J. A. Walter and K. J. Schulten, "Implementation of self-organizing neural networks for visuo-motor control of an industrial robot," IEEE Transactions on Neural Networks, vol. 4(1):pp. 86--95, 1993.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....same training method is used, regardless of the type of evaluation. Adapting the training of the output layer to the interpolation method used, may help to increase the results once more. This will be the subject of future research. Another method which is also able to extrapolate is presented in [Wal93] and associates a matrix, representing the local derivative to each competition neuron. This algorithm realizes a linear Taylor expansion. It also uses another method for the training of the competition layer, which makes it difficult to compare. It will be an interesting part in future research ....

Jorg A. Walter and J. Schulten. Implementation of self-organizing neural networks for visuo-motor control of an industrial robot. IEEE Transaction on Neural Networks, 4(1), 1993.

....robot must gain knowledge of its motion control, otherwise known as kinematics, in a given environment. This relates to sensorimotor coordination . How then can the mobile robot be acquainted with its own kinematics Our research rooted off by examining the control architectures of robot arms [5, 8, 9, 14, 15, 36, 37] to identify the underlying concepts behind automatic end effector positioning; these concepts may be applicable to our cause. In doing so, we have to observe that the direct translation of learning and sensorimotor control techniques from robot arms to mobile robots is not possible due to several ....

....Many unsupervised learning algorithms have been dedicated to the application of sensorimotor coordination on robot manipulators and mobile robots. They include evolutionary optimization or genetic algorithms [33] rule based algorithms [34] fuzzy logic [35] artificial neural networks [5, 6, 7, 8, 9, 14, 15, 30, 36, 37, 48, 49, 50, 60] and reinforcement learning [38] The choice of the learning algorithm must take into consideration all the objectives stated in Section 1.2. In particular, artificial neural network fits our purpose extremely well; it replicates the cerebellum of a mammalian brain, which performs sensorimotor ....

[Article contains additional citation context not shown here]

Walter J.A. and Schulten K.J., Implementation of Self-Organizing Neural Networks for Visuo-Motor Control of an Industrial Robot. IEEE Transactions on Neural Networks, 4(1):86-95, 1993.

....and with a fine tuned method of ff i reinitialization which aims to avoid cristallization in local minima, the SCS is able to achieve better quantization (about 1; 5dB of SNR in mean) than the GLA, however after a longer time. 3.3. 4 Neural gas algorithm Recently, Martinetz and Schulten [12, 18] have introduced a new VQ algorithm, called neuralgas . In this algorithm, for each iteration, all units are sorted according to their distance to the input. A rank of closeness k(i) is therefore attributed to each unit i, so that: ffi 0 ffi 1 : ffi k ffi k 1 : ffi N ....

J. A. Walter and K. J. Schulten. Implementation of self-organizing neural networks for visuomotor control of an industrial robot. IEEE Transaction on Neural Networks, 4(1):86--95, 1993.

....b) The third approach entails the creation of a robot trajectory given only its initial and final positions. The robot receives sensory information from the workspace and constructs an inverse kinematic mapping. This approach is used in Kuperstein Rubinstein (1989) Martinez et al. (1990) and Walter Schulten (1993). In this paper we aim at emphasizing the feasibility of applying unsupervised learning to complex robotics problems. We are particularly concerned with the problem of fast and accurate learning of single and multiple robot trajectories. The contribution of this work is twofold: i) development ....

Walter, J.A. & Schulten, K.J. (1993). Implementation of selforganizing neural networks for visuo-motor control of an industrial robot. IEEE Trans. on Neural Networks, 4(1):86-95.

....in temporal sequence learning, and specially the field of robot learning has gained relevant contributions. The vast majority of models is involved in either solving inverse kinematics for visuomotor coordination (Kuperstein Rubistein, 1989; Martinetz et al. 1990; Gaudiano Grossberg, 1991; Walter Schulten, 1993) or route learning (Denham McCabe, 1995; Gaudiano et al. 1996; Heikkonen Koikkalainen ,1997) A robot task that has received contributions from the field of unsupervised neural network is trajectory tracking (Hytyniemi, 1990; Althfer Bugmann, 1995; Bugmann et al. 1998; Barreto Arajo, ....

Walter, J.A. & Schulten, K.J. (1993). Implementation of self-organizing neural networks for visuo-motor control of an industrial robot. IEEE Trans. on Neural Networks, 4(1):86-95.

.... data, i.e. clustering problems (Kohonen (1995) Kaski and Kohonen (1996) In addition, SOMs have also been successful in applications, where temporal or sequential data are processed, for instance, in speech recognition, process control and time series analysis in medicine (Behme et al. 1993) Walter and Schulten (1993), Guimares (2000) In this paper we give a review of SOMs to several application domains, such as sleep apnea, protein sequence analysis and tumor classification. For the diagnosis of sleep apnea the temporal dynamics of physiological parameters such as respiration and heart rate, have to be ....

Walter, J.A., Schulten, K.J. (1993): Implementation of Self-Organizing Neural Networks for Visual-Motor Control of an Industrial Robot, IEEE Transactions on Neural Networks, Vol. 4, No.1, January 86-95.

....use visual feedback but are limited to a particular subtask of the hand eye coordination process. For example, in [13] an approach is given which uses visual data to plan the movement of the manipulator, but no feedback is given as to whether or not the gripper is in the desired position. In [9] a network based on a self organizing topology conserving map is used to learn a mapping from the visual input space to the arm configuration space. Many approaches place a constraint on the positioning or movement of the targeted object as does [4] which assumes that the depth of the plane on ....

Walter, J.A. and K.J. Schulten. Implementation of Self-Organizing Neural Networks for Visuo-Motor Control of an Industrial Robot. IEEE Trans on Neural Networks, 4(1):86--95, Jan 1993.

....a trivial task. Two types of approaches exist. The first type requires the human designer to explicitly model the relationships between sensors and actuators. The result is a set of equations with a number of parameters that need to be estimated. Most existing works belong to this category (e.g. [7]) The second type of approach does not employ explicit parameterized models. But rather, a general approximator is used to map the sensor space to the actuator space. A common tool to accomplish this is the artificial neural network [7] The first type of approach is effective when the system ....

....estimated. Most existing works belong to this category (e.g. 7] The second type of approach does not employ explicit parameterized models. But rather, a general approximator is used to map the sensor space to the actuator space. A common tool to accomplish this is the artificial neural network [7]. The first type of approach is effective when the system can be accurately modeled with a few parameters, but has problems when the system configuration is time varying or has a lot of degree of freedom. The second type does not have the latter limitation due to the generality of the neural ....

Walter, J.A. and K.J. Schulten. Implementation of Self-Organizing Neural Networks for Visuo-Motor Control of an Industrial Robot. IEEE Trans on Neural Networks, 4(1):86--95, Jan 1993.

....how it will be used, and how to organize it for fast and efficient utilization. This paper presents an approach to visually guided robotic manipulation based on the premises given above. There are various approaches which use visual data to guide a robotic manipulator [3, 6, 2, 10] In [7] a network based on a selforganizing topology conserving map is used to learn a mapping from the visual input space to the arm configuration space. Many approaches place a constraint on the positioning or movement of the targeted object as does [4] which assumes that the depth of the plane on ....

Walter, J.A. and K.J. Schulten. Implementation of Self-Organizing Neural Networks for Visuo-Motor Control of an Industrial Robot. IEEE Trans on Neural Networks, 4(1):86--95, Jan 1993.

....likely depend on additional criteria, such as optimizing for various side functionals. Neural network based motor control of limbs or manipulators with redundant dof is an important problem of topical interest, relevant to researchers in learning theory, robotics and neurophysiology, 2] 3] [4], 5] 6] 7] 8] 9] 10] The research reported in these references can be separated into 1 A manifold is a space which is locally, but not necessarily globally, Cartesian. For example, the surface of a sphere is a two dimensional manifold, but cannot be mapped one to one to any ....

....in the direction of that data point. The weight vectors of its neighbors, according to the topology of the net, are also adjusted, though typically by a lesser amount than the adjustment made to the winner. Applications of SOMs to the robot inverse kinematics problem are given in [41] 2] and [4]. In the research reported in these papers, either the robot has no redundant dof, or the redundancy is resolved at training time, with only a single solution along a single branch available at run time. Our results extend these prior approaches by providing a topology preserving mapping such that ....

Jorg A. Walter & Klaus J. Schulten (1993), "Implementation of Self--Organizing Neural Networks for Visuo--Motor Control of an Industrial Robot", IEEE Trans. Neural Networks 4:1, 86--94.

....is required. As noted earlier, connectionism is used in many different fields of science. For example, connectionist networks have been used for aiding astronomical work [106] assisting medical diagnosis [20] regulating investment management [121] and controlling robotic limb movement [113]. Many of these systems, however, are approached from an engineering perspective; that is, the designers are only interested in making the networks as efficient as possible (in terms of network topology, correct responses, and generalization) Consequently, this attitude towards connectionism ....

J. Walter and K. Schulten. Implementation of self-organizing neural networks for visuo-motor control of an industrial robot. IEEE Transactions on Neural Networks, 4:86--95, 1993.

No context found.

Jorg Walter and Klaus Schulten. Implementation of self-organizing neural networks for visuomotor control of an industrial robot. IEEE Transactions in Neural Networks, 4(1):86--95, 1993.

....density distribution of the given input data set. The adaptive neural gas vector quantization scheme offers significant advantages compared to other classic vector quantization methods [21] Visuo motor control of an industrial robot using the neural gas algorithm has been demonstrated in [22]. For the path planning task we consider here, it is also essential to capture the neighbor hood relationships in the sampled data in order to model the exact topology of the PCM. Methods from computational geometry provide means to discover those spatial relationships in a given data structure. ....

J.A. Walter and K. Schulten, "Implementation of self-organizing neural networks for visuo-motor control of an industrial robot," IEEE Transactions on Neural Networks, vol. 4, no. 1, pp. 86-95, 1993.

....consequence, accurate positioning of the SoftArm presents a challenging problem and can only be achieved by an adaptive control mechanism. For a more detailed introduction to the mechanics of the SoftArm see [1] In addition to the previous work on topology representing networks (TRN) in robotics [9, 1], where neighborhood preservation has been used to average over the output of several adjacent units in order to achieve a more accurate positioning, in the present study we focus on exploiting the topology to generate a motion plan from a current position to a given target in a 2 dimensional ....

J.A. Walter and K. Schulten. Implementation of self-organizing neural networks for visuo-motor control of an industrial robot. IEEE Transactions on Neural Networks, 4(1):86--95, 1993. 4

....Unix work station environment in a suitable, convienient, and efficient manner was a central aim for the development of SORMA . The next section explains the problems, that had to be addressed and shows some historic routes. 1 see e.g. Walter, Martinetz, and Schulten 1991; Walter 1991; Walter and Schulten 1993; Littmann, Meyering, Walter, Wengerek, and Ritter 1992; Walter and Ritter 1996b J. Walter and H. Ritter Delta SORMA 6 A Software Architecture: SORMA 1.1 Experiences that led to the SORMA Design The Neural network Simulation Tool (NST) is a software framework developed by Ritter (1995, 1996) ....

Walter, J. and K. Schulten (1993). Implementation of self-organizing neural networks for visuo-motor control of an industrial robot. IEEE Transactions in Neural Networks 4(1), 86--95.

....efficiency. Keywords: Service Object Request Management Architecture; shared, distributed resources; economy of reusing components; communication cost; built in interactivity; time optimal and protected invocation. 1 Introduction The experience of building and operating robotvision labs [5, 10, 12] shows, that a substantial amount of effort easily dissipates in adaption of software components to application specific needs. A lot of interesting ideas and application code is developed, but too often those remain as one of a kind pieces. Descriptions like short life time, little use and ....

Jörg Walter and Klaus Schulten. Implementation of self-organizing neural networks for visuo-motor control of an industrial robot. IEEE Transactions in Neural Networks, 4(1):86--95, 1993.

....suitable conditions, the proposed learning scheme leads to the ability of one shot learning from a single observation of a calibration point whenever the camera has been moved to a new location. This compares favorably to earlier approaches that required a longer re learning phase in that case [17]. A recent extension of the new approach to the binocular case will be reported elsewhere [15] Two features that distinguish PSOMs from the majority of other neural learning algorithms are particularly important for the practical feasibility of the hierarchical learning scheme. First, the mapping ....

Jšorg Walter and Klaus Schulten. Implementation of self-organizing neural networks for visuomotor control of an industrial robot. IEEE Transactions in Neural Networks, 4(1):86--95, 1993.

No context found.

J. A. Walter and K. J. Schulten, "Implementation of self-organizing neural networks for visuo-motor control of an industrial robot," IEEE Transactions on Neural Networks, vol. 4(1):pp. 86--95, 1993.

No context found.

J. A. Walter and K. J. Schulten. Implementation of self-organizing neural networks for visuo-motor control of and industrial robot. IEEE Transactions on Neural Networks, 4(1):86--95, 1993.

No context found.

J. A. Walter and Klaus J. Schulten. Implementation of self-organizing neural networks for visuo-motor control of an industrial robot. IEEE Trans. Neural Networks, 4:86--95, January 1993.

No context found.

J. A. Walter and Klaus J. Schulten. Implementation of self-organizing neural networks for visuo-motor control of an industrial robot. IEEE Trans. Neural Networks, 4:86--95, January 1993.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC