As the first stage of biped walking adapting to an unknown uneven surface using an anthropomorphic biped walking robot, this paper introduces a special foot mechanism with shock absorbing material that stabilizes biped walking and acquires position information on the landing surface. The new foot has three functions: (1) a function to obtain information on the position relative to a landing surface; (2) a function to absorb the shock of the foot's landing; (3) a function to stabilize changes in the support leg. Two units of the foot mechanism were produced, a biped walking robot WL-12 RVI that had the foot mechanism installed inside it was developed, and a walking experiment with WL-12 RVI was performed. As a result, decreased vibration around the pitch axis, decreased torque demands on ankle actuators on the pitch axis, increased dynamic biped walking success probability, and acquired landing surface information was achieved.
Artificial neural network (NN) can be applied to complex dynamical control system. The multilayer neural network with sigmoid function is often used in this field. But this type NN cannot learn the patterns additionally. It must learn both unlearned patterns and patterns given before. The neural network based on the distance between patterns (NDP) can memorize patterns additionally and recognize unlearned patterns. Adaptive force control using NDP is proposed in this paper. Hierarchical neuromorphic controller is used, in which the higher level NDP detects changes in environments and activates a corresponding lower level controller. Multilayer neural networks are used at the lower level for the control of unknown plant. Hierarchical structure can enlarge the range of the adaptation, and learn additionally.
Kinematics of grasping and manipulation of multifingered hand (GMMF) where multi-fingersurfaces contact with an object are discussed. There are many researches about GMMF problem, but most of them discuss on the case of fingertip contacts. In this paper the following problems have been investigated. (1) GMFF problems are classified to direct kinematics, inverse kinematics for grasping and manipulation, which are peculiar to the fingersurface contact. In each case the problems are formulated. (2) For the manipulation of an object, kinematic relationship between object and finger-joint motions are analyzed in the case when the contact motion is pure rolling (PR), twist rolling (TR), and slide rolling (SR). In each case the number of finger and degree-of-freedom of fingers necessary to grasp and manipulate the object are shown. (3) Manipulation kinematics is characterized by the equations for incremental variations of object motion and finger motion, mediated by contact point variations, unsimilar to the manipulation of manipulator arm kinematics where the equations are expressed by absolute displacement. Some examples of the computer simulation of grasping and manipulation using the above-mentioned formulation for various conditions are shown.
In this paper, first, the linear DC motor surrounded by four faces for the direct-drive hopping robot is developed. Second, the experiments of the continuous hopping by the simple hopping robot consisting of the linear DC motor and the holding springs have been carried out, and the usefulness of the linear DC motor surrounded by four faces for the hopping robot is confirmed. The rotor and the stator of the linear DC motor are designed as the body part and the leg part of the hopping robot. The body part consists of samarium-cobalt rare earth permanent magnets and yoke. The leg part is constructed with coil and coil frame. Then, the continuous hopping motion control is performed by the cycle operation of the sequential control system. This sequential control system consists of the potential energy store control and the body acceleration control. From the experimental results, it is confirmed that the continuous hopping motion control can be performed successfully.
This paper deals with a kinematic analysis of locomotive cooperation necessary for two mobile robots to carry a long bar along a general wavy road. The robots are assumed to be driven by crawler mechanism and to have a single link arm which moves only prismatically. The robots have a task to carry a long bar with keeping its height and length constant and its posture horizontal. The analysis of kinematic conditions done for a sinusoidal wavy road is expanded for a one-dimensional general wavy road, and we find that there exist 14 patterns in the relationship between two solution curves of mobile robot and reveal that two kinds of cooperation mode are needed: one is to change the direction of locomotion and the other is to alternate the robot role from “master” to “servant” and vice versa. We further develop the algorithm for the locomotive cooperation.
Contacts with insufficient degrees of freedom are defined as defective contacts. Although defective contacts are found in the power-grasp and the enveloping grasp of a multifingered robot hand, the mechanical characteristics are not yet studied in depth. In this paper, an approach to the mechanical analysis of grasps with defective contacts is presented. The relations among joint torque space, contact force space, friction cone, admissible external force space are made clear. By using the convex analysis, we represent these sets in the external or internal representation of polyhedral convex set. Finally, the validity of our approach is illustrated by a numerical example.
This paper presents a new manipulation method, graspless manipulation. Graspless manipulation refers to operations in which a robot attempts to change the position and posture of an object without grasping it rigidly. In comparing with an ordinary method, pick-and-place, graspless manipulation is complex operation, but it requires less forces and less number of finger tips for manipulation. Typical examples of graspless manipulation which have been studied are pushing and tumbling. The operations are classified by the contact states with environment, and then a new method of graspless manipulation, pivoting, is proposed. This is a same operation as a person moving furniture, which rotates an object around an appropriate vertex on the floor. In general, the operation of pivoting is complex 3D motion, but this paper shows that it can be divided to two 2D problems on some conditions. Then, this paper describes the mathematical analysis of pivoting, and then presents an experimental result.
A method has been proposed to judge if a set of base-parameter values for the dynamic model of a manipulator determines the inertial matrix of the dynamic model to be positive definite for each configuration of the manipulator or not, when continuous change of each joint variable of the manipulator is approximately considered as a finite set of discrete points. If a set of base-parameter values is judged not to do, it is physically impossible. The method can be executed on computers. The method is useful to examine a set of estimated base-parameter values. A method also has been proposed to modify the estimated base-parameter values such that the set of them determines the inertial matrix to be always positive definite if it is judged not to do.
This paper proposes an ideal control method for the adjustment of joint impedance of manipulators, by using mechanical elements such as Leaf Spring and Pseudo-Damper by a Brake. The proposed method can realize an ideal characteristics of impedance compared to conventional active force control methods which performances are limited by the responses of the servo systems, the non-linear characteristics of the force sensors and the dynamics of the manipulator. Compliant following motions in cases of sudden changes of external forces is expected. Concretely a compliance adjustment method using the spring mechanism, the structure and control of the Pseudo-Damper system, and simple control algorithms for the coordinated system are mentioned. The effect of our mechanism and methods are shown by an experiment of adjusting the joint impedance to appropriate levels including very compliant conditions and positioning, by using the 1-D. O. F. Finger Model which the new joint mechanism is applied to robot fingers.
This paper describes a new mechanism, communication system, and vision system of an in-pipe inspection robot. To date, inspection robots have had such limitations as mobility of the robot to turn in a T-shaped pipe or move in a plug valve. The new mechanism based on our dual magnetic wheels overcomes these limitations without difficulty in control. This dual mechanism, resembling a crawler, enables the robot to climb over sharp obstacles like sleeve and dresser joints. Another drawback of earlier robots is that the friction between the pipe and the cables for communication and power supply makes it difficult to move a long distance. A fiber-optic communication system can reduce friction. The spooler of the fiber-optic communication cables and batteries are mounted on the robot and the cables are rolled and unrolled when the robot is going forward and backward, respectively. The new vision system has been significantly miniaturized, enabling it to clearly view and inspect the pipe welds underneath the robot while gazing ahead for navigation. The reflection of slit light made by a light-emitting diode and a cylindrical lens is used to diagnose the pipe weld. An experimental inspection robot has been successfully made to confirm the efficiency of the new mechanism and the vision system.
This paper discusses a new method for teaching a deburring robot based on demonstration of human skillful motion. The robot is programmed to adjust the tool feedrate in accordance with the varying burr characteristics, such as burr size and material properties. This dynamic change of tool feedrate is motivated by the effective human skill in performing a deburring task. The relationship between the tool feedrate and burr characteristics is obtained from human demonstration data and stored in a computer as an associative memory. This associative memory enables the robot to select the tool feedrate that well matches the burr characteristics. Therefore, the robot motion is always effective in removing burrs and generating smooth finish of workpiece surface without severe tool wear. In order to identify burr characteristics, a laser displacement sensor has been used for direct burr height measurement, and a deburring process model has been applied for material property differentiation. The learned associative memory is stored and represented by a neural network, which can be easily incorporated into robot programming. Experimental results show that a robot can perform a deburring task in a manner similar to its human teacher.
This paper provides a wall moving mini-robot using adhesive method. In this paper, we use the ‘mini-robot’ as a kind of micro-robot larger than 10[mm] cube but smaller than 100[mm] cube, A mini-robot should have the ability of moving along the obstacles because of its size. Wall moving is the first step to pass along the obstacles. Considering the scale factor, we propose the adhesive method for wall moving of a mini-robot. While moving on a wall, the adhesion force of the adhesive that has been painted to the wall is used to support the mini-robot.