Catálogo de publicaciones - libros

Continuum Probe Designer by Acoustic Ideas Inc. is a tool that can help design the “best” phased array ultrasonic transducer for a given inspection task. Given a specific surface geometry for the ultrasonic transducer, one component of Continuum Probe Designer can determine the number of elements, and the required size and shape of each element to meet a list of ultrasonic inspection goals. Using the number of elements as a cost function, an optimization problem to find the best surface geometry for the transducer is created. Previous work has demonstrated that a (1+)-evolution strategy (ES) can be a very effective search technique for this problem. The performance of this ES was improved by starting it from “smart” (i.e. better than random) start points. Particle swarm optimization (PSO) can be used to improve the “smart” start points, and the overall PSO-ES hybrid is capable of finding feasible transducer designs from all of the start points in a benchmark test suite. This level of performance is an important step towards the use of Continuum Probe Designer as a fully automated tool for the design of phased array ultrasonic transducers.

In this paper, we describe a method for finding the pose of an object from a single image. We assume that we can detect and match in the image three feature points of the object, and that we know their relative geometry on the object. At first we present the exact pose calculation with an existing method and emphasize the limitation. Then we introduce a new method which consists of adding to the camera an inclinometer so that we reduce the number of unknown parameters and thus are able to compute the pose efficiently by using a classical iterative optimization method.

Underground mining method selection (UMMS) is one of the most important decisions that should be made by mining engineers. Choosing a suitable underground mining method to carry out extraction from a mineral deposit is very important in terms of the economics, safety and the productivity of mining operations. In reality, UMMS is one of the Multiple Criteria Decision Making (MCDM) problems and decision makers have some difficulties in making the right decision in the multiple criteria environment. In this paper, a decision support system for underground mining method selection (UMMS-DSS) has been designed and developed in order to take into account the whole related problem criteria, research the entire effects of different scenarios of all available criteria and carry out sensitive analysis when needed. UMMS-DSS was designed as to use Analytic Hierarchy Process (AHP), one of the MCDM methods, to manage those tasks and to produce acceptable solution alternatives.

A* algorithm, a kind of informed search, is widely used for finding an optimal car route, because the location of starting and ending point are known beforehand. Unidirectional A* algorithm guarantees an optimal route but requires considerable search time. On the other hand, bidirectional A* algorithm, usually known faster than unidirectional A*, does not guarantee the route found to be optimal, if the search ends when the forward and backward search meet in the middle. It may even take longer than unidirectional search to find an optimal route. In this paper, a new modified bidirectional A* algorithm which takes less search time and guarantees an optimal route is proposed. To evaluate the efficiency of the proposed algorithm, several experiments are conducted in real urban road environment and the results show that the algorithm is very effective in terms of finding an optimal route and search time.

Combinatorial auction is one of the most popular market mechanisms and it has a huge effect on electronic markets and political strategies. On large scale e-markets, we need a good approximation algorithm for winner determination that is robust for changing the distribution and the number of bids in an auction. We proposed approximate algorithms for combinatorial auctions with massively large number of (more than 100,000) bids. In this paper, we show the robustness of our winner determination algorithms for combinatorial auctions with large number of bids. Experimental results demonstrate that our proposed algorithms are robust on changing the distribution and the number of bids in an auction. Finally, we shortly describe a theoretical limitation about our algorithms that concerns with giving truthfulness of the auction mechanism.

Although most real manufacturing systems have dynamic job shop structures, there is no general analytic method that has been found for analyzing them yet and computer simulation is still an outstanding tool. One of the most difficult problems in a dynamic job shop environments is to assign the optimal due dates. Due date assignment is an important task in shop-floor control, affecting both timely delivery and customer satisfaction. The ability to meet the due dates, however, is dependent not only on reasonableness of the due dates but also on the scheduling or dispatching procedures. In this paper, an agent based dynamic job shop simulation system is designed and developed to help the decision makers who have to mainly solve the problems of selecting correct due date assignment models and dispatching rules depending on selected performance criteria in their multi machine dynamic stochastic job shop environment.

Assembly design decision making is to provide a solution of currently violating design by evaluating assembly design alternatives with the consideration of the assembly design decision (ADD) criteria and of the causal interactions with manufacturing-environmental factors. Even though existing assembly design support systems have a systematic mechanism for determining the decision-criterion weight, the system still has a limitation to capture the interactions between manufacturing-environmental factors and ADD criteria. Thus, we introduce in this paper, Bayesian belief networks (BBN) for the representation and reasoning of the manufacturing-environmental knowledge. BBN has a sound mathematical foundation and reasoning capability. It also has an efficient evidence propagation mechanism and a proven track record in industry-scale applications. However, it is less friendly and flexible, when used for knowledge acquisition. In this paper, we propose a methodology for the indirect knowledge acquisition, using fuzzy cognitive maps, and for the conversion of the representation into BBN.

An adaptation of independent component analysis (ICA) and missing feature theory (MFT)-based ASR for two simultaneous continuous speech recognition is described. We have reported on the utility of a system with isolated word recognition, but the performance of the MFT-based ASR is affected by the configuration, such as an acoustic model. The system needs to be evaluated under a more general condition. It first separates the sound sources using ICA. Then, spectral distortion in the separated sounds is estimated to generate missing feature masks (MFMs). Finally, the separated sounds are recognized by MFT-based ASR. We estimate spectral distortion in the temporal-frequency domain in terms of feature vectors, and we generate MFMs. We tested an isolated word and the continuous speech recognition with a cepstral and spectral feature. The resulting system outperformed the baseline robot audition system by 13 and 6 points respectively on the spectral features.

The cell is an entity composed of several thousand types of interacting proteins. Our goal is to comprehend the biological system using only the revelent information which means that we will be able to reduce or to indicate the main metabolites necessary to measure. In this paper, it is shown how the Artificial Intelligence description method functioning on the basis of Inductive Logic Programming can be used successfully to describe essential aspects of cellular regulation. The results obtained shows that the ILP tool CF-induction discovers the activities of enzymes on glycolyse metabolic pathway when only partial information about it has been used. This procedure is based on the filtering of the high processes to reduce the space search.

This paper addresses the method of environment recognition specialized for biped walking robot. Biped walking robot should have the ability to autonomously recognize its surrounding environment and make right decisions in corresponding to its situation. In the realization of the vision system for biped walking robot, two algorithms have been largely suggested, they are; object detection system with unknown objects, and obstacle recognition system. By using the techniques mentioned above, a biped walking robot becomes to be available to autonomously move and execute various user-assigned tasks in an unknown environment. From the results of experiments, the proposed environment recognition system can be said highly available to be applied to biped walking robot walking and operated in the real world.