This paper presents a new control model called the swarm leading for navigating multiple mobile robots as a swarm. In the control model, only a single robot, called a leader, knows a destination of the swarm, all the other robots, called followers, determine their motion by interacting with neighbor robots. If the motion of the leader is propagated to all the followers, the swarm is navigated. However, if some of the followers are left away from the other robots and isolated, they will be never navigated. Therefore, we need a stable control model that all the robots are navigated. In this paper, we develop the swarm leading control model and show the conditions that all the robots are controllable analytically using the algebraic graph theory. The proposed control model is verified with numerical experiments.

View full abstract

In this paper we describe our novel work of using graphic processing unit (GPU) to improve a homographybased visual servo system. First we propose a GPU accelerated Efficient Second-order Minimization (GPU-ESM) algorithm. With so-called "GPU-ESM tracking algorithm", our system can provide a fast and stable homography solution, approximately 20 times faster than the CPU implementation. To enhance the system stability, we adopt a GPU based Scale Invariant Feature Transform (SIFT) algorithm to deal with those cases where GPU-ESM tracking algorithm performs poor, such as large image differences, occlusion and so on. In this paper, the combination of both GPU accelerated algorithms is described in detail. The effectiveness of our GPU accelerated system is evaluated with experimental data. Key optimization techniques in our GPU applications are presented as a reference for other researchers.

View full abstract

Some living creatures not only have direct behavior and also its supportive indirect behavior. As for some spiders, they not only have direct behavior is to catch bugs directly and also supportive indirect behavior is to trap bugs. We focus on an artificial creature which can acquire building behavior for catching bugs in a 2D tile world. We do two kinds of experiments. The first is acquisition of building behavior for catching bugs. The second is acquisition of behavior by an artificial creature which can acquire direct capture and indirect capture. As a result, we show the possibility that evolutionary acquisition of building behavior for catching bugs and a relation between direct behavior and supportive indirect behavior.

View full abstract

Environment plays an important role of for behaviors acquisition in the field of Artificial Life (Alife) and evolving robotics. However, the conventional studies like K arl Sims' have not accurately considered influenced force by fluid. In this study, we approximately model the under-water environmental influence. And we propose a simulation method for artificial creature swimming in consideration of buoyancy and water drags as a virtual under-water environment based on the physics modeling. As a result of simulation, we verify that it is possible for artificial creature to swim like a real fish swimming behaviors in the under-water environment. Additionally we show the analysis of the acquired swimming behavior. Results are discussed on a relation between a shape of the artificial creature and an acquired swimming behavior.

View full abstract

There are many studies on hardware and software experiments to control autonomous robots. We have focused the elastic robot which has a body with multiple degrees of freedom. The elastic robot consists of homogeneous or heterogeneous simple modules. A feature of them is that their behavior is obtained by an each module movement based on local communications between connected modules. A physical simulation on the computer is carried out to achieve tasks autonomously for a virtual elastic robot. Our goal is to acquire a control system in evolution to achieve a composite task in specific circumstances with and without a obstacle.

View full abstract

In this study, we aim at evolving autonomous virtual creatures which have complex shapes in a complex environment. We implement a basic physics law and fluid influences with a virtual environment and evolve artificial creatures in plural environments such as on the ground, and in the water. The model, which is evolved in a different environment, obtains effective different moving behaviors, walking and swimming in each environment.

View full abstract

This paper investigates a particle swarm optimization (PSO) approach to perform a cooperative task in a multirobot system. PSO is known to be relatively easy to implement and efficient in solving various optimization problems. However, it has also been reported that the PSO approach is inefficient in solving large-scale multimodal problems. To overcome this unwanted characteristic of PSO, we have been developing a novel method that performs partial initialization in a small probability and demonstrating its basic effectiveness for benchmark functions. This paper further discusses a real world problem. We tackle the control of autonomous mobile robots by PSO. Similar to typical methods in evolutionary robotics, artificial neural networks (ANNs) are employed as controllers of each robot. PSO is utilized as a methodology for designing the ANNs. To examine our approach, computer simulations are executed using a cooperative box-pushing task. The results obtained are compared with standard PSO results. It is found that our proposed approach yields competitive results.

View full abstract

Most of snake-like robots adopt serpentine locomotion featured with smoothness and validity. But, the motion efficiency of the robot is far away from the one of the nature snake partially because of the mechanism and material differences. This paper represents that the slippage is one of the important reasons leading to low motion efficiency. To solve this problem, the passive wheel axis is suggested to be along with the normal of the serpentine curve during locomotion, which is determined by the discrete method of serpentine curve. Based on the analysis to the current discrete methods, a new control model using variable link length is proposed to improve the discrete precision. The control input of joint and link length are described. The influence of link length on the control input is also presented. This work lays a foundation to the development of new snake-like robot with the changeable link length.

View full abstract

The biopsy operation, especially in lungs, is essential to inspect whether some detected tumors are malignant or benign. But it requires a series of precise and careful actions which is time-consuming and the operator is easy to be exposed to over-radiation. We developed a distantly controlled robot for the CT-guided biopsy which is radiation-free. The robot system is easily and precisely controlled in distant from the radiation source and most actions of the biopsy operation can be automated. Also this robot has a specially designed needle insertion device which does less pain to the patient.

View full abstract

This paper describes robot that grows up. A human being changes the height and weight by growing up and the walking pattern changes along with it. Therefore, the purpose of this study is to reproduce human body growth and change in the motion pattern with the robot. A walking pattern for the figure of the robot is calculated from a technological viewpoint by the simulation.

View full abstract

Even if the ability of one individual is low such as an insect, a fish and a bird, are often takes orderly action by forming the swarm. This is called swarm intelligence. There are many researches of such swarm intelligence. Now there are many researches that it mainly uses simulation by using software. But the experiments of using the robot is a little. The purposes of this research are realization of which is based on group of nature with enhancing of crowd action to three dimensions by using airship and mobile robots, and analyze it. In this study, behavior of fish that form the swarm, became a group and lived are researched by using the airship and mobile robots. Using the robot which is equipped ultrasound sensors and pressure altitude meter and wireless communications. By these, the swarm intelligence was made into three dimensions from two dimensions.

View full abstract

This paper presents a novel design principle as well as its methodology for intelligent machine systems based on learning of hybridization in biology. In particular, we propose two principles for hybridization of physical systems: complementary principle and compatibility principle. The complementary principle states that two systems (A, B) can be integrated when A's strength corresponds to B's weakness, while B's strength to A's weakness. The compatibility principle states that two systems when they are integrated need to be adjusted at their physical component level to minimize their side effects. It is clear that the complementary principle gives a necessary condition for two systems to be integrated; while the compatibility principle gives a sufficient condition for integration of two components. These two principles are then elaborated with respect to a generic intelligent mechatronic system model, leading to three distinct hybrid physical systems: hybrid actuation system, hybrid mechanism system, and hybrid control system. Examples of the three hybrid physical systems are illustrated to highlight the two principles. The contribution of this paper is the proposed principles to design intelligent mechatronic systems through hybridization.

View full abstract

In this paper, we study a new method for robot programming: view-based teaching/playback. The motivation of its development is to achieve more robustness against changes of task conditions than conventional teaching/playbackwithout losing its general versatility. For proof of concept, the method was implemented and tested on a virtual environment. The method is composed of two parts: teaching phase and playback phase. In the teaching phase, a human operator commands a robot to achieve a manipulation task. All themovements of the robot are recorded. All the images of the teaching scenes are also recorded. Then, a mapping from the recorded images to the movements is obtained as an artificial neural network. In the playback phase, the motion of the robot is determined by the output of the neural network using scene images. We applied this view-based teaching/playback to pick-and-place and pushing in a virtual environment. Human demonstrated manipulation was successfully reproduced by a virtual robot hand with our proposed method. Moreover, manipulation of the object fromsome initial positions that are not identical to those in the demonstrations was also successfully achieved with our method.

View full abstract

The end effector of Space Station Remote Manipulator System (SSRMS) is a grappling device with flexible capture mechanism. It's developed by MDA Space Missions for large payloads grappling task in space. The capture mechanism, applying flexible cable structure to realize self-centering with lower impact, is the important component of the device and also a useful space mechanism. However, the dynamic process study of this mechanism is few reported in literature. Aiming to provide a contribution to latter research, the operation process of this mechanism is studied and molded in this paper, and the simulation results are also given.

View full abstract

In this paper, redundant manipulator control method of position and orientation for human-like movements is investigated. For orientation control, singularity-free quaternion representation is applied, and computation of orientation error with coordinate transformation is inspected. Control law for set-point regulation of position/orientation is proposed, which is based on the virtual spring-damper hypothesis, and the stability is studied using a Lyapunov-like energy function. Dynamic simulation is conducted and analyzed for validation of the proposed method.

View full abstract

Autonomous underwater vehicles have been used for relatively simple tasks such as underwater sampling, surveying and characterization of resources. One reason could be that underwater is very difficult environment to implement complex mission, but considering ground robots, underwater robot intelligence is not enough to achieve given missions. Korea Ocean Research & Development Institute (KORDI) just started a five-year project named "Development of technologies for an underwater robot based on artificial intelligence for highly sophisticated missions" from 2010 to study robotic intelligence for underwater robot systematically. This project will make a big step for 3D underwater SLAM (Simultaneous localization and mapping) under some reasonable constraints through AI based underwater environment cognition and mission based common platform technologies of hardware and software for commercialization. As the first year of the project, we are currently developing an underwater robot. This paper briefly explains the purpose of new project and addresses specifications and main features of new underwater robot as well as expected results. By the time of the conference, real pictures of the robot and experimental results could be ready to be shown.

View full abstract

Thrust is one of important performances of AUV. At designing, firstly the resistance of AUV at its maximum speed is estimated from its shape and then the thrust system, including propeller diameter, propeller RPM, driving system and required power is designed. But we can't be assured that the thrust system is rightly designed until we launch the AUV and measure its speed. If the data of the propeller open-water test is available the thrust and the torque of the propeller at certain speed can be estimated. In addition, if the motor's torque characteristics are available the maximum speed saturated by the induced propeller torque can be estimated. In this paper an easy method of estimating maximum speed of an AUV will be shown even in the case of gaining additional resistance from appendages of unplanned at the design stage. Furthermore by the method, the thrust performance changes by adjusting the diameter of the propeller will be easily investigated.

View full abstract

There are concerns about the impact that global warming will have on our environment, and that these will inevitably result in expanding deserts and rising water levels. AUVs (Autonomous Underwater Vehicles) were considered and chosen as the most suitable tool for conducting surveys concerning these global environmental problems. JAMSTEC has started to build a long range cruising AUV. The plan for its development is in several steps. The first stage was the building of an AUV, named URASHIMA, which was built in 1999, with sea trials being held since 2000. URASHIMA dived to 3,518m depth in 2001. At the end of February 2005, the vehicle was able to cruise autonomously and continuously for 317km. The second stage is our current development of several elementary technologies in way of preparation for making the next generation of vehicle. This paper introduces these elementary technologies.

View full abstract

This paper reports on the development of an automatic thrust redistribution algorithm for a redundant autonomous underwater vehicle (AUV) in the case of thruster faults. Under favorable conditions, the transformation of the thruster force vector to the generalized body-fixed force vector can be computed in a least-squares manner, which provides the minimum norm solution. In the case of some type of thruster fault, the redundancy in the system allows for seamless redistribution of the thruster forces while maintaining the desired vehicle motion. An accurate, fault-detecting thrust estimator has been previously developed that is robust to changes in source power and propeller ventilation. Such an estimator is necessary in order to quantify the health of the thrusters, which can then be used to weigh the contribution of each thruster to the vehicle motion in a continuous manner. An example of the thruster health index in its simplest form is the approximated thrust divided by the reference thrust, passed through a moving average filter, and normalized over all thrusters. This index can effectively scale back the contribution of any faulty thruster to the body-fixed, vehicle force vector solution.

View full abstract

In this paper, a mission controller for unmanned underwater vehicles is developed, which is implemented in a hybrid control architecture with the planning layer and the execution layer. In the execution layer, an ethology-based action selection mechanism is applied to the behavior-based control. By this mechanism, the controller can always choose the most appropriate behavior which maximizes robot's motivation. In addition, for the planning layer, a waypoint planner based on the Genetic algorithm is designed as a mission planner and the structure for the mission supervisor is proposed. Numerical simulations are conducted for verifying the performance of the mission controller.

View full abstract

In this paper, a novel method for the path planning and tracking of the Autonomous Underwater Vehicles (AUVs) using cameras is proposed. As the navigation of the AUVs is one of the emerging research areas in oceanic engineering, the importance of the path planning and tracking has been emphasized. Use of the visual data from cameras is one of attractive methods for underwater sensing and it is especially effective in the close range detections. In the proposed algorithm, using the vision as the primary sensor, a method for the robust path planning and smooth tracking has been implemented by generating virtual way-points based on the visually detected landmar ks. The feasibility of the algorithm has been demonstrated by the experiments using an AUV platform, K AURO, where the artificial path mar kers are used as landmar ks.

View full abstract

Fish type underwater robots have a lot of possibilities such as good mobility and high efficiency. Most of the fish type robots developed up to now have complicated mechanisms with complicated motion control. It results in that the structure and the movement of robots are different from those of fishes. The purpose of our work is to develop creaturelike flexible underwater robots by using piezoelectric fiber composite. Compared with conventional artificial muscular underwater robots, powerful underwater robots with a very simple structure can be composed by the piezoelectric fiber composite. This paper describes the development of a flexible underwater robot with pectoral fin propulsion like ray and mode shape propulsion of structural vibration, which can generate various movements.

View full abstract

This paper describes path generation and navigation system design methodology under ocean currents using multiple sensors fusion. By adopting a natural cubic spline curve generation scheme, smooth path planning can be conducted considering major ocean current direction in heuristic manner. A multiple sensors based precise navigation scheme enables underwater vehicles to be controlled with qualified navigation information such as update rate, accuracy, precision.

View full abstract

We propose a control algorithm for air cylinder actuator which has a normal PID controller with a variable offset pressure controller, so as to achieve the high speed response and less overshoot. The offset pressure is used as constant in the conventional control algorithm, which gives the stiffness to the air cylinder actuator piston. Our control algorithm therefore adopted a time-variant offset pressure in each air cylinder actuator chamber, so that the actuator properly increase or decrease the stiffness according to the target. The experimental results showed that a rise time was almost the same as that of PID controller, but tracking improved by 40% compared with a normal PID controller when the lamp of target was more moderate.

View full abstract

This paper proposes a strategy for bipedal robot walking on inclined surfaces using position and orientation based inverse kinematics algorithm. Some researchers implemented control approaches to solve bipedal walking on inclined surfaces. Generally, most of them apply control feedback at ankle joints and also introduced many more control methodologies. In this paper, inverse kinematics methodology is introduced systematically for bipedal walking on inclined floor. Positions and orientations are embedded into the kinematics calculation. In this strategy, a working bipedal robot walking pattern for flat floor must be developed first. Then, the same walking pattern can be used for the inclined floor with orientation included so that the bipedal robot is able to walk on the inclined floor successfully. This methodology will distribute the angles caused by the inclined surfaces to the appropriate robot joints. By doing this, control at ankle joints only is avoided. A 3-D dynamics simulator which is known as Robot Control Simulator and developed in our laboratory is used for simulation in order to validate our proposed method. Simulations of biped walking on 11^0 inclined floors with and without proposed method have been done.

View full abstract

Modeling and control of a rotary actuator based on shape memory alloy (SMA) wires are analyzed theoretically. The heat model and force model of the actuator is given in transfer form. A new model, well descriptive of hysteretic phenomenon between the internal variable, known as Martensite fractions(%) of SMA and stress, as well as temperature is given and written in S-function, which was integrated into Matlab/Simulink. The dynamic response under square waveform and sinusoidal signals are analyzed. In view of nonlinearity and complex parameters in SMA actuator, a classic control strategy and simple PWM is hard to implement in the SMA-based actuator system, the robust controllers based on fuzzy theory are proposed. The strategies of controller with Fuzzy gains self-tuning method and Fuzzy-PID control and normal PID are compared. Numerical simulation demonstrates that the fuzzy self-adaptive is a good choice that can be applied to practical system.

View full abstract

This paper proposes a humanoid robot's walking pattern generation method for starting swiftly by employing a feedforward controller together with a feedback controller. The proposed control scheme generates the stable walking pattern according to various walking path and shows to improve the frequency property of the carttable model which is used as a simplified humanoid robot model. However the proposed controller has a drawback such as time delay to generate the walking pattern. To overcome this problem and to reduce the time delay, we modify the desired ZMP. The efficiency of the proposed method is verified by simulations.

View full abstract

Recently, the number of people of dentists increases every year. However, the lack of skills of graduates of schools of dentistry is obvious. The problem seems to lie in the fact that the dentists have no sufficient experience of treating patients. The present study was made in order to develop a patient-robot for use in an actual clinical training. The patient has the possibility of making an unexpected motion while treating. This motion often surprises the dentist. Therefore, it is important to develop a leg of the patient-robot so that the development of the whole body motion may make the patient-robot more like a real patient.

View full abstract

Odor gas detection is important in explosives detection, environmental sensing, biometrics, foodstuff and more comfortable life. In this paper, we tried to detect odor chemicals with optical method. We used five kinds of fluorescence dyes, fluorescenin, acridine orange, pyrene, quinine sulfate, and tryptophan. As analyte, we used following four compounds, benezen, vanillin, and NPOE. It is supposed that PET or FRET mechanism will occur between fluorescence dyes and odor compounds, then fluorescence quenching will observed. Indeed, we could detect odor substances using fluorescence dyes by fluorescent quenching. Odor information could be obtained by response patterns across the fluorescence dyes. Moreover, we captured odor shape image and odor spatiotemporal images with cooled CCD camera. Shapes of target which emit odor could be roughly recognized by odor shape image. From the odor spatiotemporal images, two-dimensional odor expanse could be obtained.

View full abstract

Plume finding is the fundamental phase for the robot based odor source localization. In outdoor natural airflow environments, the fast change in both the wind speed and direction poses big challenges for a robot to find plume. This paper presents an experimental study of the spiral and zigzag algorithms for the odor plume finding in an outdoor natural wind environment. The zigzag algorithm is divided into up-flow and down-flow zigzag algorithms based on the initial motion direction. The performances of the spiral and zigzag based plume finding algorithms are compared using the indices of finding time and success rate. Experimental results show that all of these algorithms present a high successful rate, but the average plume-finding time of down-flow zigzag algorithm is the shortest among the three algorithms.

View full abstract

The issue of gas source localization (GSL) is studied in this paper. A novel method of localizing a single gas source based on a mobile sensor network (MSN) is proposed. Compared with the existing wireless sensor network (WSN) methods, the proposed MSN method only needs a few mobile sensor nodes instead of lots of stationary sensor nodes. In addition, unlike the WSN methods, the sensor nodes in the MSN are not required to enclose the gas source. In the MSN method, each sensor node sends the measured gas concentration to a central node, in which the nonlinear least square algorithm is used to estimate the location of the gas source. The MSN moves some distance towards the estimated gas-source location, and then the central node re-estimates the location of the source and the MSN moves again. Such a process is repeated till the MSN is close enough to the real source. Gas sensors are calibrated according to the relation between the variation of output voltage and the gas concentration in order to decrease the impact of the ambient temperature. Experiments in an indoor natural ventilated environment validate the proposed method.

View full abstract

In this paper, we present our initial experiments to collect a database for studying mobile robot olfaction. Mobile robots with olfactory sensing capabilities are expected to be used in various applications including gas distribution mapping and gas source localization. Owing to the turbulent nature of the airflow field and the gas distribution, these robots must be equipped with algorithms that can cope with chaotic environments. Since it is important to check the applicability of such algorithms in a diversity of environments, we propose to build a database with which the users can test the performances of their own algorithms in various environments. The database collected so far consists of two parts: a basic data set collected in a well-characterized controlled indoor environment and applied data sets collected in uncontrolled indoor and outdoor environments. A result of applying the database for testing a gas-source localization algorithm are shown as an example. We believe that such database will accelerate the research advancements on mobile robot olfaction.

View full abstract

Many rescue robots are developed for search and rescue missions in the disaster area. The rescue robots gather the information in the disaster area using sensors. Gathered information is managed by GIS (Geographic Information System) as an environment map. There are two major problems to gather the information from disaster area by using mobile robots. One is the unstable communication between operator station and disaster area, and the other is the inaccurate estimation of robots' position. In this study, we developed two new human interface systems to gather necessary information preferentially using a virtual 3D environment on demand and adjust the position errors of a robot intuitively. Furthermore simulation is carried out to validate the usability of the proposed interface by comparing with an existing interface.

View full abstract

Tracked vehicles are frequently used as search-andrescue robots for exploring areas affected by disasters. To enable good locomotion during search-and-rescue operations on rough terrain, some of those robots are equipped with active flippers. However, the manual control of such flippers requires a highly skilled operator, particularly when teleoperating with limited camera views. To eliminate this problem, we have developed a shared autonomy system that uses a manual controller for the main tracks and an autonomous controller for the active flippers. However, this system has a limitation in that the controller spreads out the flippers to maintain contact with the ground, and thus, the autonomous flipper motions hinder the turning motion of the robot. In this paper, we introduce a new autonomous controller for the flippers to support the turning motion of tracked vehicles. The new controller can be used instead of the previously developed controller, and it complements our shared autonomy system. We also validate the reliability of the new controller by performing experiments on an actual rough terrain.

View full abstract

This paper provides a trial report for displaying psuidoverlooking view for a tele-operated mobile robot. The present authors have experiences that live overlooking view of a video camera capturing images downward including the robot body and floor from higher position is extremely useful for tele-operation. To obtain this view point, a post which holds a camera in higher position is necessary. However, the authors have a fear for breaking the post when the robot is upside down in accident on the stairs or debris and wonder if height of the center of mass increases. To solve these problems, the authors propose to obtain pseud-overlooking view by means of view point transformation assuming projection surfaces. In this paper, three projection surfaces are tried and their tele-operation experiments are reported.

View full abstract

This paper presents a tracked robot composed of the proposed crawler mechanism, in which a planetary gear reducer is employed as the transmission device and provides two outputs in different forms with only one actuator. Modular design of the crawler mechanism has been conducted for enlarging its use in robot systems while considering waterproof and dustproof qualities. Modular statics for the crawler module has been investigated and a generic group of equations is proposed to describe all the configurations of the robot. Simulation results have been given numerically and the effects of some key parameters on the robot performance have been discussed in this paper.

View full abstract

We have developed a novel big danger disposal manipulator, "BIG HAND", together with crawler type mobile platform, and have reported the basic design in related field before. In this paper, we first describe the entire system implementation, and then discuss the requirement for EOD task also the control algorithms for such task in the later chapters. The "BIG HAND" is developed mainly for rescue and public security purpose, and is scheduled to be applied into public sites such as airports, stations, and so on. The manipulator has 7 Degrees of Freedom (DOF) including 1 DOF in gripper, well the crawler type mobile platform employed 2 DOFs general structure and equipped powerful combustion engine supplying long time energy to the whole system. The verified performance for the EOD task is about 4.4m maximum reach, 32kg payload in horizontal extension posture; automatic detection, scanning, grasping, and danger disposal routines.

View full abstract

This paper presents the design and analysis of a new energy-saving mobility mechanism for sensor nodes. Specifically, we propose the one-time-use leaping mechanism that utilizes the ground reaction force generated when releasing multiple precompressed springs mounted underneath individual sensor nodes. Toward enhancing coverage and connectivity in senor networks, our focus is placed on how to control the leaping distance and direction. For the purpose, both the number and pattern of release are mathematically analyzed. Through extensive simulations, we validate the effectiveness of the proposed leaping mechanism for the enhancement of network connectivity, with no additional power consumption.

View full abstract

Rapid information gathering during the initial stage of investigation is an important process in case of disasters. However this task could be very risky for human rescue crews, when the infrastructure of the building has been compromised or the environment contaminated by nuclear, biological, or chemical weapons. To be able to develop robots that can go inside the site instead of humans, several area of robotics need to be addressed and integrated inside a common robotic platform. To facilitate such integrations, we developed a highly maneuverable mobile robot which software and hardware architecture has been designed keeping in mind collaboration between research and development groups of different interests. In this paper, we described the modular interoperable and extensive hardware and software architecture of Quince, a high degree of mobility rescue robot.

View full abstract

In this paper, a new locomotion method using selfexcitation wire-swing motion is proposed. A wire with a robot hanging from it is modeled to a simple 2-link-pendulum, and self-excitation pendulum motion is implemented. The reference angle is defined by the cycloid curve trajectory since the cyclic motion is a robust swing motion, and the required torque trajectory is confined to make the motion smooth so that the torque trajectory is optimized in the multi-links robot' s constraints. The resolved momentum control can manipulate the multi-links robot as a single mass model and it can also control the variation of representative momentum. The length of the wire has to be extended smoothly and quickly to a planned throwing point. We developed a new winch with an Electro-Rheological clutch, whose response to brakes is high. The throwing timing is designed simply, taking into consideration the flight and the friction of the winch. For safety landing, the landing velocity has to be as small as possible. Thus, the direction of the landing velocity is planned to be the same from the landing point to the winch, so that the brake of the winch can efficiently reduce velocity. These methods were implemented to the dynamic simulator and the real multi-link robot, and the results show the feasibility of the high-mobility with the wire-swing motion.

View full abstract

In this paper, we have proposed a behavior selection framework using GSPNs (Generalized Stochastic Petri-nets). We exploited two navigation schemes, a sensor-based reactive scheme, viz., DWA, and map-based tracking using trajectory tracking. Our selection framework is distinct from previous behavior selection schemes because the exploited behaviors are able to ensure the completion of navigation tasks. We designed two sub-systems to monitor the states of each navigation scheme. The proposed selection framework in this study was simulated using the Player/Stage simulator. We carried out performance estimation of the two exploited navigation schemes. The results of performance estimation clearly showed that both navigation schemes are required for robust and efficient navigation in real-world environments. The average navigation time under the proposed behavior selection framework decreased by about 45% in comparison with the use of a single navigation scheme. These results clearly showed that our behavior selection framework is more efficient and robust than a single navigation scheme.

View full abstract

This research deals with outdoor SLAM based on local elevation map matching and proposes a new terrain classification method for efficient map matching. The main map used for outdoor SLAM is an elevation map which consists of 2-D grids with elevation information on each cell. 3-D environmental data acquired by tilting a 2-D laser scanner are used to build a local elevation map and classify the local terrain. The experimental results show that the proposed method can enhance performance of outdoor SLAM.

View full abstract

In this paper, we address the problem of representing dynamic environment as accurately as possible using inexpensive and error-prone sonar sensors. For a robot which should move to a goal with avoiding dynamic obstacles, the robot should have a capability to build a map which represents dynamic environment as exactly as possible. Under the necessity, we propose an efficient and accurate grid-mapping approach that is based on the sound pressure; we call the map the sound pressure map. The performance of the sound pressure map was confirmed through real experiments. Even with using sonar sensors, the sound pressure map can build the best grid map in terms of representing dynamic environment.

View full abstract

Line is the most important feature in the feature based mapping. Although it has been widely used for, but we insist that there is no proper line model that is good for SLAM applications. To validate our claim, we firstly derive five necessary conditions that a good model should satisfy. Then we evaluate the previous line models and show that none of them satisfy all the necessary conditions. Finally, we propose a new line representation that fulfills the fore-mentioned conditions.

View full abstract

In this paper, we propose a landmark-based map management method for RBPF-SLAM in non-static environment. The reliability of landmark is tracked using the negative information from the sensors. Based on the current reliability value and the phase transition of reliability, landmarks are classified into three types. The confident landmarks that represent the static parts of the environment are remained on the map along with the conditional landmarks that observed occasionally due to the environmental changes. On the other hand, spurious landmarks that represent the disappeared objects are eliminated. Result of experiment in real non-static environment verified the performance of the proposed map management method.

View full abstract

This paper proposes a novel human fall detection and prevention method for a walking aid. A three-wheeled omni-directional cane robot was developed previously for aiding the elderly walking. The relative position between the legs of user and the Center of Gravity (COG) of user play an important role in the fall detection when using the cane robot. The COG of user can be estimated from the angle of an inverted pendulum which represents human model. The angle of the inverted pendulum is computed by the support leg position, the hip position, and the force that the human pushes the cane. The fall direction and relative position between the human and the robot play an important role in the fall prevention. The proposed method is verified through experiments.

View full abstract

This paper presents a method for teleoperated mobile robots to rapidly adapt to a behavior policy. Rapid policy adaptation cannot be achieved when significant data are not differentiated from insignificant data in every process cycle. Our method solves this problem by evaluating the significance of data for learning based on the change in degree of confidence. A small change in the degree of confidence can be regarded as reflecting insignificant data for learning (that data can be discarded). Accordingly, the system can avoid having to store too frequent experience data, and the robot can adapt more rapidly to changes in the user's policy. In this paper, we confirm that by taking advantage of a significance evaluation of each proposition of each sensor level data, a robot can rapidly adapt to a user policy. We discuss the results of an experiment on a mobile robot in which the user switched policies between 'avoid' and 'approach'.

View full abstract

To perform important tasks and guarantee human safety, we can use remote controlled "Teleoperated robots". Teleoperated robots have been used in a variety of critical situations ranging from military missions, undersea operations, and rescue activities. For users of teleoperated robots, we present an interaction interface consisting of a surrounding display interface for enhancing a user's perception of the remote environment and using human gestures for intuitively controlling a robot. We present the architecture of our new interaction interface, which components are needed, and how to implement the interaction space to enhance user perception. We also propose methods for increasing user understanding and reducing the size of data transmitted from a teleoperated robot to the user. The results of this work will be used for our future research such as with a multi-teleoperated robot control and virtual human-robot interaction systems.

View full abstract

Surgical teleoperation systems are being increasingly deployed. There are, however, remaining issues such as nonlinear characteristics of the interaction between the slave robot and soft tissues, and difficulty in employing force sensors in the surgical end-effectors of the slave. These issues make it difficult to offer a general approach to designing the overall control structure. This paper addresses these issues by proposing an optimized controller which guarantees robust stability and performance. The environment, i.e., soft tissues, is characterized with the nonlinear Hunt-Crossley model. The overall teleoperation system is modeled as a linear parameter-varying system. A gain-scheduling control scheme is adapted to design a performance-optimized controller while guaranteeing robust stability. The developed gain-scheduling control scheme shows good tracking capacity and high transparency in varied experimental conditions. Error of the transmitted impedance is significantly lower compared to other conventional control schemes for frequency band less than 2 Hz which is frequently recommended for surgical teleoperation.

View full abstract

In this paper, a new type of 2-DOF spherical wrist mechanism is proposed for minimally invasive robot surgery. It is composed of the spherical 4-bar mechanism whose ground link is connected to the end link of slave robot arm by a revolute joint. Due to the hybrid serial and parallel kinematic structure, for the given desired orientation workspace, it may be designed compactly with enhanced stiffness characteristics compared to fully serial or parallel wrists. For the realization of symmetric orientation workspace, optimal synthesis is performed so that the spherical 4-bar mechanism is designed as a straight line mechanism on a sphere. Then, the 2-DOF spherical wrist is realized by connecting the ground link of the synthesized 4-bar linkage to the end link of slave robot arm using a revolute joint. For the proposed spherical wrist, screw-based Jacobian is derived and instantaneous kinematic relation is also suggested via the theory of screws.

View full abstract