The general evaluation method of various kind of safety strategies for welfare and human-care robots is first proposed in the world. The impact force and impact stress are chosen as safety evaluation values to quantify safety. The dangerous index and safety index are defined to make quantitative evaluation of the effectiveness for each safety strategy. Moreover, this method enables us to know the contribution of each safety strategy to the overall safety performance of welfare robot. As a result, design optimization of the safety robot is discussed successfully.
Learning new behaviors is a crucial problem in behavior-based robots. This research proposes a new method of reinforcement learning, called Instance-Based Classifier Generator (IBCG), for the acquisition of reactive behaviors. In IBCG, the learning system successively memorizes a newly experienced state-action pair as an action-rule. Utility of each rule is estimated by the original temporal credit assignment procedure, which is designed so that the cooperative rules leading the system to an eventual reward should self-organize. Learning capability of IBCG is experimentally examined through a task of mobile robot navigation in both simulated and real environment. The results demonstrate that the robot with IBCG acquired behaviors such as light-seeking, collision-avoidance, and wall-following.
This paper describes a method for making description of complicated scenes including different-sized objects (some objects may have textures on their surfaces) by using stereo with resolution control. In such scenes, some objects may occlude other objects and, thus, surfaces of the occluded objects can be observed as some small surfaces. In case of observing the scenes in a low resolution, the boundaries of small surfaces may not be extracted. In a high resolution, a surface boundary can be fragmented into small pieces if the surface has textures on it and if the brightnesses of some parts of the texture near the boundary are similar to the brightness of the other side of surface separated by the boundary. Thus, it is difficult to obtain scene descriptions by associating object boundaries in single resolution observation because stable boundaries may not be obtained. We propose a method of describing complicated object scenes, in which boundaries of each surface are followed while the resolution of stereo cameras in the observations is selected depending on the complexity near the following boundary in order to obtain stable ones. The description of the object surfaces are constructed from only those stable boundaries. We proved that our method is applicable by some experiments of real scenes.
A micro inspection robot for 1-incn pipes was developed. The robot can undertake visual inspections inside piping and collect small objects. It is 110 [mm] long, has an external diameter of 23 [mm], and weighs 16 [g] . The robot is equipped with a high-quality micro CCD camera and a two-digit hand for manipulating small objects in pipes. It is propelled by a micro electromagnetic motor. The wheels, which are driven by planetary reduction gears and worm gears, press against the pipe wall and enable the robot to travel inside even in vertical pipes and curved pipes. It travels at a speed of about 6 [mm/s] and the pulling force exerted is approximately 1 [N] . In this report the robot system, the various micro technologies used in the development of the robot, including the micro actuator, micro reduction gears, micro robotics hands, and micro CCD camera, and the specifications of the robot are presented.
This paper deals with a new mobile micro-robot using centrifugal force. Many mobile micro-robots using PZT elements have been developed, though their PZT elements generally require some high voltage amplifiers with electric cable lines. Therefore, it is a little difficult for the mobile micro-robot to move about in long and thin pipes. My proposed micro-robot consists of many brush fibers, two coreless motors and a button electric cell. These coreless motors with eccentric weights generate centrifugal forces which transmit to the brush fibers with elastictity, and consequently occur driving forces of the micro-robot. First, I derive a basic mathematical model of the mobile micro-robot using centrifugal forces. Next, some simple prototypes of this robot are developed, and the characteristics of high velocity is examined by experiment. Finally, I verify the validity of the proposed mathematical model by comparing simulation results with experimental results .
A well-known difficulty of the non-linear optimal feedback control is that it results in non-linear two-point boundary value problems (TPBVP), which in general, can not be solved analytically. Therefore, exploration of solution algorithms is a key problem to establish a practical technique of non-linear optimal feedback control. Stabilized continuation method is shown to be suitable for real time optimization in non-linear receding horizon state feed-back control which can be applied to various problems in the field of engineering. Once a performance index and a mathematical model are given, optimization yields a feedback law that achieves the best performance in terms of the given performance index. This paper addresses non-linear receding horizon state feedback control law and its application to non-linear mechanical link systems. The contributions of this paper are: (i) outline of receding horizon control law and continuation method, (ii) comparison between receding horizon control and gradient method, and (iii) application to non-linear mechanical link systems.
This paper presents a robot symbolic analysis system by Maple called ROSAM II for the symbolic modeling of robots. The ROSAM II permit to generate the symbolic models of forward kinematics, inverse kinematics, forward dynamics, inverse dynamics, base parameters and so on for serial robots, tree-structured robots and closed-loop robots. It has been developed under the computer algebra software Maple V and its GUI for settling robot parameters has been implemented using the C++ language under Windows95/NT. In this paper we represent a design concept of the ROSAM II and an overview of algorithms generating symbolic models. An example for the robot symbolic analysis is also shown.
Considering walking robot's legs as manipulators, the multi-legged walking robot can be treated as grasping manipulators. It means to separate the multi-legged walking robot into body part and leg parts. The body part is rigid and it is forced to move by legs. The leg part is a set of manipulators, which give some forces to the body part. In this point of view, we can treat multi-legged walking robot's problem as dextrous manipulation, which realizes desired body motion. In this paper, we propose a method of force-torque distribution of walking robot's legs.
With the increase in the number of senior citizens, there has been a growing demand for human-friendly wheelchairs as mobility aids. This paper proposes the concept of an intelligent wheelchair to meet this need. It can understand human intentions by observing the user's nonverbal behaviors and can move as the user wishes with minimal human operations. It also describes our experimental robotic wheelchair system. Human intentions appear most on his/her face. Thus, the experimental system observes the human face, computing the direction of the face. As the first step toward the intelligent wheelchair, we have performed experiments on controlling the system's motion according to the direction of the face. Experimental results prove our approach promising.
Generation of an environmental map is one of important tasks for robot navigation. We have proposed a method to generate a static environmental map and estimate the egomotion of a robot by using an omnidirectional image sensor, based on the assumption of unknown translational motions of the robot. However, since both robot and objects move in the environment, the map generation and the robot egomotion estimation by using a single camera are difficult because of ambiguity of correspondence caused by occlusion. In this paper, we improve our previous method for treating applicable to a dynamic environment. The method can detect a moving object and find occlusion and mismatching by evaluating estimation error of each objects location.
In this paper, we propose a new method to drive SMA (shape memory alloy) -pipe active forceps for minimally invasive surgery. Two actuation methods of SMA pipe have previously been introduced, namely, hot/cold water circulation, and electric heating. The former suffers from rather slow response. Although the latter shows better response, it is unsuitable for surgery of current-sensitive areas. In this paper, optical heating and gas cooling are newly proposed for heat transfer of the shape memory alloy pipe. The use of optical heating and gas cooling allows to actuate forceps in dry conditions, which simplifies sterilization and eases manufacturing process. It also opens the way to applying this forceps to surgery of current-sensitive areas, such as cardiac and cerebral surgery. The prototype active forceps is fablicated using KTP/YAG laser as light source, thin optical fiber (core diameter: 0.3 [mm] ) as light guide, and CO2 freezing gas as coolant. Mathematical analysis provides fundamental heating/cooling performance. The prototype experimentally shows approximately 3 [sec] of heating/cooling speed.
This paper provides a grasp planning for a multifingered hand-arm robot in the presence of obstacles. When the hand grasps an object, the arm and the hand must not collide with obstacles. It is difficult to generate a grasp configuration for a multifingered hand-arm robot even though geometric information about an object and obstacles are all known, since a multifingered hand-arm robot has many degrees of freedom of motion. We propose two methods for generating the grasp configuration and select the one method according to the environment around the object. One method is called the hand configuration method and gives priority to the hand configuration. Another method is called the arm configuration method and gives priority to the arm configuration. We separate the grasp planning into several subproblems. The feature of our grasp planning is to use heuristics in order to generate the grasp configuration rapidly. The effectiveness of our grasp planning is clarified by simulation results with 3D graphics.