We propose a novel consistent control of both the legs and the inter-segment joints for a centipede-like multi-legged robot. The proposed method is integration method of Follow-the-Contact Point (FCP) gait control for the leg control and control for the inter-segment joints. Finally, the usefulness is verified by the result of physical simulation of the robot.

View full abstract

The purpose of this study is to generate the walking gait for 2 legs robot in the operating field with obstacle. It is expected that 2 legs robot can avoid obstacle more smoothly in the same way that animal and human adjust stride normally to step over obstacle. Stepping points are determined optimally under the condition of maximizing walking speeds or minimizing energy for walking. The problem of gait generation is reduced to a combinatorial optimization problem solved by using genetic algorithm. Orbits of toes and hip between stepping points are generated by using conjugate gradient method. The stable walking condition is obtained by adding the control of the upper part of the body.

View full abstract

In this paper, we propose a joint stop mechanism in addition to the cross link mechanism in our previous research. This mechanism consisting of the joint stop, torsional spring and spacer is introduced into the knee joint of our bipedal robot. The aim of this research is to address the reduction of power consumption in walking by the proposed mechanism. The function of joint stop mechanism is to temporarily fix the knee joint to a certain angle when performing extension walking. As a result, we succeeded to cut down energy consumption by assisting torques to actuators by the joint stop.

View full abstract

Recently, many biomimetic quadruped robots have been developed. Most of them use many actuators for running, which causes them to become big in size. We have developed a middle size of quadruped robot by using only one electric motor as an actuator for running. We also adopt to the robot the mechanism which is similar to Theo Jansen mechanism to make its gait biomimetic. In order to put our mechanism into practice, we propose "Modified Gallop Gait" that is derived from bounce gait and gallop gait. The results of the experiment show that the robot runs as the modified gallop gait while its pitch angle oscillates periodically to realize its stable running.

View full abstract

This study deals with sweeping algorithms of robotic vacuum cleaner appliances. Specifically, some sweeping manoeuvres of such robots in market are discussed and evaluated by numerical simulations. Then, they are tested by experiments employing small mobile robot.

View full abstract

We have been developing the lightweight and small size cleaning robot. This robot consists of a fan which draws up dust, the tires with the motor, an obstacle avoidance mechanism and the various sensors such as the infrared sensor, and could move the top surface of the thin plumbing, and could clean up that surface. By using this equipment, inefficient and dangerous work at the high place can be done in a safe and in a short time.

View full abstract

This paper describes a path planning that considers the robot shape with a transported object for an omnidirectional transportation mobile robot. The path planning considering the robot shape is usually treated in a Configuration Space. However, a mapping to the Configuration Space needs much computational complexity to check the collision of the robot and obstacles. Therefore, we propose a method for reducing the computational complexity using the domain decomposition according to the collision condition that the robot posture depends on or not. The effectiveness of the path generated by combining a distance transform and a potential method is verified through some simulations.

View full abstract

The maintenance and management of mountain forest needs special knowledge and skill, and contains various dangerous tasks. Thus, the shortage of workers causes the problem of increasing unmanaged forests. In this research, we are developing "Teleoperated mountain forest maintenance robot system", which assist human work semi-autonomously. In this paper, we will discuss and propose a performance improvement method in recognizing process of the surrounding objects in the mountain forest environment. The proposed method is based on the data integration of the distance and the received light intensity which are obtained by LRF(Laser Range Finder). The experimental results illustrate the validity of the proposed method.

View full abstract

This paper presents a concept of aerial remote inspection robots for steel structures. Unlike conventional aerial robots, the presented design allows active contacts to the target structure and conducting vibration testing while staying on the structure. Two types of vibration probe feet, i.e., electric-field-driven and magnetic-field-driven probes, which can adhere to the surface of the target and give vibratory forces, and the corresponding principles of damage detection are presented.

View full abstract

In recent years, production operations by robots are increasing in the industrial field. However, it is important to ensure the safety of a human-cooperation-robot, because it is at risk for injuring on a human by clashing. Besides, unmanned transport vehicles become popular in the industrial field. Using unmanned transport vehicles as human-cooperation-robots enables transportation of heavy parts and burden reduction. In this study, an omni-directional cart is used as a human-cooperation-robot. The purpose of this study is to develop of disturbance estimation system with disturbance observer in omni-directional cart. Thereby, the external force applied to the cart can be estimated. Then, operating cart appropriate to the external force, the high-security omni-directional cart is developed.

View full abstract

Conveyers are widely used in transportation and manufacturing fields. In case that a conveyed object requires changing its traveling direction, conventional systems used moving flippers, jack-up sliders, etc. Other systems using omnidirectional wheels or rollers with other actuators are also proposed. In contrast, a method using special roller mechanism is proposed in this paper. The mechanism called Partially Sliding Roller (PSR) has one active (or passive) rolling DOF and one semi-passive sliding DOF along rolling axis. Several PSRs can not only transport object but also shift actively to rolling axis direction with the passive DOF. The idea of PSR followed by models of proposed conveyer mechanism and brief experiments to evaluate the method are described.

View full abstract

This paper describes a systematic method for design and implementation of distributed autonomous controllers in cooperative multi-robot systems. Based on our previous concepts of non-centralized grouping robots, and bionic self-reproducing, self-assembling, and self-organizing molecular mobile robots, a totally distributed coordination algorithm is developed to provide robustness to multi-robot manufacturing and assembly missions using Petri net based formal modeling methodologies. In a bionic assembly factory environment, where a number of autonomous mobile robots function with own controller and inter-robot communication facilities, each robot plans the subtasks sequence, decides the workstation for the next assembly step with general knowledge of the plant layout, and at the station performs the assembly subtask autonomously or cooperatively with other robots, avoiding resource competition and sharing operation experience. Embedded Petri net models of autonomous task execution and implementation issues are presented for a prototyping experimental system.

View full abstract

We have proposed high accuracy linear positioning mechanism using singular configuration of manipulator. First, we proposed basic mechanism using singular configuration. After that, we developed linear positioning mechanism based on the basic mechanism. By analysis of the theoretical precision of the mechanism, we show the mechanism can do positioning more accurately and speedy than simple crank slider mechanism and linear actuator. In this paper, we confirm the accuracy by peg-in-hole experiment. In this experiment, we have realized high-accurate positioning with high-speed motion between two holes.

View full abstract

The purpose of this study is to develop the high speed checkweigher with electromagnetic force balance system. We approximate the checkweigher by a spring-mass-damper systems as the physical model, and the equations of motion are derived. The validity of the proposed model can be confirmed by comparison of the simulation results with the realistic responses. The proposed model offers practical and useful information to examine control scheme.

View full abstract

Demands for the development of soft robots are now increasing because they can work well under unstructured environment by deforming their bodies. However, systematic design methodology for controlling soft robots with a large number of bodily degrees of freedom has not yet been established. To address this issue, we have focused on Plasmodium of true slime mold, which moves adaptively by deforming their amoeboid body. In our previous work, we constructed a model that allows large deformation of the body on the basis of autonomous decentralized control. In this study, we developed an amoeboid robot to validate the proposed model in the real world.

View full abstract

Many quadrupeds change their gait from a walk to a trot, and finally, to a gallop depending on the locomotion speed. These locomotion patterns are generated via the coordination between limbs, i.e., interlimb coordination, and are partly controlled by an intraspinal neural network called the central pattern generator (CPG). However, a proper understanding of interlimb coordination mechanism remains elusive. Here, we demonstrate the gait transition experiments with a simple quadruped robot, which limbs are coordinated by an unconventional CPG model that we previously proposed. By only changing a parameter of locomotion speed, our robot exhibited spontaneous gait transition from a walk to a trot, and then to a bound, even in the absence of direct neural interaction between the oscillators of the CPG. Very interestingly, we found a canter -an asymmetric, three-beat gait- in the transition from a trot to a bound. In this paper, we discuss the generation mechanism of canter gait.

View full abstract

Snakes exhibit adaptive and versatile behavior under unpredictable real-world constraints. Although numerous snake-like robots have been developed thus far, they could not reproduce the innate behavior of real snakes. To address this issue, we previously proposed an autonomous decentralized control scheme in which three-dimensionally effect is added to curvature derivative control. In this study, we developed a robot to investigate the validity of the proposed control scheme.

View full abstract

Snakes can exhibit various locomotion patterns in response to the environment. To elucidate the mechanism underlying this versatile locomotion, we previously proposed a model based on autonomous decentralized control and showed via simulation that it could reproduce various locomotion patterns. Further, we extended the model to reproduce concertina locomotion in narrow space as well. In this paper, we developed a robot to validate the extended model in real world.

View full abstract

Most robots are vulnerable to changes in their own morphology such as failures. To tackle this problem, we have focused on ophiuroids, which have simple body with pentaradial symmetry, as our model. Although they are composed of primitive decentralized nervous system, they can coordinate their arm movements adequately even when the arms are arbitrarily cut off. In this study, we aimed to clarify the fundamental mechanism underlying this resilient ophiuroid locomotion. For this purpose, we performed behavioral experiment using ohiuroids with different number of shortened arms, and proposed a simple autonomous decentralized control scheme on this basis. The validity of the propose control scheme was confirmed via simulation.

View full abstract

To make an exploration with multi-robot system, it is important to calculate each robot's position, that is, to realize a localization. However, the growth of the number of robots will cause a computation explosion. Then, we proposed the simple localization method for a multi-robot system which consists of a big number of robot in our previous work. In this paper, we analyzed the error of the multi-robot localization with our proposed localization method. Especially, we analyzed not only the absolute distance error but also relative distance error and concluded that this method is successful whenever a small number of Land-Mark nodes are used.

View full abstract

Insects exhibit surprisingly adaptive locomotion under unstructured environments despite their limited computational resources. Such locomotor patterns are generated via coordination between leg movements. Thus far, numerous studies have been devoted to elucidate this mechanism from biological as well as robotic viewpoint; however, it has not yet clearly understood. To tackle this problem, we employ a unique approach: we reconsider interlimb coordination mechanism of hexapod locomotion from quadruped locomotion. In our previous study, we proposed an unconventional CPG model of quadruped locomotion that fully exploits physical communication between legs via load sensing. Our quadruped robots with this CPG model well reproduce various gait patterns observed in actual quadruped animals. Based on this result, we intend to apply the CPG to the interlimb coordination of hexapod locomotion.

View full abstract

In this paper, we propose an autonomous distributed control method for modular robots that operates based on only local communications. The robot's task is self-reconfiguration to mimic the shape of the nearby object detected by proximitysensors (mounted on the modules). Each module decides in only local communications the need for action and the direction of movement, and create meaningful operations for the system. We designed a two-dimensional lattice type modules of three-layer structure. Deformation mechanism uses a slide mechanism. The communication among neighboring modules uses the optical communication of the phototransistor and LED. The availability of the rule is demonstrated by simulations.

View full abstract

In this paper, we propose a modular robot system generating actions based on photo-detection. We particularly propose the chain action generation by propagating light. Many of the modules consist of non-identical modules, easily replaced and connected. We make up autonomous decentralized control system by generating motivations of the modules to transform local structure with LED and photo transistor. In the first, by communicating light between the electrically-independent modules, the modular robot acts meandering motion without microcomputer. Herewith, the modular robot is downsized, power saving, low-cost, excellent environment adaptability and under simple control. In the future, the modular robot will snake narrow and winding road by proximity sensor.

View full abstract

In this paper, we investigate gathering control for sheepdog system by multiple mobile robots. This work was motivated by so-called sheepdog harnessing, in which a mobile robot is developed to gather a flock of ducks or sheep and maneuver them to a specified goal position. Based on discrete model of sheepdog system, we examine the validity of cooperative control in order to drive all sheep to one goal position and that of distributed control in order to maneuver them to a different goal position.

View full abstract

Development methods of the mobile robot system for simultaneously performing system improvements and practical use have been proposed. The system has constructed the architecture of a mobile robot with a variety of functions as a method of realization. And, it was exemplified usefulness and practicality as architecture oriented design.

View full abstract

A visual monitoring with mobile robot has broad applications to enhance a route navigation, complex task implementation, and simply recording of the human and robot activities etc. However, an efficient camera work has belonged to the art of human skill. In this study, we introduce three simple evaluation factors, which allow the monitoring robot to capture an adequate picture composition for better task implementation. Also, we demonstate a dynamic localization of monitoring robot in search of better picture composition.

View full abstract

Visual recognition in multi-robot systems has a peculiar problem in that an observation from a different viewpoint gives a different perspective. Also, a representation of the same target object is highly affected by such a viewpoint or environmental condition. Hence it has become a challenge to realize cognitive sharing of the object or ROI between the robots without a help of artificial markers. In this paper, we propose consensus-making algorithms to acquire the viewpoint-invariant representation of color, shape and geometrical relations. Experimental results show the ability of cognitive sharing significantly improved by the proposed method.

View full abstract

This paper describes a behavior acquisition by a composite artificial neural network (CANN) and a region divided artificial neural network (RDANN). The CANN includes a high-level ANN, which is used for deciding which low-level ANN is activated to calculate output signals. The RDANN is a special case of the CANN, in which the structure or the selection function of the high-level ANN is given by a designer. To examine the relation between parameters of the high-level ANN and learning effectiveness, a CANN is applied for pole balancing problems. The controller, a CANN, a RDANN or a common ANN, is optimized by the particle swarm optimization. Experimental results show that the angle of the pole is important for selection by the high-level ANN. Results of additional experiment show that the RDANN that is designed to select low-level ANNs by the angle of pole is the most effective for pole balancing.

View full abstract

In object pushing manipulation by camera mouted two-wheels-robot, we propose a way of extraction and estimation of element that affect behavior of object from image feature. We verify the way of object behavior estimation based on estimation of movement of feature points with respect to robot motion.

View full abstract

This paper describes a novel neural network model for pattern recognition and behavior selection. In the case of constructing a system which can adapt to changing environment and achieve some task, it is necessary for the system to be able to predict current situation and select proper behavior by utilizing memory data and limited observed data. This function is realized by connecting three neural networks in series. Each neural network has a role of prediction, recognition and behavior selection, respectively. In this report, the neural network model is explained, and a simulation result is shown for confirming the idea.

View full abstract

In this report, we propose a reinforcement learning that a target agent can learn an unknown environment efficiently by applying a knowledge acquired by another source agent with a different form. Concretely, A Target agent has three tables, Target Q-table, Source Q-table and State-Transfer Table(STT), Target Q-table is unlearned and represented for Target's state-action pair, Source Q-table is already learned and for Source's state-action pair, STT is unlearned and for transferring Target state into Source state. It sets Target Q-table and (STT and Source Q-table in series) in parallel Target's Q-table and Source's Q-table have the almost same size. STT has a smaller size. We expect that the proposed method converges early in detail, since in the learning, STT with Source Q-table converges early but roughly, and Target Q-table converges slowly but in detail.

View full abstract

In this work, we considered the influence of task complexity in the shape evolution of the locomotive robot. Robots were evolved in a virtual physical environment for achieving various tasks. Acquired shapes of robots were classified in 4 types according to the number of legs.

View full abstract

This paper proposes a new predictive control system using recurrent RBF networks (RRBFN) and Fuzzy rules. This system is constructed from a prediction system and a Fuzzy control system. The prediction system predicts the state of the controlled object on time t+n. The Fuzzy control system controls the controlled object based on the prediction results of the long-term prediction system. We test the proposed method under an outfielder problem in order to investigate its efficiency.

View full abstract

We have confirmed that a reinforcement learning becomes more efficient with multiplex learning spaces, a whole and one or two of partial learning spaces. In this paper, we extend the number of partial spaces to N, and study about the learning efficiency. We investigate the learning inefficiency in a case of increasing the number of unavailable partial learning spaces in experimental simulations. We confirmed that increasing did not influence the learning in an inefficient way, and that only one available partial learning space influenced the learning in an efficient way. As a result, we found that more multiplexing becomes more efficient for the learning, even if the partial learning spaces include some unavailable ones.

View full abstract

A walking assistance apparatus which can promote gait training for motor palsy patients and outdoor walking for the elderly was developed. We proposed the control method according to the purpose of the user. For example, to promote gait training, this apparatus can select the impedance control method. User's legs are assisted along the trajectory with the ideal gait as well as walking. To promote outdoor walking for the elderly, this apparatus can select the torque control method. The apparatus calculates the necessary torques of each joint according to the posture and motion of the apparatus and user. In the case of the impedance control method, by adjusting the natural frequency of the ideal dynamic equation for the flat steps under the soles, the user felt it was easier to walk by frequency entrainment. The torque control method proposed to use real-time acceleration data instead of using the template trajectory. By using this method, the subject could walk on the slope and stairs continuously. From the result, the impedance control method is good for gait training, and the torque control method is good for walk promotion.

View full abstract

The control method of the contact motion in a robot arm for life support has been developed. We used a 2 DOF robot arm and FSRs for the detection of the contact. If the gripper closes and the contact location is the same as the working direction, the arm doesn't stop or return the motion at the time of contact. If the gripper closes and the contact location is different from the working direction, it stops or returns the motion at the contact. We showed the effectiveness of the proposed control method in the basic motion experiment.

View full abstract

This paper describes basic structure of a novel joystick system operated by tongue. This system is designed to give the force feedback to the user during interaction with the environment. The elastic plates adjust the supporting stiffness of the stick. Through this function, this system can give the user the different reactive forces according to the contact situation between the controlled device such as robot hand and the environment. We fabricated a prototype system and examined basic performances of proposed mechanism.

View full abstract

Dementia is one of the most serious problems in aging society. Aerobic exercise has good evidence for prevention of it. Walking and cycling are popular and practical aerobic exercises. However, elderly people have problems in the walking and cycling, which come from lacks of physical / mental functionalities and traffic environment. For them we have developed virtual reality (VR) walking / cycling system. In this paper, software system, which can easily and rapidly construct big-scale computer graphic data for the VR walking / cycling system, was described. The performance of the software was evaluated with benchmark tests. Some applications of the software are also introduced.

View full abstract

In this paper, we present a new in-pipe robot with independent underactuated parallelogram crawler modules, which can automatically overcome inner obstacles in the pipes. The parallelogram crawler modules are adopted to maintain the anterior-posterior symmetry of forward and backward movements, and a simple differential mechanism based on a pair of spur gears is installed to provide underactuated mechanisms. A central base unit connects each crawler module through foldable pantograph mechanisms. In addition, an operation controller to make the robot turn is developed. To verify the basic behavior of this robot, primary experiments in a pipe with different diameters, a pipe with a partial step, bend pipe, and T-branch were conducted.

View full abstract

In this paper, we developed a novel wheeled mechanism integrated with an actively actuated lug. The drawbar pull generated by using various lug sinkage lengths on the sandy terrain is firstly measured on a prototype mechanism. Based on measured results, we proposed a strategy of controlling lug sinkage length for reducing the fluctuation in drawbar pull. This method did not based on terramechanics model and need not measure physical properties of the terrain. The validity of the proposed method is finally verified by comparing the generated drawbar pull of developed wheel module with that of a wheel with a fixed lug.

View full abstract

There are many robots that work on sandy soil. Most of these robots are based on wheel mechanism equipped with protrusion devices called lug or grouser. These lugs are used for preventing slip or stuck, and improving the traveling performance of the robot. Lug's motion affects lug-soil interaction forces. Therefore, in this paper, we verify the relationship between lug's motion parameters and the traction force generated by the bulldozing lug by experiment. We focus on three motion parameters (inclination angle, moving velocity, sinkage length of lug) in this paper. We obtain approximate curves that describe how traction force changes during the bulldozing.

View full abstract

It is necessary to know the air inflow of a suction cup for the designing of a new passive suction cup, which not only could easily attach and detach, but also could keep long adsorption time. This paper proposes a method to calculate the air inflow and its speed, which is easy to implement since it only requires a pressure sensor. Before calculating the air inflow, the factors affecting it should be known. In order to find this factors, experiments are conducted to find the effect of diameter, material and pulled height of displacement of a suction cup.

View full abstract

We introduce the opinions of welfare service workers who conducted a social experiment of the elderly support community model with applying the elderly single population support sensor system for the preventing solitary death. When we did the interviews after this experiment to these welfare service workers, was able to hear a positive opinion to build a better society and to improve the welfare society in relation to this system of the sensor units and the local community network.

View full abstract

Home oxygen therapy (HOT) patient always needs to carry an oxygen cylinder at the time of going out. The present oxygen cylinder is not necessarily lightweight, and if it unites with a conveyance cart, it becomes as mass of about 5 kg. And this load becomes burden and reasons of giving up going out physically and mentally. So, in this paper, we describe the power assistive robotic cart development which carries the oxygen cylinder to reduce the load of HOT patients.

View full abstract

This paper describes the outline of a new "watch over assist system", which helps to prevent elderly to become bed-ridden, by promptly recognizing falling and simultaneously notifying to persons who give care. The system is featured by 1) Real-time falling detection by using low-cost temperature distribution sensors, 2) Walking gait feature sensing by using low-cost active infrared sensor, 3) Life pattern detection and behavioral analysis by using ubiquitous sensors, 4) Stuffed toy-like bi-directional friendly user interface for elderly persons.

View full abstract

In this paper, we analyzed gait motions with a prototype of a walking assist device "Raku-walk", for evaluation of the device. The Raku-walk is a new three-wheeled walker which can be used in sitting position. Using 9-axis motion sensors, normal gait motions without the Raku-walk and gait motions with the Raku-walk were measured and analyzed. From the results, differences between normal gait motions and motions with the Raku-walk were observed.

View full abstract

Most of robots working in nuclear power station adopt tether cable for communications. To prevent the cable from being entangled on the obstacles, cable reel mechanisms have been developed with dust-proof and waterproof designs for easing the cleaning of radioactive material from the robots. Problem of the existing cable reel mechanisms is that they may malfunction under large payload. Focusing on this challenge, we propose a novel cable reel mechanism (an oscillating level winder arm) with new designed dust-proof and waterproof mechanism. In addition to this, we make a fundamental experiment to inspect the validity of the oscillating level winder arm.

View full abstract

This paper discusses on a buoyancy and depth control module for underwater or amphibious robot. The buoyancy control applies recirculating compressed air ballast mechanism that consists of a cylinder, an external bladder called air bag and hoses. An air pump and a valve are installed in the cylinder. The system can manipulate the volume of the air bag by using the air inside of the cylinder to fill the air bag, thus to change the buoyancy force of the system. Then, pressure depth sensing method is used for depth controlling. This method uses pressure sensor to detect ambient pressure of a robot in the water and calculates its current depth. This buoyancy and depth control module will help robot to move up and down and provide depth control during underwater mission.

View full abstract

In this paper, a biomimetic approach based on central pattern generator (CPG) is utilized to control body shape of a snake-like robot for serpentine locomotion. The CPG model is based on the phase oscillator, in which only one control parameter is used to control the locomotion. The novelty of the approach is the online body shape transition of the snake-like robot which is useful in adapting to various space width especially for rescuing and searching tasks. The major parts of this paper include: 1) a brief introduction of the phase oscillator model and the proposed CPG network, 2) a proposed activation function to control the body shape transition of the snake-like robot, and 3) a turning motion and straight-path motion control strategy. Simulations are used to demonstrate its effectiveness with explanation on setup and control strategy for future experiment on physical snake-like robot.

View full abstract

Eccentric paddle mechanism (ePaddle) is a novel locomotion mechanism for accessing challenge amphibious terrains. A planetary gear transmission chain is applied in this mechanism. In this paper, we propose a systematic method for modeling the kinematic and dynamic models of this gear transmission.

View full abstract