In this paper, the authors propose a learning control method of the compensative trunk motion for a biped walking robot which has a trunk, based on the ZMP (Zero Moment Point) stability criterion, for the cases of the ZMP within and outside the stable region. And the authors develop a biped walking robot with a ZMP measurement system and a support device. By computer simulation and learning control experiments, the authors confirmed the convergency of the learning method and the change of the convergence rate with the change of the weight coefficient. By the learning control experiments for the case of ZMP outside the stable region, the authors showed that even though the walking state of the robot itself changes, by supporting it with a human and its learning with the ZMP and the support force, stable walking even without the support of a human is realized at last.
BUNRAKU is a Japanese traditional puppet show. In this paper we introduce a virtual reality techniques to BUNRAKU so that people can enjoy it interactively. The proposed system allows us to play BUNRAKU together with virtual reality in real time through a telephone line by preparing the same stages at distant locations. If a human puppeteer controls the puppet at the one place, the sensors detect the puppet's movement, and these data are sent to the other place, where the puppet's action is reproduced in real time by the robot according to the received data. A shamisen (accompaniment instrument for BUNRAKU) is also available in this system to control the puppets, by using the sound analyzer as system input device. Since the system's output is actually moving an object (the robot), it looks much more realistic than a CRT image display.
This paper describes a position estimation method of a wheeled mobile robot by integrating the informations in an odometric dead reckoning and a laser navigation system. The dead reckoning regularly gives the robot's position by the revolution counts of both side wheels. The laser navigation system succesively observes the bearing angles relative to the corner cube reflectors fixed in the robot's environment. The chi-squared hypothesis testing is applied to reliably identify the corner cubes. The identified angle measurements modify the robot's position calculated by the dead reckoning based on the Extended Kalman filtering. The plant model is introduced from the kinematic equation concerning the dead reckoning, which regards both the robot's position and the wheel's radius as the state variables and the encoder measurement as the input variable. The measurement model is built concerning the bearing to a corner cube reflector in environment observed by the scanned laser. The proposed method enables the robot to accurately estimate the position even under uncertainty of the wheel's radius and the robot's motion with slippage in the cluttered environment. The simulation and experimental results justify the proposed method.
This paper proposes a planning method for multiple mobile robots system. It has two characteristics as follows: (1) Each robot plans a path individually. There is no supervisor. (2) The concept of cooperative motion can be implemented. A two layered hierarchy is defined for a scheme of individualrobot path planning. The higher layer generates a trajectory from the current position to the goal. The lower layer called “Virtual Impedance Method” makes a real-time plan to follow the generated trajectory while avoiding obstacles and avoiding/cooperating other robots. This layer is composed of four modules called “watchdog”, “deadlock solver”, “blockade solver” and “pilot”. The local equilibrium is detected by the watchdog and canceled by the deadlock solver or the blockade solver. Simulation results indicate eff eciveness of the proposed method.
This paper studies the performance of both P-type and PI-type learning schemes for robot motion control. P-type learning is one of the simplest scheme that does not use differentiation of velocity signals. Recent theoretical results show that a forgetting factor in P-type learning guarantees the robustness against initialization errors, fluctuation of dynamics, and measurement noises. In this paper, experiments are carried out to confirm the theory, and design guidelines of learning control parameters are clarified on the basis of the experimental results. Moreover, a new PI-type learning algorithm that uses positional signals is proposed, and the uniform boundedness of output signal is shown theoretically. The proposed PI-type learning scheme improves the learning speed and reduces trajectory errors considerably in comparison with the P-type learning scheme.
A learning control scheme for a class of robot manipulators whose endpoint is moving under geometrical constraints on a surface is proposed. In this scheme, the input torque is composed of two different input signals updated at every trial by different laws. One is updated by the angular velocity error vector which is projected to the tangent plane of the constraint surface. The other is updated by the force error at the manipulator endpoint. A theoretical proof of the convergence of force errors when velocity and position trajectories are in a neighborhood of the desired trajectories is presented. Computer simulation results by using a 3 DOF manipulator are presented to demonstrate the efficiency of the proposed method and the convergence of force trajectories besides position/velocity trajectories. A new type of learning that uses a feedforward torque input calculated by an approximate dynamics model of the manipulator is proposed to accelerate the speed of learning. The efficiency of the proposed method is illustrated by computer simulation.
In this paper, a trajectory tracking control problem of a planar 2 d. o. f. flexible manipulator is discussed. First, the equations of kinematic relationship and equations of motion are derived, using spring/mass model. From these equations, a relationship between tip trajectory and joint input torque is derived. To realize the desired trajectory, the fourth derivative of the desired trajectory with respect to time has to exist. Then a trajectory tracking controller is designed for real-time control of the flexible manipulator neglecting some small terms. A simulation result shows the effectiveness of this method.
In travelling crane systems, the load-swing sometimes has an oscillation component perpendicular to a travelling direction by accidental disturbances. This paper analyzes the dynamics of the load swing with the rope length as another control channel to the crane system and proposes a control strategy both to suppress the load swing and to control the trolley position. In the control, the rope length is variated to move the load upwards and downwards following to a sinusoidal function so that the swing direction of the load rotates. At the same time, the trolley is controlled to eliminate the oscillation component of the load growing in its travelling direction and to approach to a desired position of the trolley. The control is evaluated by numerical experiments and proved to be effective.
An experimental study is presented on the parameter identification of a typical industrial robotic manipulator, PUMA260. First, parameter coefficient equations are derived from a linearized equation of the robot to clarify its base parameters and a motion planning strategy is proposed for accurate identification. Next, the effects of manipulator configuration, angular velocity and angular acceleration on the identification are experimentally investigated and desirable motion conditions are determined. Moreover, the feasibility of a reduced dynamic model consisting of parameters that dominantly contribute to joint torques is studied. Finally, the proposed parameter identification technique is validated by manipulator trajectory control experiments.
In this paper, we present a robust and flexible automatic inspection method of insulators in outdoor power plants. In outdoor environment, images of the objects to be inspected are varied both by weather and by time of inspection. To overcome this difficulty, we introduced two features to our method. One is that geometric features of insulators and their relations are modeled both for finding insulators and for detecting foreign articles on them. The other is that fifteen features are integrated into one inspection result. These two features make our method robust even in the outdoor environment. Our method consists of three steps; (1) extraction of each insulator from a scene. (2) extraction of features such as projection, histogram, line fitting etc, and (3) detection of defective regions by integrating belief value of these features. Though outdoor real image experiments, we comfirmed that our method has as much reliable as human inspectors for detecting articles attached on the insulators.