In biped locomotions there are several kinds of walking motion patterns. By selecting one of these motion patterns appropriately according to the change of circumstances, a robot may be afforded abilities of adaptivity in walking. However, if there is no large change of circumstances around the robot, the biped locomotion can be realized by repeating the same motion. Therefore, first of all it is most necessary for the biped locomotion robot to generate one walking motion pattern.
In general, even if one walking motion pattern is given, it can not be realized by only simple servo-mechanisms, since dynamic interferences of links of a robot, frictions, and changes of a mechanical constraint condition with walking motion strongly affect the motion of the robot. To overcome these difficulties we apply a kind of learning control to the biped locomotion robot. According to this control scheme, a desired input pattern which can realize a walking motion is constructed as the robot repeats walking motion. After several trials, the robot can obtain the walking motion pattern.
In this paper, some experimental results obtained by this learning control are shown. Moreover, we suggest that an input pattern which generates another walking motion can be directly constructed from the input pattern obtained previously by learning.

View full abstract

We formerly developed the motion oriented robot language (IML) with the characteristic of functional programming language such as LISP and APL. But, IML didn't process feedback information from various sensors for improving adaptability to working environments, so the task which was performed adapting to the change of the environment couldn't be programmed in IML.
In this paper, as the sensory function of a robot we think about force and on-off sensor and describe the IML which is extended to process these sensory information, and it is shown that task programming becomes easy by introducing the playback of the teaching data with the force control mode. We could confirmed the effectiveness of IML by making really a robot perform some tasks programmed in IML.

View full abstract

The method of controlling an industrial robot is, at the present, a semiclosed loop for controlling the movement of a robot arm by detecting the rotational angles and velocities of a driving motor. Robot systems of such structure are comprehensively analyzed by simulation, considering not only arm dynamic system, but also control system, driving system and transmission system. As a result, the following facts are made clear in relation to the vibration of robot arm and its reduction.
(1) Characteristics of robot arm vibration
When the rigidities of transmission system is thoroughly weaker than those of control system, the lowest natural vibration of arm dynamic system is predominant. This is also a primary natural vibration of the mathematical model in which the links are rigid and springs are provided at joint. As joint springs, stiffnesses are as equal as the rigidities of transmission system. When the rigidities of control system is thoroughly weaker than those of transmission system, the vibration is caused by the rigidities of control system.
(2) Reduction o f robot arm vibration.
In the first place, it is necessary to find whether the vibration is derived from the rigidity of control system or that of transmission system. If it is due to the rigidity of transmission system, the effective measure is to raise it or lower the rigidity and the damping of control system. If it is due to the rigidity of control system, , it is effective to raise the damping of control system.

View full abstract

This paper describes the problem of similarity on walking machines in the energetic efficiency.
First of all, five non-dimensional parameters which control the energetic efficiency of walking machines are introduced for nine physical parameters by applying the techniques of dimensional analysis.
In order to check those influences on the energetic efficiency, the basic considerations on actuator arrangement and computer simulations are performed for n-legged walking model. The influence on the ground reaction forces caused by the leg inertia forces is especially considered in this simulation.
Simulation results and basic explanations on similarity are concluded in the following way: (1) Energetic efficiency (specific resistance) is largely influenced by the body height ratio, even though the same stride ratio is assumed. (2) Energetic efficiency (specific resistance) is influenced linearly by the leg/body mass ratio. (3) Ground reaction forces due to the leg inertia force are not always negligible. (4) Six-legged machines (minimum duty factor=0.5) consume the fewer energy than four-legged machines (minimum duty factor=0.75) in high walking speed, while the result becomes opposite in extremely low speed. (5) Even though two walking machines have the similar configurations and same leg/body mass ratio, energetic efficiencies for both machines are not equal gererally, and same energetic efficiencies are only realized in the limited walking speed.

View full abstract

This paper describes a very high level robot language processor for the blocks world which consists of task-oriented level and object-oriented level, and shows a “build arch” example.
At first, the task-oriented level processor translates its input like“build arch”into the concrete goal state descriptions as its output by referring to knowledge about the goal task in the concept model. Then, the object-oriented level processor translates the goal state descriptions into the source code of the motion level processor AL/L by making task plans automatically and computing the position and the orientation of each object.
If a concept of the goal task is predefined in the concept model, these processors can expand the goal task into the source code of AL/L. Therefore human programmers can write their programs without considering the details of the environment and the robot.

View full abstract

This short paper discusses a method to simplify the trajectry teaching of robots along object profiles by using the ultrasonic proximity sensor. To control a robot manually, this system employs a force operated joystick which can provide the velocity command in the direction of the robot hand, and the proximity information is f edback to the joystick as a reaction force.
Using this force feedback, operators can guide the robot hand along the object surface as if the endeff ector was compliantly pressed to it. Then the labour involved in the fine positioning will be extreamly reduced and the efficient teaching will be performed.

View full abstract