Humanoid robot

As for humanoid robot, how to simplify complicated walking dynamics to help construct stable walking controller has been studied intensively.  From a view point of making a stable controller for the bipedal walking based on knowledge of control theory, it looks like to be not easy because of dynamics with high nonlinearity and coupled interactions between state variables with high dimensions.  ZMP-based static walking is one of potential approaches, which has been proved to be a realistic control strategy to realize stable walking of actual bipedal humanoid robot, since it can guarantee that the robots can keep standing by retaining the zero moment point within the convex hull of supporting area.  Avoiding complications in dealing directly with true dynamics without approximation, inverted pendulum has been used frequently for making a stable controller, simplifying the calculations to determine input torque. Further linear approximation of the humanoid robot into simple inverted pendulum enables researchers to aim at realizing stable gait through well-known control strategy. These previous discussions are all based on simplified bipedal model, which avoided to discuss the effect of foot and slipping motion existing usually in real world.  What the authors think is more important is that the dimension of the dynamical equation will change depending on the walking gait's varieties. When the foot's underside surface contacts flatly with the ground without slipping, the foot does not move of course against the ground, having the foot excluded from the dynamical equation.  Our research has begun from such view point of aiming at describing gait dynamics as correctly as possible, including slipping of the robot on the ground. But our modeling approach adopted Newton-Euler Method that calculates all links dynamics with full dynamical freedom of six (3 for position, 3 for rotation). Furthermore the NE Method helps to include easily the slipping freedom of foot into the description of the dynamics, as has been written in this paper, and also being useful for solving forward dynamics calculations With the increase of degrees of redundant manipulators, weights are also increasing. So, in this situation, it is not suitable for only basing on kinematics when high-accuracy control is required. Therefore, we propose a new concept Dynamic Reconfiguration Manipulability (DRM) which is a measure of how much a dynamical system can potentially produce a motion in a workspace with normalized input torque, by combining the Dynamic Manipulability with Reconfiguration Manipulability. This new measure represents how much the dynamical system of robots possesses shape-reconfiguration acceleration ability in workspace by unit torque input for all joints during executing primary tasks.

  • 腕振り歩行
  • 不整地歩行
  • DRMを用いた運動解析
  • 腕振りを用いたジャンプシミュレーション

Visual servoing

Nowadays, in the field of robot vision, a control method called a visual servoing attracts attention. Visual servoing, a method of controlling a robot using visual information in the feedback loop, is expected to be able to allow the robot adapt to the changing or unknown environments. In this research, two cameras were mounted on the end-effector of the robot arm, in order to use visual information. A hand-eye configuration is inevitably affected by hand’s dynamical oscillations, then it is hard to keep target’s projected position at the center of camera’s view. So Hand/Eye-Vergence system was proposed on a view point that the visual serving controller of the hand and eye-vergence should be separated.

  • 実験映像
  • 6自由度ビジュアルサーボイングの実機での実験映像

Grinding robot

Recently industrial robots are used for many purposes, especially as machining facilities.For example, there are welding, assembling and grinding operations. Many of grinding robots are controlled by using feedback with force sensors. However, the force sensorused for force detection by means of strain gage will be damaged if a shock and big force should be applied. So, this research aims to achieve a new method for grinding a target object into desired shape with force-sensorless feed-forward control. However, a problem that vibration occurs during the grinding work has emerged, making the accuracy of the grinding become worse. Therefore, this report considers a method that changes the gain of the position control for suppressing the vibration. We experimented with an experimental device and compared current data with earlier data to confirm the proposed stabilizing method is effective. Further, the effects caused by the change of contact angle for theobject surface to grind are examined , and the results showed our method can reducevibrations during grinding.

  • 実機を用いた実験映像

Catching fish robot

We report relative comparison on leaning speed of fish in an antagonistic relation of prey and predator-the prey is a fish and the predator is a robot seeking to catch the fish by a net attached robot’s hand through visual servoing. It was confirmed that the fish have found escaping strategy by itself, e.g., staying at corners of a pool where the net is inhibited from closely approaching to the corners to avoid the net clashing to the pool wall. The effectiveness of the conceived escaping strategies by fish have been measured as learning speed that describe decreasing tendency of how many fish could be caught in constant time when the fish caught being released immediately to the same pool. To overcome such fish’s ability to conceive new strategies for escape, in this paper, chaos and random have been added to the net motion, which are experimentally examined to judge whether chaos and random can decrease the fish’s learning speed.


Multi joint manipulator

The multi joint manipulator can work dexterously by using the complicated-complex structure and the object, since the robot with the shape of the complexity increasing redundant degree as adding to the number of links. However,the more the number of links increases, the more manipulator's weight and dynamic interference between links increases. Therefore, the control accuracy of end-effecter worsens. And then in order to control, the bigger power and energy will be needed.Here, I think a manipulator just can save energy and more precisely like a human-being who can do some highly accurate task by a little power with elbow-contacting or wrist. So, we prove the effectiveness of with elbow-contacting by the simulation.

Underwater Vehicle

Nowadays, the role of AUV is playing as one of the essential tools in human society in terms of countless domain in applications such as inspection of underwater structures (dams, bridges), and underwater cable tracking. Even though many researches have been conducted and published worldwide, rapidly increasing demands to be addressed in many spheres for researchers have been expanding. In spite of developing technology related to power storage devices, there has been still limitation for underwater vehicle especially for operations that take longer duration than power capacity of underwater vehicle. Recharging unit with underwater docking function can enable the AUV to operate independently of a surface vessel for extended periods. Therefore, the role of docking operation came in picture not only for battery recharging application but also other novel applications such as sleeping under mother ship, or new mission uploading. Moreover, docking capacity can be extended to provide navigation for other underwater vehicles on the way of their mission too. However, there are a number of challenging issues in achieving these applications which request high accuracy and robustness against disturbances that are provided by the underwater environment. To achieve those tasks, we have developed visual-based docking system for underwater vehicle using stereo vision. Recently, due to the progress in computer vision, vision-based system has been highlighted as promising navigation system. Like the land and space systems, a number of researches on underwater vehicle using visual servoing have been conducted in worldwide recently. Each of them is with different merits and limitations depending on intended applications. Most of researches are based on the monocular vision. In contract, we have developed 3D pose tracking system for docking operation using stereo vision providing high homing accuracy. To the best knowledge of authors, our proposed system is the world first initiated research using two cameras as stereo vision in underwater vehicle environment. We have developed efficient and robust real-time autonomous docking system by means of visual servoing using stereo vision, named as Three-Dimensional Move on Sensing (3D-MoS). The system recognizes a relative pose between ROV and target object utilizing 3D model-based recognition and Multi-step GA. We conducted regulating performance experiments in which the defined relative pose between the ROV and target is maintained using proposed system. Then the robustness of visual servoing is evaluated against different disturbances while the ROV is controlled by visual servoing. Finally, we conducted docking experiment in the sea of Wakayama prefecture to evaluate how much our 3D-MoS system would be robust against natural sea environment. According to the experimental result, it was confirmed that docking performance in sea using proposed system was achieved successfully with centimeter level accuracy in recognition and visual servoing.


Underactuated Flight Object

The helicopter is applied in large field because of flight ability such as vertical ascent, vertical descent and hovering. Especially manned helicopter is used for rescue, emergency activity and firefighting at the time of disaster, and unmanned helicopter is precious sources of information in the danger spot where people cannot approach. But construction of helicopter is complex and sensitive to the influence of the wind. There are mainly two models in a helicopter. One is a single-rotor helicopter having main rotor and tail rotor. Tail rotor is to generate anti torque. Another one is a twin-rotor helicopter having two main rotors. Our laboratory has an experimental device of three degree-of-freedom underactuated flight object like twin-rotor helicopter. Twin-rotor helicopter has an advantage in the safety, because this model has a characteristic that operation stability against roll is high. This device controls the angles of vertical and rotation direction by thrust gained by two rotors. Controlling an underactuated flight object has attracted a lot of attention, due to the fact that flight object is an underactuate nonlinear system. It is considered that it is possible to contribute for reducing weight, lowering the cost, and the energy saving if the system can be controlled with the number of control inputs less than the number of the system outputs. The control strategy for the experimental device is examined by two degree-of-freedom PID control. This method aims to adjust target-tracking and disturbance-reduction ability independently by restraining the excessive control inputs.


Temperature Control of the Aluminum Plate

The aims of this research for three dimensional temperature control by expanding the model of one dimensional aluminum plate into two dimensional. This research uses two degree-of freedom generalized predictive control because the concept of prediction is acceptable and reasonable for industry. GPC technique has been first proposed by Clarke and others in 1987. The control method has features that the objective function includes prediction and control horizons, and control signals are computed by receding their horizons. With these features, the control strategy has been accepted by many of practical engineers and applied widely in industry. Although our method can achieve to design the output response for the aluminum plate model independently of modeling error or disturbance, the previous model was just one dimensional model. Further, the temperature control for the purpose of temperature uniformity by Nano and others in 2002 has studies gradient of temperature control method by using the PID control. Two DOF GPC using temperature control is unique in other ones. We expected that this method can be utilized in industrial fields. So this research aims for three dimensional temperature control by expanding the model of one dimensional aluminum plate into two dimensional one in order to develop thermotherapy machine and to produce products which are made from thermoplastic materials.

  • アルミ板の温度制御

Robotic IVR (Interventional Radiology)

In recent years, medical technology is progressing in human society. As one of them, Interventional Radiology (IVR) is paid much attention. IVR is a surgical method which cures illness by inserting a needle or a catheter into a body while seeking its position with CT image or an ultrasonic image. As kinds of the blood circulatory system IVR, there are an intra-arterial injection treatment and a vascular embolization. The intra-arterial injection treatment is the medical procedure which injects an anticancer drug from the inside of a blood vessel to a tumor. The vascular embolization is the medical procedure which closes a blood vessel which caused bleeding or a tumor vessel. As kinds of the non-blood circulatory system IVR, there are radiofrequency ablation (RFA) and a drainage. RFA is the medical procedure which cauterizes a tumor by stabbing with a needle and generating radiofrequency waves. It is mainly used the medical treatment of lung cancer or liver cancer. The drainage is a medical procedure which excretes the pus and the blood which accumulated into the body. IVR has the advantage that invasiveness is low and the medical treatment in local anesthesia is possible as compared with the fact that a laparotomy surgery requires general one. Therefore the number of IVR surgery has been increasing in recent years. In this research, IVR with RFA to lung cancer is focused on. Since lungs have membrane layers including air, a cancer in lung is not reflected in an X-ray photograph and an ultrasonic image. Therefore, in IVR for lung cancer, CT equipment is widely used. In the current IVR for lung cancer, radiation exposure to medical doctors has been concerned, since doctors conduct the treatment with holding a needle beside a CT equipment. Thus, radioactive exposure to medical doctors during puncture should be reduced through supporting a needle by a robot. However, any surgical robots for RFA has not been developed yet. In addition, not only reduction of radioactive exposure but also reduction of the number of puncture by position control using CT image information is expectable. In this research, a robot which supports a needle of RFA to lung cancer is proposed.


A lot of researches on inverted pendulum cart are conducted in recent years. And it is attracted as a personal vehicle which realizes energy saving for many practical applications, such as the Segway and the P.U.M.A. The personal mobility has the advantage of energy efficiency for transportation because the cart is small and light. However, since most conventional personal vehicles require a certain level of physical ability from the driver, it is not suitable for elderly and disabled people to drive. Therefore, an inverted pendulum cart with a sliding mechanism for posture control is developed as a personal mobility available for anyone. And we aim for realization of acceleration and deceleration while keeping the angle of the cart perpendicular.


The demand that a robot manipulates deformable objects, such as cables and wires surround peoples grows. It is important for a robot to recognize these forms by using image information. Form estimation algorithm of deformable object based on image information is proposed in previous research.