فهرست مطالب الیپس مسیحی

Nowadays, the diversity and expansions of problems in various branches of science has greatly increased and finding the answer to such problems in a short period of time is considered as a very essential challenge. Using artificial intelligence can significantly accelerate the solving process for complex problems and considerably reduce the response time. The pursuit-evasion problem is one of the problems that can have a high level of complexity due to the nature of the factors involved. Factors causing complexity include the number of factors involved, the range of member’s vision and barriers exist on the playground. Various algorithms have been proposed to solve the pursuit-evasion problem so far, each has its own strengths and weaknesses. In this paper, the pursuit-evasion game is examined by using the linguistic geometry algorithm specifically in a problem with 9×9 playground dimensions and to examine the generalization of the algorithm's performance in problems with different dimensions. It has been shown that this approach can improve the response speed by more than 90%. In this work, the factors affecting the realism of the pursuit-evasion game are examined more carefully in linguistic geometry and finally the problem is solved by simplifying the problem space to a number of subspaces in which the movement paths of each element of the game are clear. We show that, despite the problem space became more complex, the linguistic geometry algorithm creates about 91% improvement over the other algorithms.

Roads are one of the fundamental infrastructures for nations’ development, and their maintenance is of utmost importance. However, roads are subject to gradual deterioration due to vehicles’ continual run, climate change, and miscellaneous damages. Hence, road management centers in various countries, design and analyze a range of maintenance policies depending on road conditions through continuous monitoring. While deciding the time and type of road maintenance, has been traditionally done selectively by experts, regarding the large number of fragments in road networks, repeated and non-algorithmic nature of the decision making process, as well as the need for high precision to avoid over-budgeting, this task should be performed preferably by means of decision support systems. In order to determine appropriate actions for maintaining road fragments, pavement assessment indices must be measured at first, and then the right policy for maintaining each fragment or the whole road network must be planned based on the estimated maintenance costs and the allocated budget. In this paper, a fuzzy expert system is developed as a decision support system to assist road maintenance managers in their decision process by enhancing the speed and precision of policy making.

In this paper, we deal with a procurement problem in a decentralized two-echelon supply chain, in which a buyer (manufacturer) aims to procure a bundle of needed items from a number of suppliers. The problem is modeled via a bi-level programming model, in which the buyer acts as a leader and the suppliers separately act as followers on lower level. To solve this bi-level mathematical model, a hybrid algorithm based on particle swarm optimization (PSO-A*) is proposed. The proposed mechanism, by satisfying the partner's constraints, is able to reach a near-optimal solution which persuades the partners to contract. In this paper, a comprehensive pattern is proposed for embedding the negotiation process in mathematical models and their solution procedure. This study aims at developing a bi-level programming to deal with a negotiation-based procurement problem, according to the realistic assumptions, in which the buyer is considered as a leader and makes optimal decisions according to supplier's proposals in lower level as followers. Such a mechanism provides an alignment among supplier's production planning and order allocation to avoid instantaneous orders, inability of suppliers to supply orders, and impose high inventory cost. In addition, it supports the partnership with valued suppliers through suitable order allocation by taking supplier's capacities into consideration. This research has been done based on the assumptions derived from the interviews with the experts in supplying automotive parts company called SAPCO and a number of its partners. To evaluate the performance of the proposed algorithm, the results of the PSO-A* algorithm are compared with those of PSO-Exact and PSO-Greedy algorithms. Based on computational analysis, it can be observed that the PSO-A* algorithm is more efficient compared to the PSO algorithm in which its lower level sub problems are solved through an exact solver; it is also more effective compared to the PSO-Greedy algorithm.

Multi-robot systems are preferable for tasks that are inherently distributed in space, time, or functionality. For the problems that can be decomposed into independent subproblems, using a multi-robot system offers a potential for reducing the overall task completion time. For effective employment of multi robot systems, it is necessary to properly implement Task Allocation, which is an NP-hard problem. In this paper, a two-layer architecture for exploring and covering an unknown environment by multiple heterogeneous robots is developed. At the first layer of the architecture, the Multi-SRT algorithm is developed for exploration and covering of the environment and the Multi-Tangent-Bug is used for online path planning and obstacle avoidance in a distributed manner. In the second layer, by means of a centralized approach, a Fast Genetic Algorithm (FGA) is proposed for solving the multi-robot task allocation problem. Performing each task enhances the utility of the system, and completing all tasks is the ultimate goal of the system. For evaluating the efficiency of the FGA, a number of scenarios were run and the results were compared to NSGA-II algorithm. Simulation results showed the reliability of the developed architecture at the first layer and the precision and quality of the task allocation at the second layer.

Warehouses and distribution centers are essential components in supply chain and their management has a particular importance. In the traditional approach for collecting the items of orders in warehouses, operators walk or drive toward the shelves and collect the ordered items. However, since 2006 a new system has been deployed in some large distributing warehouses like Amazon Inc., in which shelves are mounted on mobile platforms and are carried by small mobile robots toward operators who pick the ordered items. Advantages of this system compared to traditional system are increased flexibility, accuracy, and speed of preparing the received orders. On the other hand, the mathematical model of this system –which we call it ‘Automatic Reverse Warehousing System (ARWS)’– is introduces as a trade solution, and no research papers have been published about it. In this paper, this system will be studied from the viewpoint of industrial engineering. Then, its components and their relationship with each other and their two major subproblems, namely, allocation and routing will be identified, and their interrelations will be investigated. The model is solved for minimizing the overall cost and finding the best paths of shelves through a Genetic Algorithm and maximum flow approach.

The goal of a query-by-example music information retrieval system is retrieval of the target song corresponding to user-provided example from a particular dataset. The example can be a few second piece recorded from any music source such as TV or even a noisy environment e.g. gym. In this paper, a query-by-example system for music retrieval using genre recognition is proposed whose goal is to show the effect of genre recognition to achieve the accurate and rapid performance in such systems even in presence the background noise. This system includes two basic blocks: genre recognition and matching-retrieval. A binary decision tree performs the genre recognition and matching-retrieval uses two Euclidean and Kullback-Leibler (KL) distances along with a score level based decision fusion. The proposed system is evaluated on the well-known GTZAN dataset (prepared by George Tzanetakis) and by two random groups of pure and noisy queries. The results show the accuracy of 97% and 86% for two pure and noisy query groups, respectively, in retrieval time of 525 ms with Euclidean distance. These values are 97% and 82% in retrieval time of 380 ms with KL distance.

This paper addresses the cooperative object transportation by a multi robot distributed system, which is a difficult problem due to path planning and robot cooperation challenges. In this problem, a number of robots should transport an object to a goal point safely while avoiding obstacles and utilizing a proper coordination and cooperation mechanism. The proposed method has a two-layer structure which benefits from both centralized and decentralized architectures. The global level takes advantage of full knowledge of environment to plan an optimal path using the new Optimally-Connect Random Tree (ORT) method, and the local level performs some local processes to reduce the system’s overall processing load and cost and increase its robustness. The required coordination between the robots is realized via radio communication, and for local path planning of the robots a combination of potential fields and TangentBug algorithms has been used. The proposed method has been implemented on multiple KUKA youBot mobile manipulators in the Webots simulation software, and its performance has been evaluated through various experimentations and the results of implementing and comparing the ORT and Rapidly-exploring Random Trees (RRT) showed the advantage of the proposed method.

In the rail fleet Sizing problem, it is aimed to determine the optimal number of each vehicle type in transportation fleet, while optimizing system objectives. This issue has been investigated as a single-objective problem in the context of rail transportation with the assumption of homogeneity in the fleet, despite the existence of different goals in real systems. In this paper, a dual objective function is adopted after studying real-world consideration. According to the experts at the IslamicRepublic of Iran Railway Company (IRIR), reduction of the number of delays in response to requests during the course of planning is an important additional objective function, which is included in our model. The solved problem is dynamic and demands for wagons and the transportation times are assumed to be deterministic. In this paper, after presenting the mathematical model, the importance factor of each objective function is calculated taking into account theprescriptions of the experts at the IRIR Research Center and based on aggregate weighting method. In order to solve the model and find the solutions on the Pareto front, three methods based on Genetic Algorithms, Simulated Annealing, and a hybrid of those are designed and their parameters tuned, by which the rail transportation system of the IRIR is solved and the results are discussed.

The problem of navigation among movable obstacles (NAMO) is to find a collision free path for a robot while the robot is able to manipulate and transfer some objects (if possible or needed) to clear its path toward the goal. NAMO is a NP-complete problem and is in a class of motion planning problems that have dynamic environments. In this domain, an optimal plan for the robot can be defined with respect to many different criteria such as length of the transit and transfer paths, number of manipulated obstacles, total number of displacements of all the objects and time. In this article, we have tried to design an algorithm capable of solving a wide variety of NAMO problems, using concepts like, visibility graph and penetration depth. Using the recursive function to solve some existing problems in the literature shows significant reduction in number of transferred movable obstacles as well as total number of displacements of all the objects.

Production planning problems are generally modeled as mixed integer programming problems; and solved through heuristic algorithms، because of their innate high computational complexity. In this paper، a mixed integer nonlinear programming (MINLP) model is designed for multi-item، multi-period production planning to replenish orders of the buyer and minimizing the supplier’s costs. It is supposed that the order quantity is constant، and ordering occurs at once. This model has been developed according to the realistic assumptions of SAPCO Company، which is a major supplier of automotive parts in Iran، and one of its partner companies. In addition، an efficient heuristic algorithm based on A* search has been proposed to solve this mathematical model. The proposed search algorithm does not need an initial solution; also، it can overcome the memory overhead through bounding the stored states in its open-list. Actually، in spite of the simplicity of the proposed algorithm، which is established based on the simple inventory management equations; it is able to generate efficiently optimal or near-optimal solutions in comparison with an exact solution method، a greedy search algorithm، and simulated annealing algorithm as a metaheuristic algorithm.

This paper deals with motion planning of Tractor-trailer robots، which are car-like robot dragging several trailers with no driving force. Each trailer has a nonholonomic kinematic constraint which increases the complexity of the path planning problem. We solved this problem by implementing the Equivalent Size concept، which depending on the size، number، and link-point positions of trailers، transforms a tractor-trailer path planning problem into a single car-like robot path planning problem. In this paper a new path planning algorithm is proposed for car-like robots which utilizes the Fast Marching Method (FMM)، which is a numerical method for solving the Eikonal differential equation، and the concept of Virtual Obstacles. The algorithm is fast، works independent of the shape of obstacles، and is easy to implement. To evaluate the quality of the solutions the algorithm is compared with the grid search and nonholonomic RRT algorithms. The results showed that the new method has by far lower runtime compared to the other algorithms، while producing short and smooth paths.

Dynamic facility layout problem (DFLP) is locating departments in different time zones where the location of each department may be varied at the beginning of each period. Traditionally, this problem is constructed based on the minimization of total handling cost, while more objectives could be considered in designing an efficient layout, like activity relationship, department closeness and etc. In this article DFLP is considered by two minimization of the total handling cost and maximization of total activity relationships. This problem is then solved by a new hybrid heuristic algorithm using local search supplemented by simulated annealing. This algorithm is so powerful to generate pareto solutions in such a way that both of convergence and diversity properties are evident.

In this paper a new method is proposed for path planning of planar manipulators amid obstacles through mathematical programming in a way that the robot’s links avoid collision with obstacles throughout their motion from an initial to a goal configuration. After inputting the workspace geometry، the shortest feasible path for the robot’s end-effector is planned toward its goal position using Generalized Visibility Graph، which is then interpolated into subgoal points that should be sequentially reached by the end-effector. Next، a Mixed Integer Nonlinear Programming (MINLP) model with the objective of minimizing the distance between the end-effector and the subgoals is successively solved and the angle of each link is determined such that it does not intersect obstacles. In order to enhance the safety of clearance from obstacles، they are enlarged by an offset. Also، the proposed method has been modified and tuned aiming to reduce the number of constraints and 0-1 variables، which led to reduced runtimes.

In parallel with advances in robots hardware, the motion planning science has become increasingly indispensable and complicated, and enables robots to enhance their capabilities in applications where the presence of humans is impractical or hazardous. The covering and minesweeping operations of outdoor areas are examples of such applications. This paper deals with the problem of covering and minesweeping of unknown environments by multiple mobile robots which have cooperative interactions. For searching the space, the new method of multi sensor-based random trees (MSRT) is developed, and for planning the robots’ paths toward the scattered items and their collection, a mathematical model similar to the multiple traveling salesman problem is proposed, which together with the k-means clustering, the A* search, and the visibility graph techniques plans the route of each robot. Comparisons with the counterpart mathematic model solved to optimality showed that the proposed method produces suboptimal solutions in much shorter computational times.

Statistical Process Control (SPC) charts play a major role in quality control systems, and their correct interpretation leads to discovering probable irregularities and errors of the production system. In this regard, various artificial neural networks have been developed to identify mainly singular patterns of SPC charts, while having drawbacks in handling multiple concurrent patterns. In this paper, a new compound method is proposed for automatic identification of concurrent patterns of SPC charts. First, by using a wavelet transform the complex concurrent pattern is decomposed into its underlying singular patterns, and then by implementing the Principal Component Analysis (PCA) jointly with a Probabilistic Neural Network (PNN), pattern types are identified. The results obtained from simulated data exhibited a success rate of 94.83% in recognizing concurrent multiple patterns.