing and weeding with the aim of increasing the yield of organic cultivated paddy fields. The prototyped machine is compact sized and is a long stick into the ground while floating in water. The performance of the prototype was investigated the experiments in actual paddy fields. As the results, it showed sufficient weed suppression performance, but, the yield did not increase through indirect proof of the gas.

This paper presents a feature point matching to generate mosaic image that expresses a cultivation environment, as a part of behavior strategy of tomato harvesting robot. Since that mosaic image requires tomato fruits and stems in front face, in this study, an IR image suitable for detecting the object in front is used. Proposed method uses robot moving distance and location of original feature point to decide detection area to search for feature point. Compare features of feature points in this area and original feature point, then decide matching point. Feature point matching of two consecutive infrared images was performed. By adding detection area of feature point, it is possible to reduce false matching.

We studied shaft tillage cultivation as practiced by an autonomous robot. Through previous studies, we found that leg locomotion has a higher compatibility with the shaft tillage method than wheeled locomotion. Therefore, we are currently developing a legged robot for shaft tillage cultivation. In this article, we will report the progress on developing the real scale model of the legged robot. In particular, we describe the development of a backlash free servo module for the swing leg motion, the power transmission method of the standing leg and the sensing method of the ground contact.

This research presents Giacometti gait mechanism for the Giacometti robot. Motivated by the Giacometti structure, which is thin and long, the robot has leg design that is non-stiff and deformable which makes it difficult to walk with conventional walking gait. Using the proposed gait, walking experiment was conducted with various input pressure and gait frequency and could achieve stable walking at 0.05m/s.

The teleoperation system for the hexapod robot and its footstep planning algorithm are proposed to assist the teleoperation of the walking on the environment cluttered with obstacles. In this paper, ARA* (Anytime Repairing A*) algorithm is implemented to the footstep planning for real time calculation. ARA* is one of the anytime algorithm to allow tuning heuristic search. By finding a sub-optimal solution in early time, it modifies the solution in the permitted time. The footstep patterns are limited in the conventional method, but since the search of the allowable landing point around the candidate of the footstep pattern is implemented, the pattern of obstacle avoidance is found easily without detour. The simulation result shows the effectiveness of the proposed teleoperation system.

In this paper, we developed hexapodous quasi-passive dynamic walking robot “Hexapod-Duke”. Experimental results show that this robot can walk with tetrapod gait shown by small creatures like insects on the level ground. Through investigation of walking experiments, we confirmed that the roll angle trajectory of the middle leg component and that of sum of all leg components superimposed are same and the swing of hind leg component was the largest among all legs. As a result of analyses, we conclude the effect of the middle leg has the biggest influence on how the robot walks in the hexapodous quasi-passive dynamic walking and considered why the robot's swings of hind leg component is the largest of the three.

This paper proposes hexapod robot gait control based on Follow-the-Contact-Point (FCP) gait control and human-in-the-loop control. Robots need surveying and recognizing a passage plane and planning a path and moving in unknown environment. However recognizing and planning are difficult and take a time for robots, so human-in-the-loop control is took into account. In order to put it into practice, HMI and the hexapod robot which equips FCP gait control, depth sensor are developed. By human deciding contact points and planning path, the hexapod robot can walk on uneven terrain. And we can get data of evaluation about passage planes and to plan a path like a human.

An omni-directional static gait for our working six-legged robot is proposed. This robot can stand up with four lower legs and can use two upper legs as arms; its body is vertical. The robot walks with four legs statically in this upright posture. The center of gravity of the robot in upright posture is high. The robot inclines its body by moving up/down three support legs, thus shifting the center of gravity above the support polygon. Then the robot raises one swing leg, moves its body in any direction, and grounds the swing leg. The experimental results show that the robot can walk in any direction by the proposed method.

In this paper, a new method setting workspace for tripod gate is proposed to realize smooth walking. Generally, frequent stepping change of the leg occurs when turning direction near the movable limit of the work space. To avoid the stepping request, the buffer area is provided to the outside of the work space of the support leg. The proposed method is implemented to the real hexapod robot, and the effectiveness is shown by comparing the energy consumption

In this paper, we propose the method of tomato fruit and pedicel recognition method using 3D camera for tomato harvesting robot. Recognition method is constructed by 2 parts; tomato fruits detection and tomato pedicel detection. To avoid detecting stem and pedicel of neighbor fruit, we propose a robust estimation method in pedicel direction calculation. We evaluated the proposed method in the greenhouse with real tomato trees, experimental result shows its robustness.

In recent years, aging of agricultural workers has progressed, attention is paid to ICT conversion of agriculture and Functional food with health maintenance effect. Raspberry contains a large amount of compounds that have a high health maintenance effect, which takes time and labor for cultivation. The purpose of this research is to construct an image processing system that recognizes color and position information in order to check whether harvesting of raspberries is possible under an indoor environment. As research contents, RGB values of colors of berries at various stages were calculated from spectral analysis results of berries and compared with actual fruit photos. We also classified colors in the image using machine learning and verified its accuracy.

The purpose of this study is to develop a measurement system for smart forestry by using a multicopter with a three-dimensional laser scanner sensor. The measuremnt by the laser scanner on which a multi-copter flying requires attitude correction of measurement data to improve presicion. To acheive it, we utilize IMC sensor's information.

Japan is a forestry country where forests account for about 70% of the land, and introduction of robots is expected in the field of forestry. However, there is already a method of measuring height using aircraft and laser scanner, but the diameter of breast height is investigated by human hands despite the lack of personnel. Therefore, in this research, we propose a trees diameter measurement system using multi rotor helicopter. We aim to stably fly through the forest and measure the breast height diameter of trees in the forest. We propose a tree detection algorithm using LRF data.

This paper presents study of the robot which is able to harvest tomatoes. To harvest the farm products such as tomatoes is performed by farmers now. So many farmers have problem that much cost and time are spent. Therefore, to solve this problem, we developed the robot that performs tomato harvesting. We made the robot which can harvest the tomato on the rails last year. So we will make it possible to run in rough terrain.We designed the base part of the robot by considering the inclination of the vehicle body when the robot go up the stair and impact which the robot will receive when the robot go down the stair. In addition, our laboratory took part in a tomato harvesting robot competition held in Kyushu Institute of Technology in December 2016.

Agricultural population in Japan is decreasing and getting aging. With these matters, the ratio of the elder people who have accidents during farmwork is getting so much. As for the number of accidents in agricultural machines, a case of string trimmers is most. The Japanese government has been promoting automation for agricultural machine to prevent the accidents and realize work-saving and safety for farmer. This research describes a mobile mechanism for a weed cutter robot on levee slope. The proposed mechanism has variable camber angle structures for each four wheels. The usefulness of the proposed system was confirmed by experiments which have uniformly slope angle with artificial turf. At the end of this paper, modelling for this system was performed by using bicycle model and the validity of the model was evaluated.

This paper reports the design of the hind legs of the four-legged robot we are building. Each hind leg of the robot has three joints, and a motor to drive the ankle joint, which is the most distant from the body, was installed on the link between the knee and the ankle joints. However, a heavy motor on such a remote position from the body increases the moment of inertia of the leg, which prevent fast movement and demands the motor driving the hip joint a large amount of torque. To solve the problem, we adopted a new design, in which the motor for ankle joint was installed in the body. We experimentally confirmed that the error of the ankle joint decreased.

Many robots for searching in uneven ground have been developed. However, carrying robots in uneven ground is dangerous because of risks that operators may be injured. If robots are able to throw in uneven ground from safety area, it is safe and easy to carry robots. In addition, they are possible to reach the unreachable areas in step overcoming ability of robots. The purpose of this study is developing a throwable robot for search in uneven ground. Thus, we have developed quadruped robots with spherical shell which protects legs from external force when they land. In this time, we developed a new robot which has better impact resistance and step overcome ability than preceding model. In this paper, we report modeling new model and examination of improvement that step overcoming ability.

This study aims to develop a quadruped robot capable of walking and running at various speeds based on biological concepts. The quadruped robot is equipped with a mathematical central pattern generator (CPG), consisting of four coupled neural oscillators, basically producing a trot pattern. The system based on biological concepts contains sensory feedback to the CPG, measuring the body tilt. This spontaneously gives rise to an unprogrammed walk at low speeds, a gallop at high speeds, in addition to trot at medium speeds. We also confirmed the robustness improvement using the system. We assume that the oscillations of the four legs produced by the CPG and the body oscillation varying according to the current speed are synchronized along with the varied phase differences to keep balance during locomotion through postural adaptation via sensory feedback. As a result, each gait was created and the robustness improved.

The purpose of this study is to clarify what kind of influence of flexible trunk structure on its gait or gait transitions of quadruped locomotion. First, we developed a quadruped walking robot whose trunk part can be replaced easily. Each leg of the robot was controlled by independent phase oscillators and implemented TEGOTAE-based control, which generated a variety of gaits by making good use of the interaction between its body and the environment. We conducted walking experiments using trunks that have various degrees of freedom while walking speed was gradually changing and analyzed gaits of the developed robot.

Various robots aim to search information in uneven ground have been developed. Carrying robots into uneven terrain has a risk of injury for operators. If robots can be thrown into uneven surface, the operation will be safe and easy. We have been developing a quadruped robot with spherical outer shell, called "QRoSS" which is able to land on the irregular ground. There is flat ground in the disaster site and rotational locomotion is faster than walking on the flat terrain. The object of this research is to construct rotational locomotion as one of a complex operation. The moving speed of rotation was derived from the dynamics model.

Energy loss is a problem in legged robots. Energy loss in a legged robot occurs in actuator's negative work and collision with the ground. It is said that in animals there are mechanisms to reduce the loss of mechanical energy by elasticity in a body. This paper investigates the relationship between the energy loss and the each passive trunk joint and motion of the leg.

In current agriculture, demand for efficiency by automating agricultural work is increasing. For that reason, there are researches about automating agricultural work such as automate traveling of agricultural vehicles. In this study, to implement autonomous travel system for hovercraft for agriculture, we designed and developed a self-position estimation system using inertial sensors. The developed system consists of acceleration sensor, gyro sensor and microcomputers and estimates a self-position using acceleration and angular velocity obtained from these sensors. In order to evaluate the performance of the developed system, we installed it in the hovercraft and experimented in paddy fields and swimming pool. Afterwards, the self-position of the hovercraft was estimated using the data obtained by the experiment.

Currently, the proportion of senior citizens involved in agriculture is increasing. As a result, the state of the art in mechanization and automation of agricultural work is progressing. In regards to tilling works, walking tractors for use in home gardens still require pushing while plowing. It requires considerable physical strength to operate walking tractors and this could pose problems for women and senior citizens. Therefore, in this paper a small tillage robot for home gardens is developed. The developed robot is small and can be used in tight spaces. In addition, the need for pushing is eliminated by using remote control. Various agricultural crops can be accommodated by adjusting the tillage depth. The robots' depth adjustment capability is inspired by the foraging motion of the mallard during feeding. As a result, convenience, safety, and multi-functionality can be ensured for cultivating crops in home gardens.

We have developed a weeding robot for many years. This robot weeds by moving in the rice field. It has a unit of Global Positioning System (GPS), in order to move automatically in the whole field. However, positioning by GPS is not accurate well. In this study, we investigate factors for inaccuracy GPS positioning through experiments in order to improve positioning by GPS. From experimental results, in independent positioning, individual differences are a major factor of positioning dispersion. On the other hand, in positioning by Real-Time Kinematic-Global Navigation Satellite System (RTK-GNSS), positioning errors are within 50 cm. In case of these errors, the robot can estimate its position by filtering algorithm. Therefore, the robot can move automatically in the rice field.

In our laboratory, we have developed a remote controlled mower robot, AMR-D01. However, it has several weak point such as it takes time to work, grass which is not mowed hit against front body because its blade diameter is small. In this paper, an oscillating mower mechanism for mower robot is proposed in order to increase the swath of the robot. This mechanism is based on Lever-crack mechanism, which has a rotary link, a transmission link, an oscillation link, a fixed link. Characteristics of the mechanism, such as angle of oscillation and quick return ratio of the oscillation link, required torque of the rotary link, with ratio patterns of length of each link is evaluated by simulating. From the results, proper design and parameters are selected. Proposed mower robot have the advantages of reduced mowing time and costs using the oscillating mower mechanism.

Raspberry fruit contains a large amount of compounds with high health maintenance effect, so it will be a high added value product. This will save the declination of agriculture in Japan. With that back ground, we propose to automate cultivation and harvesting of raspberries. In this paper, especially, we describe our proposed automated-harvesting system.

In recent years, the introduction of robot technology to the primary industry is very important, and introduction to the agricultural field is also expected. Among them, tomatoes have a long harvesting work time and high labor load. Therefore, automation of harvesting by robots is desired. In this research, we are targeting tomato production facilities with large-scale infrastructure. We are developing tomato harvesting robot aiming at practical application at Hibikinada Greenfarm Co., Ltd.. In this paper, we focused on end effector for harvesting tomato. When harvesting tomatoes, it need to have a stem end. In the previous study, it was pluck a tomato from a stem, but stem end took off from tomatoes. In order to harvest a tomato from cluster with stem end, we developed an end effector using suction mechanism and cutting mechanism.

Quadruped robot has better stability of posture and walking than bipedal walking robot. However, there are certain conditions to get stable posture. As we regard the foot in contact with the ground as a point, the supporting leg polygon is made by connecting the points with lines. If the centroid projection point of the robot is not in the polygon, the robot falls. And, as the center of gravity projection point is closer to the side of the supporting leg polygon, reduces the stability margin I become unstable. Therefore, in order to perform a stable posture, feet position is limited. As a result, freedom of attitude is limited. So, we will it is possible to obtain the inertial force by rotating a disk, to be used inertia rotor. We designed the optimum regulator based on the experimental machine we made and simulated it.

A robot that acts in a human living space needs the ability to surmount plains, steps, irregular terrain. I thought that a walking robot with high running performance was more suitable. An overlapping articulated walking robot having degrees of freedom in which the first joint axes of the lower leg and the upper leg are arranged on the same straight line acquires a stride which takes full advantage of the movable range of the leg without enlarging the robot It is possible to do. Based on the above, this research proposes an overlapping articulated quadruped walking robot adaptable for stair climbing as a representative obstacle in human living space. In this paper, we report the fact that legs with multistage telescopic mechanisms indispensable for that purpose were produced and verified.

Quadrupeds exhibit versatile gait patterns in response to morphology, environment conditions, and locomotion speed. The goal of this study is to understand the mechanism of adoptive interlimb coordination underlying quadrupeds. In our previous study, we designed interlimb coordination rule from the viewpoint of Tegotae from body support and propulsion. Tegotae is a Japanese concept describing how well a perceived reaction matches an expectation. As a result, we reproduced gait transition from walk to trot to gallop via two-dimensional simulations. In this study, we designed a physical quadruped robot in order to verify the proposed model in real world. Particularly, we focused on limb structures that can distinguish between vertical and horizontal components of ground reaction force. In addition, we designed compliance of the limbs for high speed locomotion.

This paper proposes a method for decentralized walking control of a centipede-like multilegged robot with active inter-segment joints. Based on plannnig the path of the head segment, the following segments walk with generating a consistent desired trajectory by using a cubic spline curve. Both trajectory generation and control for legs and inter-segment joints are performed only from local information. Finally, we show the availability of the proposed method by simulation using v-rep.

In this paper, frictional constraints of biped robot’s sole is investigated and verified. Conventionally, the studies of frictional constraint in walking have only considered in translational slip. But there are not only translational slip but also twist in actual biped walking. In this study, a method to calculate the frictional force and torque is suggested as the quantitative expression of frictional constraint based on a biped robot model. The frictional constraints were verified the proposed calculation approach of the frictional force and torque by generating pseudo slips.

Electric fences are installed to protect agricultural crops from animals. But, it becomes useless when electric leakage occurs due to growth of weed. Therefore, we developed a device for detecting and notifying the leakage, and developed a self-propelled weeding robot which automatically mow the weeds along the electric wire in order to prevent leakage. As a result, leakage detection and notification became possible when the voltage of the electric fence was lower than the reference voltage. Also, it enabled self-propelled weeding robot mow the weeds along the electric wire.

Accidents in farm work by brush cutter have reported annually. In this study, to aim at development autonomous mowing system, the method for mowing system is proposed to implement autonomous running by range specification and verify the effect of the method on actual environment. This method includes target track generation that generate the zigzag line in specifying range, moving method limited rectilinear propagation and pivot turn, and self-position estimation separated for only during rectilinear propagation and for only during pivot turn. Finally, the demonstration experiment is performed on actual environment.

In multiple robots environment, it is an essential technique that the robots avoid collisions with each other. To achieve efficient collision avoidance, it is desirable that each robot avoids surrounding robots considering not only their positions but also their behaviors. We have developed spatially seamless local communication system in order to exchange behaviors between the robots. This system ensures spatially seamless communicable area towards robot's surroundings by rotating a transmitter and a receiver. However, a bit rate of communication using the system is insufficient. Although the system has redundant frame structure to detect interference between signals from different transmitters, it causes decrease of the bit rate. In this paper, we aim to improve the bit rate by reducing frame length of transmission pulse while retaining interference detection function.

In this study, we propose a prototype test system for variable flexible spacecraft by using multi robots. The system is designed for gravity compensation of the prototype model of spacecraft which has deployable membrane structures or flexible structures. The system is composed of multi robots. Each robot has parallel mechanism to follow the fast motion of the structural vibration and deployment. The robots are controlled by the force control to support the structures as well as the decentralized autonomous control. The validity of the proposed system and the most suitable support point for this system are verified by numerical simulations.

This paper discusses cooperative transport system using three position controlled manipulators. However, cooperative transport using multiple manipulators has several problems. The most important problem is an internal force occurring in a conveyed object due to mutual position error of the manipulators. Therefore, we suggest a cooperative transport system using wires as mechanical compliance to solve the problem. The object is fixed by using six wires and can build the system which internal force does not produce. In addition, we devised the correction method of object's position and posture error at the positioning point due to the position error of the base of manipulators that occurred when transported the object.

Legged animals exhibit adaptive and resilient locomotion through their inter-limb coordination. Our motivation is to elucidate the inter-limb coordination mechanism underlying various legged animals from a unified viewpoint. Thus far, many studies have focused on animals with a few legs, e.g., bipeds and quadrupeds, and those with many legs, e.g., myriapods. However, the two extremes were not connected because those with intermediate number of legs, e.g., isopods, have been less studied. Accordingly, in this study, we focus on the locomotion of sea roaches which have 14 legs. As a first step, we observed the locomotion of sea roaches and found that they change their gait pattern depending on their walking speed and morphology.

Animals exhibit versatile locomotion patterns in response to the environment. To understand the underlying decentralized control mechanism, this study focuses on centipedes which exhibit highly versatile locomotion patterns. It is known that centipedes move by using their legs and body undulation. In addition to this, we report here that centipedes can move by combining expansion and contraction of the body trunk with leg motion when they climb up slippery sticks. We hypothesized that this behavior emerges by exploiting local forces generated at the junction between the body trunk and the legs as well as the ground reaction forces. On this basis, a simple autonomous decentralized control rule is proposed, and it is demonstrated via simulation that the above-mentioned behavior can be reproduced by the proposed mechanism.

Many animals locomotion is emerged by their autonomous distributed neural network called CPG (Central Pattern Generator). The purpose is to elucidate the common interlimb cooperative mechanisms for the locomotion with different number of legs using a multi-legged robot in which interlimb nerve coupling are eliminated and CPG is isolated on each leg (isolated-CPG). This paper reports that an interlimb cooperative walking was emerged by new control method using a actual robot.

In this study, towards the neurophysiological mechanism elucidation of the aggressive behavior of crickets, in order to examine the internal state of the aggressive behavior. Therefore, we have developed a small mobile invasive device that can intervene in aggressive behavior of cricket. In this paper, by useing a small mobile invasive device, it is showed that the locomotion of crickets greatly influences the probability of aggressive behavior.

Animals can change their locomotion patterns in response to changes in their body properties. Clarifying the underlying control mechanism will help develop highly adaptive bio-inspired robots. To address this issue, this study focuses on undulatory locomotion of animals having one-dimensional body structure. They propagate bodily waves to attain propulsive force effectively during undulatory locomotion, yet the direction of the wave propagation depends on the friction property between the body and the environment. To understand the essential mechanism underlying the friction-dependent wave propagation, here we propose a decentralized control mechanism by focusing on “Tegotae”, a Japanese concept describing how well a perceived reaction matches an expectation. Specifically, Tegotae function, which quantifies Tegotae, is defined based on tangential and normal component of the friction force, and the control scheme is designed such that the Tegotae function is increased at each local controller.

Snakes can alter their locomotion patterns in response to the environmental changes. Inspired by this ability, various snake-like robots have been developed thus far. However, they could not reproduce the innate behavior of real snakes. To address this issue, in our previous work, we proposed an autonomous decentralized control scheme based on Tegotae, a Japanese concept that describes how well a perceived reaction matches an expectation. It was demonstrated via simulation that two qualitatively different locomotion patterns, concertina locomotion in narrow aisles and scaffold-based locomotion in unstructured terrain, can be reproduced by using the Tegotae-based control without any parameter change. In this paper, we developed a robot to validate the Tegotae-based control in the real world.

Progress of artificial intelligence is remarkable, but application to a safe system has not been sufficiently discussed. We propose the following as "" and describe problems to be expected. (1) Apply AI independently as safety-related system. (2) Apply AI as part of safety-related system. (3) Apply AI as a human substitute.

In the safety of service robots, we clarify the differences and problems from current functional safety and machine safety frameworks when using artificial intelligence that learns and infer such as deep learning. Specifically, we discuss the difficulty about errors and faults and calculation of their probabilities, which were necessary for conventional safety analysis.

In order to develop safe robots, it is important to carry out risk assessment without fail. In order to realize risk assessment with few mistakes it is essential that specifications are clarified. It is effective to use SysML in order to clarify the specification of robot of concept level. In addition, we created a model diagram necessary to develop modeled systems for risk assessment. In this report, we focus on the battery system and describe its procedure.

In recent years, CAN (Controller Area Network) is used in various situations to control a distributed system. For example, CAN is implemented to automobiles, autonomous driving vehicles and medical robots. The adopting of CAN will continue for a while. However, CAN has vulnerabilities and there is a possibility of receiving unauthorized access from the outside of a system. It is necessary to design a system considering vulnerability of CAN. In this research, we conducted attacking experiment to a CAN system. First, we constructed a CAN system for a mobile robot. Then, we attacked a specific node by using DOS attack. Finally, we demonstrate spoofing of the node after DOS attack.