Recognition of the current location is important for a mobile robot to follow the planned path. Landmarks, which are assumed in the environment around the robot, are used to support this recognition. A mobile robot basically estimates its current location by dead-reckoning using the information of internal sensors. Information obtained by landmark measurement is used to cancel the estimation error which will grow as the travelling distance increases. In this paper we propose a method to select the most suitable landmark among the measurable landmarks from a current position by evaluating the update of the estimation error distribution. We use the extended Kalman filter to update the distribution of estimation error. We show the update characteristics of two types of landmarks, such as line landmarks and point landmarks. For the evaluation of the update we use the width of the distribution ellipse perpendicular to the direction of the path. Through the navigation simulation the effectiveness of the proposed method is confirmed.
We have been developing a novel method to program a robot, an APO (Assembly Plan from Observation) method. A human performs assembly operations in front of the APO system's TV camera. The APO system recognizes such assembly operations and generates an assembly plan to repeat the assembly operations using its robot arm. Our system recognizes object configurations after each of assembly operation and then extracts a task model based on a face contact relation. Motion parameters necessary to execute the operation are calculated from oject configurations. A task model is obtained reasonably well from a noisy object recognition result. However, motion parameters are affected by errors in a recognition result. The errors in motion parameters may cause failure of the task execution. This paper proposes a method to correct erroneous motion parameters based on face contact contact relations. In order to maintain the relationship among motion parameters of previous operations, we define an operation dependency list (ODL), a symbolic list of homogeneous transformations to represent operations. We implement this method in the APO system, apply the method to several assembly examples, and verify the effectiveness of the method.
A novel definition of singular points of mechanisms is proposed in this paper. The linear equation of infinitesimal motion of a mechanism is derived by differentiating the equations of displacement in loops of a mechanism. Three matrices are determined as the coefficients of three vectors of infinitesimal displacements of input-, output-, and passive-joints. Singular points are classified in accord with the ranks of the matrices and the relation of inclusion of the linear spaces determined by the matrices.
The authors have been using the ZMP (Zero Moment Point) as a criterion to distinguish the stability of walking for a biped walking robot that has a trunk. In this paper, the authors propose a learning control algorithm of the compensative trunk motion that makes the actual ZMP get closer to the desired ZMP. The convergency of the algorithm is confirmed by computer simulation. By the learning experiments with the biped robot, the reappearance of the measured ZMP is shown and the convergency of the algorithm is confirmed. The change of the convergence rate with the change of the weight coefficient multiplied to the errors between the measured ZMP and the disired ZMP is confirmed by the simulation and the experiments. And also the reasons are discussed.
We have experimentally examined to estimate the parameters which appear in the dynamic model of PUMA 560 applying the two identification methods ; “step-by-step method” and “simultaneous method”.The instrumental variables method is also used in applying the “simultaneous method”. To evaluate the accuracy of the estimates we simulated the motion of the manipulator using the estimates and compared the simulated trajectories with measured trajectories. It can be concluded that the “step-by-step method” is more precise way to estimate the parameters than the “simultaneous method”, but the “simultaneous method” taking advantage of the instrumental variables method is neary as precise way as the “step-by-step method”. We have also described in detail the contents of the work which is needed to obtain the estimates about each identification method, and compared the identification methods about the labour and consuming time on the computer. If the instrumental model is constructed appropriately we can obtain the good estimates using the instrumental variables method in the “simultanous method”. We have proposed one method to determine the parameter values in the instrumental model which was already proposed. It is time consuming but very easy.
A new approach is presented for designing compliance matrices for assembly tasks using linear programming. Compliance plays a significant role in performing manipulation tasks which involve contact with rigid objects such as assemblies. However, there is no useful method for designing a compliance matrix which' can be adapted to each task. In this paper, compliance is treated as a linear mapping from a force to a corrected motion and is constructed on the basis of geometric information. First, a basic framework of assembly task strategies is described. This is a hierarchical framework based on primitive motions. The primitive motion is defined as a motion that accomplishes a transition from an initial state to a goal state, given a single compliance matrix and a single motion command. The presented method designs compliance matrices for primitive motion. Next, the three conditions required of an appropriate compliance matrix, stability, feasibility, and error correctability, are discussed. These conditions are then formulated as a set of linear inequalities with respect to compliance matrix elements, and the compliance matrix design problem is reduced to a linear programming problem, with this set of inequalities as constraints. Finally, the usefulness of this method is shown by a box palletizing task.
We have been using the ZMP (Zero Moment Point) as a criterion to distinguish the stability of walking for a biped walking, robot which has a trunk. In this paper, we introduce a control method of dynamic biped walking stabilized by trunk motion compensating for the three-axis moment on an arbitrary planned ZMP. We developed a biped walking robot WL-12 RV (Waseda Leg-12 Refined V) and perfomed a walking experiment with WL-12 RV using the control method. As a result, we realized fast dynamic biped walking (walking speed 0.54 s/step and step length 0.3 m) on a flat floor. This walking speed is about 50 percent faster than that with WL-12 (Waseda Leg-12) which compensates for only the two-axis (pitch-axis and roll-axis) moment by trunk motion.
In this paper, we will design path tracking controller for an articulated vehicle with double trailers using time scale transformation and exact linearization technique, which are developed in nonlinear control theory. The controller will be designed in the following way. Firstly, we will describe the dynamics of the vehicle in a state equation whose time scale is the distance along the desired path (not the actual time t) . Then, we will exactly linearize this state equation using coordinate transformation and nonlinear feedback, and design a linear stabilizing controller for the linearized state equation. We will show, in the simulation, that the proposed controller works satisfactorily, and the articulated vehicle with double trailers follows the desired path even when it is moving backward. We can design path tracking controller for an articulated vehicle with multi trailers using the same strategy.
This paper is concerned with some control experiments of one-link flexible manipulator by using a hardware experimental system. A torque servo actuator has been developped with an accurate torque sensor and a torque servo controller. The torque servo actuator (TSA) is insalled on the articular joint to achieve a fine control. To realize a stable control of the structurally flrxible manipulators, two control methods, i. e., PD-control and pseudo resolved acceleration control (pRAC), are examined through the experimental system. The effectiveness of TSA and pRAC are successfully demonstrated.
This paper is concerned with some control experiments of two-link flexible mainpulator by using a hardware experimental system. A torque servo actuator (TSA) has been developped, which is insalled on each articular joint to achieve a fine torque control. To realize stable control of structurally flexible manipulators, three control methods, i. e., extended local PD-control, pseudo resolved motion rate control and pseudo resolved acceleration control, are examined through the experimental system. The effectiveness of TSA and the control methods are successfully demonstrated.