Browsing by Subject "Backpropagation"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
Item Open Access Multi-stage neural networks with application to motion planning of a mechanical snake(IEEE, 2003-06) Fidan, B.; Sezer, Erol; Akar, M.An efficient approach to nonlinear control problems in which the plant is to be driven to a desired state using a neural network controller in a number of steps is by representing the whole process as a multi - stage neural network. In this paper, an explicit formulation of the back propagation algorithm is developed for such networks. Later, this approach is used to build up a path planner for a mechanical snake (a robot composed of a sequence of articulated links). This path planner, together with a tracking algorithm, is shown to get the mechanical snake out of a collision-free closed region.Item Open Access Neural networks for improved target differentiation and localization with sonar(Pergamon Press, 2001) Ayrulu, B.; Barshan, B.This study investigates the processing of sonar signals using neural networks for robust differentiation of commonly encountered features in indoor robot environments. Differentiation of such features is of interest for intelligent systems in a variety of applications. Different representations of amplitude and time-of-flight measurement patterns acquired from a real sonar system are processed. In most cases, best results are obtained with the low-frequency component of the discrete wavelet transform of these patterns. Modular and non-modular neural network structures trained with the back-propagation and generating-shrinking algorithms are used to incorporate learning in the identification of parameter relations for target primitives. Networks trained with the generating-shrinking algorithm demonstrate better generalization and interpolation capability and faster convergence rate. Neural networks can differentiate more targets employing only a single sensor node, with a higher correct differentiation percentage (99%) than achieved with previously reported methods (61-90%) employing multiple sensor nodes. A sensor node is a pair of transducers with fixed separation, that can rotate and scan the target to collect data. Had the number of sensing nodes been reduced in the other methods, their performance would have been even worse. The success of the neural network approach shows that the sonar signals do contain sufficient information to differentiate all target types, but the previously reported methods are unable to resolve this identifying information. This work can find application in areas where recognition of patterns hidden in sonar signals is required. Some examples are system control based on acoustic signal detection and identification, map building, navigation, obstacle avoidance, and target-tracking applications for mobile robots and other intelligent systems. Copyright © 2001 Elsevier Science Ltd.Item Open Access Target classification with simple infrared sensors using artificial neural networks(IEEE, 2008-10) Aytaç, T.; Barshan, BillurThis study investigates the use of low-cost infrared (IR) sensors for the determination of geometry and surface properties of commonly encountered features or targets in indoor environments, such as planes, corners, edges, and cylinders using artificial neural networks (ANNs). The intensity measurements obtained from such sensors are highly dependent on the location, geometry, and surface properties of the reflecting target in a way which cannot be represented by a simple analytical relationship, therefore complicating the localization and classification process. We propose the use of angular intensity scans and feature vectors obtained by modeling of angular intensity scans and present two different neural network based approaches in order to classify the geometry and/or the surface type of the targets. In the first case, where planes, 90° corners, and 90° edges covered with aluminum, white cloth, and Styrofoam packaging material are differentiated, an average correct classification rate of 78% of both geometry and surface over all target types is achieved. In the second case, where planes, 90° edges, and cylinders covered with different surface materials are differentiated, an average correct classification rate of 99.5% is achieved. The method demonstrated shows that ANNs can be used to extract substantially more information than IR sensors are commonly employed for. © 2008 IEEE.