The corners and the middle points, which are extracted as features from the line approximation of a given pattern, are overlaid on a radial grid to form the input array for training a backpropagation network for class...
详细信息
The corners and the middle points, which are extracted as features from the line approximation of a given pattern, are overlaid on a radial grid to form the input array for training a backpropagation network for classification. The proposed method is shown to be simple and robust by extensive testing of its performance on patterns both with and without noise.
This paper presents a texture segmentation algorithm based on a hierarchical wavelet decomposition. Using Daubechies' four-tap filter, an original image is decomposed into three detail images and one approximate i...
详细信息
This paper presents a texture segmentation algorithm based on a hierarchical wavelet decomposition. Using Daubechies' four-tap filter, an original image is decomposed into three detail images and one approximate image. The decomposition can be recursively applied to the approximate image to generate a lower resolution of the pyramid. The segmentation starts at the lowest resolution using the K-means clustering scheme and textural features obtained from various sub-bands. The result of segmentation is propagated through the pyramid to a higher resolution with continuously improving the segmentation. The lower resolution levels help to build the contour of the segmented texture, while higher levels refine the process, and correct possible errors.
The proposed update technique for computer emulated resistive grids propagates the node activities to and fro, thereby generating within only a few sweeps a well-spread potential distribution suitable for performing a...
详细信息
The proposed update technique for computer emulated resistive grids propagates the node activities to and fro, thereby generating within only a few sweeps a well-spread potential distribution suitable for performing a gradient search for path planning in configuration space. The typical shapes of the forbidden areas in the manipulator's configuration space make this technique particularly fast. Tests with a real manipulator confirmed the validity of the approach.
Adaptive input precompensators in conjunction with nonlinear controllers for multi-link flexible manipulators are considered in this paper. In an earlier paper, we had shown that application of a nonlinear inner-loop ...
详细信息
Adaptive input precompensators in conjunction with nonlinear controllers for multi-link flexible manipulators are considered in this paper. In an earlier paper, we had shown that application of a nonlinear inner-loop control reduces the variations in frequencies due to the geometrical configuration for multi-link flexible manipulators. This results in a better performance when input preshaping or any other controller based on a linear model is designed. To improve the performance of the system to parameter variations (e.g. changes in payload), an adaptive version of the advocated controller is utilized. This is achieved by estimating the time of application of the impulses for on-line preshaping and in the case of payload uncertainty, estimation of the payload and real-time adjustment of the nonlinear inner-loop based controller. Frequency domain Time-Varying Transfer Function Estimate (TTFE) and Empirical Transfer Function Estimate (ETFE) system identification algorithms are proposed for estimation of vibrational modes and unknown payload. Experimental results on a two-link flexible manipulator with adaptive nonlinear control and preshaping are provided to show the effectiveness of the advocated controllers. Overall, the present paper completely generalizes the adaptive input preshaping technique for multi-link flexible manipulators.
This paper shows that categorical color developed in Yue Du (1994) is a good representation of color space for some robot vision tasks. The authors use SCARF as a test case to compare 'equivalent' RGB, categor...
详细信息
This paper shows that categorical color developed in Yue Du (1994) is a good representation of color space for some robot vision tasks. The authors use SCARF as a test case to compare 'equivalent' RGB, categorical color, and intensity robot vision algorithms on a difficult robot road following task. SCARF uses a full three-dimensional representation of color space to formulate road and off-road colors. Performing categorical color or intensity projection, the authors can model the road and off-road colors as single dimensional probability density functions. This greatly reduces the time to process the color image. Moreover, the authors show that categorical color is a good representation in that it outperforms an equivalent intensity algorithm on the robot tasks. Categorical color, while it cannot outperform the original RGB algorithm, performs well enough to be used in this particular robot application. Its effectiveness is verified by many experiments on difficult road image sequences.
The exploitation of analog VLSI techniques combined with computervision knowledge offers spectacular possibilities. Limitations of current VLSI technologies do not allow to create sensors with extremely complex pixel...
详细信息
The exploitation of analog VLSI techniques combined with computervision knowledge offers spectacular possibilities. Limitations of current VLSI technologies do not allow to create sensors with extremely complex pixel architecture, but the coupling of external CMOS analog processing units is a great solution for rapid low level segmentation processes. This paper presents a novel sensing approach where photo-transduction, multiresolution feature extraction, scale-space integration, and edge tracking combined with sub-pixel interpolation are performed on a mixed-signal (digital-analog) VLSI architecture. The paper also discusses how we implement the curvature primal sketch into the system for higher level scene representation. The main sensory part of this integrated image acquisition system is a CMOS sensor called Multiport Access photo-Receptor (MAR). VLSI also provides means to integrate analog computing, digital controller, and DSP co-processor modules which define a powerful sensory chip set for focal plane image processing. A current version of the MAR sensor which implements 256/spl times/256 pixels includes 16 analog spatial filters which simultaneously compute multiresolution edge maps. This novel smart image sensor approach with associated low level segmentation capability presents good opportunities for real time automated process for the particular case of unstructured environment.
We present an integrated methodology that can be used to extract the depth from a sequence of images. The method combines the ability of stereo processing to acquire highly accurate depth measurements and the efficien...
详细信息
We present an integrated methodology that can be used to extract the depth from a sequence of images. The method combines the ability of stereo processing to acquire highly accurate depth measurements and the efficiency of spatial and temporal gradient analysis. As a result of this integration, depth measurements of a high quality are obtained at a speed approximately ten-times higher than that of stereo processing. Without any a priori information of the locations of the points in the scene, the correspondence problem in stereo processing is computationally expensive. In our approach, we use the spatial and temporal gradient analysis, which has been shown to provide the depth with a high efficiency but limited accuracy, to guide the stereo matching process. Experiments for lateral and axial camera translations have acquired depth results with a mean error of less than 3 percent.
A neural network (NN) control technique for robot manipulators is introduced in this paper. The fundamental robot control technique is the model-based computed-torque control which is subjected to performance degradat...
详细信息
A neural network (NN) control technique for robot manipulators is introduced in this paper. The fundamental robot control technique is the model-based computed-torque control which is subjected to performance degradation due to model uncertainty. NN controllers have been traditionally used to generate a compensating joint torque to account for the effects of the uncertainties. The proposed NN control approach is conceptually different in that it is aimed at prefiltering the desired joint trajectories before they are used to command the computed-torque-controlled robot system (the plant) to counteract performance degradation due to plant uncertainties. In this framework, the NN controller serves as the inverse model of the plant, which can be trained online using joint tracking error. Several variations of this basic technique are introduced. Backpropagation training algorithms for the NN controller have been developed. Simulation results have demonstrated the excellent tracking performance of the proposed control technique.
In this paper a new neural network (NN) control technique for non-model based PD control of robot manipulators are proposed. The main difference between the proposed technique and the existing feedback error learning ...
详细信息
In this paper a new neural network (NN) control technique for non-model based PD control of robot manipulators are proposed. The main difference between the proposed technique and the existing feedback error learning (FEL) technique of Kawato et al. is that compensation of robot dynamics uncertainties is done outside the control loop by modifying the desired input trajectory. By using different NN training signals, two NN control algorithms are developed. One is comparable to that in the FEL technique and the another involves the Jacobian of the PD controlled robot dynamic system. Performances of both controllers with different trajectories and PD controller gains are examined and compared with that of the FEL controller. It is shown that the new control technique is superior.
This paper highlights a ranging sensor that can be calibrated at standoffs of 1 to 3 feet with a depth of field of /spl ap/1 ft. Tests at a 16" standoff have produced a 0.004 inch standard deviation in height mea...
详细信息
This paper highlights a ranging sensor that can be calibrated at standoffs of 1 to 3 feet with a depth of field of /spl ap/1 ft. Tests at a 16" standoff have produced a 0.004 inch standard deviation in height measurement errors. The data density and field of view are programmable. The sensor is positioned robotically, allowing it to provide accurate, shed-range measurements prior to manipulation tasks. The sensor head is relatively inexpensive (/spl ap/$10 k) and is designed for in situ calibration. This approach is intended to allow replacement components to be installed, calibrated, and validated without decontamination. Calibration, validation and data acquisition are all designed to be performed remotely and in a highly automated fashion. These in situ features of the design are targeted for the interest of industry with the aim of producing a sensor having a longer lifetime and a lower maintenance cost.
暂无评论