This paper describes a new wireless packet transfer mechanism called MACS that supports open and distributed multi-robot environments. In point-to-point communication in such an environment, node name management, routing decision, and collision occurrence are serious problems. MACS provides a group communication facility where members of the group are selected dynamically and autonomously by user-defined conditions. It implies that the node name management or the routing problem does not occur. Furthermore, MACS avoids the occurrence of packet collision by introducing a new time-slot assignment mechanism called TDMA/TP (TDMA in Temporal and Partial area) . The paper also describes our implementation and performance evaluation of MACS on our autonomous mobile robots. The results suggest that MACS successfully performs conditioned multicast communication with dynamically selected members in a multi-robot environment.
In this paper, we discuss the control of a 3D (3-dimensional) spatial flexible manipulator. In the 3D spatial flexible manipulator, object parameters for the control change depending on the arm configuration. Therefore, gains in the controller for the vibration suppression have to be changed configuration-dependently. In this paper, we propose a configuration-dependent adaptive controller in which the joint accelerations are used as commands to suppress the vibration. We discuss two configuration-dependent control schemes. One is a scheme which adds damping to the vibration system. We call this scheme as ‘active damping control.’ The other is a scheme which uses the optimal regulator. These schemes are combined to be implemented on a 3D flexible manipulator. Experimental results show the effectiveness of the control schemes. In the experiment, we find discontinuous gains of the optimal regulator for some arm configurations. Also the plausible reasons for discontinuity are briefly discussed.
Tele-existence is an advanced type of teleoperation system that enables a human operator to perform remote manipulation tasks dexterously with the feeling that he or she exists in the remote anthropomorphic robot in the remote environment. In order to extend the function of a human, an extended tele-existence system with virtual reality technology was proposed. We improved our tele-existence experimental system and made an extended tele-existence experimental system with virtual environments for training of tasks. In this paper, the configuration of the system is reported. The performance of the system is evaluated by experimental operations.
This paper proposes a navigation system for a mobile robot moving in an outdoor environment with brief route information. Our approach to visual navigation is based on the ability to find and track road boundaries in the environment. The brief route information for navigating the robot is represented with motion directions (e.g. right, left) at the intersections and metric distances between the intersections. The system consists mainly of two modules. One is to move along a road autonomously. The other is to monitor the robot actions and to navigate it so that the robot does not deviate from the route indicated by the route information. Generally speaking, it is difficult to navigate the robot in an outdoor environment since the ground is not flat and the images taken through the camera contain some noises. We have evaluated its navigational capability with experimentations in the real environment. As a result, it is confirmed that the system can robustly navigate the robot over a long path.
This paper proposes a simple method for applying popular PID position-controlled industrial robots to cooperative tasks with force interactions among them. For handling single object by multiple robots, force control using torque controllers is generally used to avoid excessive inner forces, however, a popular industrial robot has a PID positioncontroller. Thus it is usually very difficult to realize cooperative control with industrial robots. This paper proposes a method to introduce flexible mechanisms at the tips of robots to control forces at the end-effectors. After the installation of flexible mechanisms at the tips of robots, a closed-loop structure including flexible parts is generated. First, the number of degrees of freedom of the closed-loop structure is analized by modeling the installed flexibility as virtual joint variables. Next, the minimum number of springs, required to avoid both excessive inner forces and vibrations caused by the installed flexibility, are derived. Finally, the minimum number of actuators to control the object in arbitrary directions of 3 D space is derived for installed robots.
In this paper, we present a contact motion control method between a robotic manipulator and a workpiece. A contact motion is a very short time phenomenon and the dynamics of the system is changed at the instance of collision. It is therefore difficult to control a contact motion with linear control theory. In previous works, we have applied a learning control scheme to solve these problems for one or two degree of freedom manipulators and have shown some simulation results to verify feasibility of the method. With the proposed method, a dead-beat response of contact force has realized after some iterative operations. In this paper, we discuss the feasibility of the proposed method from an experimental point of view. Firstly, we show a design indication of the settling time which causes a non-overshoot response with finite power. Secondly, we study the relationship between the number of observation and learning convergency in case of existing measuring noise. Thirdly, we discuss an influence of modeling error which is caused by compliance of gears or links of a manipulator. Some experimental results illustrate validity of our algorithm.
This paper proposes an effective strategy of holding a swinging object by two arms for automating heavy parts assembly. Instead of human arms, this research uses two flexible arms to avoid impact forces at the moment of contact. The strategy has three modes, which are following, approaching and cooperation. One arm which touched the target object first follows the object so as to keep the contact. From the position and the deflection of the arm, the exact position of the object can be measured. In the approaching mode, the other arm is also controlled to touch the object by the feedback of the distance to the object. After the holding, two flexible arms reduce the vibration of the object by cooperative control. All the modes are realized by the same structured controllers, which control the desired force generated at the tip and the position. Experimental results show the effectiveness of the strategy.
This paper discusses a dynamic active sensing system capable of detecting the contact point between a flexible beam and an object. The proposed sensing system is simply composed of an insensitive flexible beam, a torque sensor, a joint position sensor, and an actuator. The dynamics (such as moment of inertia and stiffness) of the sensing system changes when the motion phase switches from non-contact to contact. We show that the natural frequency during contact is generally a multi-valued function of the contact location, which is not desirable from the viewpoint of determining the contact location uniquely. It is also shown that by adding a proper mass at the free end of the beam, we can make the function close to a single-valued one. The additional mass also contributes to reducing the oscillation frequency, which releases us from treating high frequency signals. We design the hardware model based on this idea, and verify the basic principle experimentally.
This paper describes modeling of ambiguous matchings in stereo vision and a viewpoint selection for resolving the ambiguities. Many iterative or active methods have been proposed for determining matches between features. These methods, however, do not consider the cost of recognition. In general, visual recognition of an environment requires much computation and the recognition result includes uncertainty. A trade-off, therefore, must be considered between the cost of visual recognition and the effect of visual information to be obtained. In order to consider such a trade-off, a new segment-based stereo matching method is proposed. The initial matches are established by using a local disparity histogram. All matches including ambiguous ones are kept with their reliabilities in the model. The ambiguities will be resolved by subsequent observations if necessary. An efficient method is proposed for determining whether a next viewpoint can resolve an ambiguity. Experiments with real data show the usefulness of the method.
The purpose of this research was to make a machine or a robot jump over a big obstacle or a ditch. They sometimes need to jump to move over these obstacles. We plan to do fundamental researches on a jumping machine to (1) make the jumping machine jump as high as possible, (2) control the direction of the jumping, (3) control the attitude of the machine in the air, and (4) make it land softly. In this paper the method of making the jumping machine land softly in two dimensional space is described. The optimum soft landing of the jumping machine in two dimensional space is defined. This definition prevents the jumping machine from falling and receiving the big force from the ground. We apply this test to the jumping machine when it jumps and lands in two dimensional space and test the validity of the soft landing by experiment.