Path design is one of the bases of autonomous control of mobile robots. Smoothness is necessary for the path from the viewpoint of dynamic robot motion. Direction of the robot on the path is important from the viewpoint of its interaction with its environment. So not only points through which the path goes, but also a tangent at each point are preferable for specifyng the path in a space because most of wheel-type mobile mechanisms cannot control their direction independently from their position on the path. In order to satisfy these conditions a path design using B-spline curves is presented. which has a continuous 2 nd derivative, that is, a continuous curvature, and which passes through given points with a tangent specified at each point. To control a mobile robot to move along the designed curve an algorithm is presented, which decides speed pattern for a two-wheel-driven-type mobile mechanism with a minimum travel time. Feasibility of using the designed spline curve for trajectory control is demonstrated using an experimental hardware robot with two wheels driven independently.
As space robots are used in zero gravity environment, there are some technical problems; the control scheme of the manipulator considering the reaction movement caused by the manipulator motion, the recognition scheme of 6 D. O. F. motion of the target, and so on. To solve these problems, we have constructed the ground simulation system combining numerical simulation and servo mechanisms. On this system, dynamics of the space robot and the target is solved based on the momentum conservation law, and the relative motion between them is realized. Using this simulation system, we can develop space robots efficiently. In this paper, we describe about the configulation of this system, simulation algorithm, results of evaluation tests.
To realize intelligent robot, it is necessary to establish an effective method for visual and tactile recognition. Much of the existing work on tactile recognition has assumed that tactile sensors derive only data on object surfaces so as to generalize recognition algorithm. With tactile sensors which provide local features of contact areas, e. g, vertex, edge, etc., it is possible to prune inconsistent interpretations more effectively than existing methods. This paper proposes a method for object identification using local feature sensors. In the method, interpretations at each step of identification process are represented as a set of subsets of 3-D space where a sensor to be located. With the progress of identification process, the set will be narrower. Consequently local constraints become strict and inconsistent interpretations will be efficiently pruned. In the pruning, local constraints are compared considering the error bounds of measurement in sensor data.
This paper presents a method for building a 3-D world model for a mobile robot from sensory data derived from outdoor scenes. The 3-D world model consists of four kinds of maps: a physical sensor map, a virtual sensor map, a local map, and a global map. First, a range image (physical sensor map) is transformed to a height map (virtual sensor map) with respect to the mobile robot. The height map is segmented into unexplored, occluded, traversable and obstacle regions from the height information. Moreover, obstacle regions are classified into artificial objects or natural objects according to their geometrical properties such as slope and curvature. A drawback of the height map-recovery of planes vertical to the ground plane-is overcome by using multiple height maps which include the maximum and minimum heights for each point on the ground plane. Multiple height maps are useful not only for finding vertical planes but also for mapping obstacle regions into the video image for segmentation. Finally, height maps are integrated into a local map by matching geometical parameters and by updating region labels. We show the results obtained using landscape models and ALV simulator of the University of Maryland.
In this paper, a new vibration control method for a space flexible manipulator is proposed from the viewpoint of practicability. We considered a N link SCARA flexible manipulator. The features of the proposed method are the following: (1) Basically, the control system consists of simple local feedbacks of joint angle and joint torque. Therefore, it is easy to adapt this method to constructing a controller and sensor systems in a practical system. (2) The asymptotic stability of this feedback scheme is proved for a distributed system. (3) On the other hand, the proposed method is not an optimal control method because, a dynamic compensator is not used, therefore this method is a low authority control. First the dynamics of a manipulator is derived by Hamilton's principle. Then, the control method is derived based on the artificial potential. The stability of the proposed method is also proved by the Lyapunov method of stability analysis. The validity of this method was proved by the experiments using a two link SCARA flexible manipulator “TESRA-I” Experimental results showed that the manipulator was controlled precisely without vibration.
A discrete-time learning control for robotic manipulators is studied system theoretically with a pulse transfer functions. Firstly, the learning stability condition for the discrete-time learning control is derived in frequency domain. From that condition, we derive the new learning stability condition which is applicable for the case of which output number is larger than input number. This is more common as in the case of voltage input position and velocity outputs. Secondly, we examine the learning algorithm with the position signal. When the sampling period is small, the learning stability is degraded because of the system's zeros which are generated by discretizing Thirdly, we examine the learning algorithm with position and velocity signals. In this case, we can stabilize the learning control systems which are unstable in the case of the learning with only the position signal. Finally, we show the simulation results on the trajectory control of robotic manipulators using the discrete-time learning control. The simulation results agreed well with the analytical ones.
Fundamental data as for partitioning of functions between human operators and computers in teleoperation systems are offered in this paper. These data were experimentally obtained by using a simulator. The simulator is composed of three components: a hypothetical robot manipulator capable of moving in two dimensional plane; an aiding device; input devices necessary in a manual mode of teleoperation. The aiding device can operate the robot hand to substitute a human operator for a while. It makes clear that the aiding device is very useful, especially in the case that human operator is a beginner or that task is too difficult to do, for example, the degree of freedom of the manipulator's motion is high.