This paper presents a method for automatic sensor placement for model-based robot vision. In such a vision system, the sensor often needs to be moved from one pose to another around the object to observe all features ...
详细信息
This paper presents a method for automatic sensor placement for model-based robot vision. In such a vision system, the sensor often needs to be moved from one pose to another around the object to observe all features of interest. This allows multiple three-dimensional (3-D) images to be taken from different vantage viewpoints. The task involves determination of the optimal sensor placements and a shortest path through these viewpoints. During the sensor planning, object. features are resampled as individual points attached with surface normals. The optimal sensor placement graph is achieved by a geneticalgorithm in which a min-max criterion is used for the evaluation. A shortest path is determined by Christofides algorithm. A Viewpoint Planner is developed to generate the sensor placement plan. It includes many functions, such as 3-D animation of the object geometry, sensor specification, initialization of the viewpoint number and their distribution, viewpoint evolution, shortest path computation, scene simulation of a specific viewpoint, parameter amendment. Experiments are also carried out on a real robot vision system to demonstrate the effectiveness of the proposed method.
The fact that muscles are composed of different Motor Units (MUs) is often neglected when investigating motor control by macro models of human musculo-skeletal-joint systems. Each muscle is associated with one control...
详细信息
The fact that muscles are composed of different Motor Units (MUs) is often neglected when investigating motor control by macro models of human musculo-skeletal-joint systems. Each muscle is associated with one control signal. This simplification leads to difficulties when mechanical and electrical manifestations of the muscle activity are juxtaposed. That is why a new approach for muscle modelling was recently proposed (Journal of Biomechanics 2002;35:1123-1135). It is based on MUs twitches and a hierarchical genetic algorithm (HGA) is implemented for choosing the moments of activation of the individual MUs, thus simulating the control of the nervous system. Its basic benefit is obtaining the complete information about the mechanical and activation behaviour of all MUs, respectively muscles, during the whole motion. Its possibilities are demonstrated when simulating fast elbow flexion. Three flexor and two extensor muscles, each consisting of approximately real number of different types of MUs, are modelled. The task is highly indeterminate and the optimization is performed according to a fitness function that is an assessed combination of criteria (minimal deviation from the given joint moment, minimal total muscle force and minimal MUs activation). The influence of the weight of the first criterion (the one that reflects the importance of the movement accuracy on the predicted results), is investigated. Two variants concerning the muscle MUs structure are also compared: each muscle is composed of four distinct types MUs and the MUs twitch parameters are uniformly distributed. (C) Elsevier Ltd. All rights reserved.
A method of hierarchical genetic algorithms (HGA) is used for optimizing the element spacing and lengths of Yagi-uda antennas. This scheme has the ability of handling multiobjective functions as well as the discrete c...
详细信息
ISBN:
(纸本)0780378296
A method of hierarchical genetic algorithms (HGA) is used for optimizing the element spacing and lengths of Yagi-uda antennas. This scheme has the ability of handling multiobjective functions as well as the discrete constraints in the numerical optimizing process. Together with the technique of Pareto ranking scheme, more than one possible solution can be obtained. It has been found that the number of dipoles of the antenna can be minimally identified with multi-facet design criteria as well as the imposed stringent constraints on the antenna design. Furthermore, This added feature also enables a design tradeoff between cost and performance without extra computational effort.
One of the most challenging problems that occurs in the design of real-time embedded systems is to take into account the stochastic nature of different timing parameters which affect the system performance. In this pa...
详细信息
One of the most challenging problems that occurs in the design of real-time embedded systems is to take into account the stochastic nature of different timing parameters which affect the system performance. In this paper we propose a stochastic framework for hardware-software co-synthesis whereby the task execution times, the data communication times and the input arrival. times are all assumed to be random variables. Based on this framework, a hierarchical genetic algorithm has been developed which explores the state space of possible architectures and pro duces a set of evolved ones which are optimized with respect to cost and performance. This is coupled with a stochastic scheduling algorithm which generates high performance stochastic schedules for the tasks and estimates the overall deadline meeting probability of the system.
A wireless local area network (WLAN) is designed for an IC factory in Hong Kong using the hierarchical genetic algorithm (HGA), The HG,S is capable of handling multiobjective functions and discrete constraints. Becaus...
详细信息
A wireless local area network (WLAN) is designed for an IC factory in Hong Kong using the hierarchical genetic algorithm (HGA), The HG,S is capable of handling multiobjective functions and discrete constraints. Because of this uniqueness, together with the adopting of a Pareto ranking scheme, a solution can be reached even when skewed multiobjective functions and constraints confinements are being imposed. It has been found from this study that a precise number of base stations can be identified for the WLAN network, while it can satisfy a number of objectives es and constraints, This added feature provides a further design tradeoff between cost and performance at no extra effort.
A hierarchical approach for nesting two-dimensional shapes based on geneticalgorithms is described. For the higher-level search, a representation that facilitates genetic search based on recombination is developed. A...
详细信息
A hierarchical approach for nesting two-dimensional shapes based on geneticalgorithms is described. For the higher-level search, a representation that facilitates genetic search based on recombination is developed. An alternative to overlap computation based on assembly of polygons is used at the lower level-of search. Two implementations to find minimum-area enclosures for polygons, with and without a constraint on the width of stock, are discussed. Sample output illustrating the effectiveness of the approach is provided.
暂无评论