Search

An analytically tractable potential field model of free space is presented in this paper. The model assumes that the border of every two-dimensional (2-D) region is uniformly charged. It is shown that the potential and the resulting repulsion (force and torque) between polygonal regions can be calculated in closed form. By using the Newtonian potential function, collision avoidance between object and obstacle thus modeled is guaranteed in a path planning problem. A local planner is developed for finding object paths going through narrow areas of free space where the obstacle avoidance is most important. Simulation results show that not only does individual object configuration of a path obtained with the proposed approach avoid obstacles effectively, the configurations also connect smoothly into a path. Index Terms - Artificial potential field, free-space representation, obstacle avoidance, path planning.

The sample-path optimization technique, used for minimizing a function X(sub infinity) that is an almost-sure limit of a computable sequence of random functions Xn, is examined. The method is presented in terms of a sequence of extended-real-valued stochastic processes Xn(theta) that converge almost surely to a function X(sub infinity)(theta). A possible advantage of sample-path optimization is that, since we are considering the deterministic function Xn(omega,.), we can use deterministic optimization techniques which have been devised in the first half of the century.

An algorithmic approach for testing whether a linear time-varying model is perfect-controllable is used to demonstrate that, under some weak conditions, target path controllability of a general nonlinear model is correspondent to that of the equivalent time-varying linearization about a certain trajectory. This finding indicates that for testing the target path controllability, a policy maker may limit himself to work with the linearization of a non-linear model.

A unified treatment of central path algorithms is described and ways of arriving at methods of center, penalty function methods, potential reduction methods and primal-dual methods are presented. Analysis of methods show the close relationship between good properties of points near the central path and nice primal-dual properties at these points. Different algorithms can be obtained from different parametrizations of the central path, which can then be expressed through polynomial algorithms. Thus, linear programming problems can be approached using primal-dual properties rather than the central path.

Hargens and Herting (1990) note that studies evaluating the reliability of referee assessments of manuscripts have underestimated the reliability of the review process for two reasons. First, past studies have assumed equal intervals between the categories used in assessing manuscripts (e.g., accept, accept after minor revision, revise and resubmit, and reject). Second, the procedures used have measured the reliability based on a single referee's evaluation for each manuscript rather than on a composite of two or more referees. This paper shows how to estimate the reliability based on one, two or any number of referees, and we apply these procedures to estimate the reliability of journal reviews based on one, two, or three referees for each of the journals examined by Hargens and Herting.

Life-history traits such as longevity and fecundity often show low heritability. This is usually interpreted in terms of Fisher's fundamental theorem to mean that populations are near evolutionary equilibrium and genetic variance in total fitness is low. We develop the causal relationship between metric traits and life-history traits to show that a life-history trait is expected to have a low heritability whether or not the population is at equilibrium. This is because it is subject to all the environmental variation in the metric traits that affect it plus additional environmental variation. There is no simple prediction regarding levels of additive genetic variance in life-history traits, which may be high at equilibrium. Several other patterns in the inheritance of life-history traits are readily predicted from the causal model. These include the strength of genetic correlations between life-history traits, levels of nonadditive genetic variance, and the inevitability of genotype-environment interaction.

A heuristic for calculating the path for maximizing the expected utility of a stochastic acyclic network is presented. The heuristic reduces computational complexity by reducing the size of the network. The method obtains a group of attractive solutions by combining Monte Carlo simulation and shortest path techniques. The emphasis is on shortest route problems in which outcomes are measured by a general, nonlinear utility function. The global optimum is found by the enumeration of all feasible paths. A reduction algorithm based on stochastic dominance was developed to speed up calculations and evaluated using Monte Carlo simulation. Some 70 test problems involving 20 to 60 nodes were generated randomly and analyzed. The results suggested that the heuristic resulted in considerable computational savings as the network increased in size, and that the quality of the reduced network solutions was superior to the quality of those derived from the original formula.

A novel path-following algorithm that utilizes a uniform linear sensor array is proposed for signal source localization. The proposed algorithm consists of an algebraic relation between the actual source bearing and range and the bearing of a source under a plane wave assumption. This algebraic relation is applied as a path to follow the peak of the two-dimensional MUSIC spectrum. It is shown that the algorithm is computationally efficient and is able to reduce the number of searches to n independent one-dimensional searches following the known algebraic path.

The model most widely used for path analysis of grain yield in small grains assumes bidirectional causal pathways between yield components; however, because yield components develop sequentially, it is questionable from a biological standpoint whether a yield component could influence others that develop earlier. The purpose of this research is to describe the theory and application of an alternative model for path analysis of grain yield that takes into account the sequential development of yield components. Path analysis was performed assuming bidirectional and unidirectional causal relationships among yield components of eight spring barley (Hordeum vulgare L.) cultivars grown at Palmer and Fairbanks, AK, in each of two years. Yield and the yield components spikes per square meter, kernels per spike, and kernel weight were estimated. The unidirectional model seems to reflect operative causal pathways more realisitically, and results in simpler estimates that present a clearer picture of the relationships among grain yield and yield components. Of the three yield components, spikes per square meter resulted in the greatest positive contribution to yield, while the effect of kernels per spike on kernel weight was the most negative compensatory response among yield components., Two models for path analysis of grain yield in barley were developed under two assumptions of yield component causal relationships. One model assumed bidirectional and the other unidirectional relationships among yield, spikes per square meter, kernels per spike and spike weight. The unidirectional model seemed to give a simpler and more realistic picture of yield and yield component interactions. Among the yield components, spikes per square meter had the most positive effect on yield, while the effect of kernels per spike on kernel weight was the most negative compensatory response.