[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

This paper proposes a method that acquires cooperative behaviors based on the estimation of the state vectors. In order to acquire the cooperative behaviors in multi robots environments, each learning robot estimates the local predictive model between the learner and the other objects separately. Based on the local predictive models, robots learn the desired behaviors using reinforcement learning. The proposed method is applied to a soccer playing sit-uation, where a rolling ball and other moving robots are well modeled and the learner's behaviors are successfully acquired by the method. Computer simulations and real experiments are shown and a discussion is given.

View full abstract

Recently, investigation of underwater structure and fresh water environment such as lake and dam which are closely related with human life have got much attention. Even though Autonomous Underwater Vehicles (AUVs) are anticipated to be very useful tools for mentioned missions, problems of conventional navigation method prevent AUVs from accomplishing them. In this paper, a new underwater navigation method for AUVs is proposed based on Artificial Underwater Landmarks (AULs) which are recognized by a vision system implemented in the robot. When an AUV cruises by deadreckoning, position error accumulated gradually depending on characteristics and direction of navigation devices reliability. When the error is within an acceptable range for finding a specific AUL along the track line, the AUV can correct the position and direction by refering to those of the AUL. Then, the AUV continues to cruise by deadreckoning again. Since the underwater vision system is requested to have high reliability, a new un-derwater image processing strategy is introduced based on analysis of underwater vision environment. Configuration of the AUL, shape and color of targets on it are designed to optimize image processing and recognition. Experiments were carried out to demonstrate the efficiency of the proposed navigation method by the intelligent AUV “Tri-Dog1”.

View full abstract

In this article, we will develop pneumatic actuators composed of multiple elastic tubes. Elastic tubes with mechanical constraints have a capability of performing a single various motion while multiple motions cannot be performed. Actuators composed of single-motion tubes can realize multiple motions by controlling air pressure imposed on individual tubes. First, we will investigate the motion of elastic tubes with mechanical constraints. Second, we will present the concept of pneumatic group actuators. Next, we will develop two their prototypes; single-stage pneumatic group actuator and double-stage pneumatic group actuator. Then, we will formulate the kinematic properties of the prototypes. Finally, we will experimentally demonstrate the behavior of the two prototypes.

View full abstract

In this paper, we propose a successive gait-transition method for a quadruped robot to realize all-directional static walking. The gait-transition is successively performed among the crawl gaits and the rotation gaits while the foots hold in common positions before and after gait-transition. The transition time of the gaits is reduced through carefully designing the foot position of the crawl gait and the rotation gait while limiting the foot in a rectangular reachable motion ranges. Experimental test was executed to show the validity and the limitation of the proposed gait-transition method.

View full abstract

This paper reports an evaluation about autonomous behaviors of an interaction-oriented robot, which will work in our daily life as our partner. To develop and improve such an interaction-oriented robot, it is necessary to find out the evaluation method of the human-robot interaction. We tried to evaluate the robot named“Robovie”, which has a human-like upper torso, a sufficient physical expressing ability, and abundant sensors for communicating with humans. Robovie autonomously exhibits playing behaviors such as a handshake, hug, and short conversation, based on visual, auditory, and tactile information. For the evaluation, we installed three behavior patterns“passive”, “active”, and“complex”into Robovie. As the result, “passive”pattern brought the best impression. We also analyze the dynamic aspects of the interactions with a concept of“entrain level”, then we suggest interaction-chain model for human-robot communication.

View full abstract

Recently, techniques of arranging micrometer-sized objects with high repeatability under a scanning electron microscope (SEM) are required to construct micro-devices. Since micro-objects tend to adhere to other objects mainly by electrostatic force, manipulation techniques by adhesive force are often taken in basic researches in which micro-objects are studied. These techniques, however, still do not achieve the desired repeatability under the real condition because many of these are used just for the empirical effectiveness and some even include the trial-and-error ways. Thus, in this paper, we analyzed micro-object operation theoretically by considering adhesion and rolling-resistance factor in the kinematic system consisting of a sphere, a needle-shaped tool, and a substrate. In this analysis, we clarified that it is possible to control the fracture of the contact interfaces of the system by determining the tool-loading angle reasonably. Based on the analysis, we also proposed the reasonable way of pick and place operation of a micro-sphere under an SEM.

View full abstract

This paper presents an interface for choreographing human and animal characters without reference motion. Its basic function is to enable animators to generate a natural motion clip by dragging a link to an arbitrary position with any number of links pinned in the global frame as well as other constraints such as desired joint angles and joint motion ranges. Each constraint can be switched on or off, strengthened or weakened for each joint at users' will. The interface is based on a new online inverse kinematics technique that allows more flexible attachment of pins and various types of constraints. Although the basic method can create natural motions from scratch, editing or retargetting captured motion requires only a small modification to the original method. We also demonstrate the powerfulness and usability of the proposed method by a number of example motion clips. The same methodology can be applied to flexible and intuitive manipulation of humanoid robots with large degrees of freedom in real time.

View full abstract

We have discussed a motion control of the legged robot Emu. In the research, we implicitly assumed that Emu stands uprightt on a level surface. However it is difficult to prepare the surface in practice because of an inclined plane. If we experiment in controlling Emu on the inclined plane under the assumption, Emu may fall down. Therefore we focus on an attitude control on a ramp of unknown slope. According to the stability analysis, we found that our proposed control system had a property of parametric stability. Thus we analyze a parametric stability of a precise model of the experimental apparatus of the legged robot Emu and verify the parametric stability through some simulations and experiments.

View full abstract