Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.



Published on

  • Be the first to comment

  • Be the first to like this


  1. 1. Ansari Muqueet Husain, Shaikh Mohammad Sohail, V. S. Narwane / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 5, September- October 2012, pp.1698-1701 Path planning of material handling robot using Ant Colony Optimization (ACO) techniqueAnsari Muqueet Husain*, Shaikh Mohammad Sohail**, V. S. Narwane*** *(Department of mechanical ENGG, Mumbai University, mumbai-08) ** (Department of Information Technology, Mumbai University, mumbai - 08) *** (Department of mechanical ENGG, Mumbai University, mumbai-77)ABSTRACT The present work utilizes Ant Colony introduced by Marco Dorigo in 1992 in his PhDOptimization (ACO) technique for the generation thesis [2]. Several ant species are able to select theof optimal motion planning sequence. The present shortest path, among a set of alternative paths, fromalgorithm is based on ants behavior, pheromone their nest to a food source. Ants lay a chemical trailupdate & pheromone evaporation and is used to called pheromone when they walk to attract otherenhance the local search. This procedure is ants to take the path that has the most pheromone;applied for proposed a method for path planning this mechanism manifests an effect of positiveof mobile robot motion in warehouses for feedback. ACO algorithm is a stochastic, distributedmaterials handling with starting from any and collective approach that has been used to solvelocation to reach a certain goal. Our technique as different hard combinatorial optimization problemsbased on the well-known environment of the such as the Traveling Salesman Problem (TSP),warehouse. To validate the proposed algorithm, Quadratic Assignment, or the Vehicle Routingthe program has been developed in Visual Problem. The first ACO algorithm was the AntC++.This technique can generated feasible, stable System (AS), which was designed to solve the TSPand optimal robotic materials handling sequence In this paper, we propose a heuristic evolutionaland then path sequence can satisfying the algorithm ARPP which is based on ACS frameworkmaterials handling constraints with minimum and applies visibility graph for mobile robot globaltravel time. The solution is either optimal or near path planning.optimal. The rest of paper is organized as follows. In section 2 we introduce path planning ofKeywords - ant colony system (ACS), Obstacle warehouse robot. In section 3 we give problemavoidance, Ant based Robot Path Planning representation including the visibility graph model(ARPP), pheromone, visibility graph. and ARPP algorithm in section4 its simulation results, then the conclusion of properties of ARPPI. INTRODUCTION are given. Mobile robot path planning or navigation isan important application for robot control systems II. PATH PLANNINGand has attracted remarkable attention from many 2.1THE WAREHOUSE ROBOTresearchers. As a result, many interesting research The complete system in the warehouseresults have been obtained. A difficult issue in robot robots consists of a single mobile robot withnavigation or path planning in an unknown manipulator with grippers for handling, and pickingenvironment with static or dynamic obstacles is to the materials from the racks or shelves. A fewfind a globally optimal path from the start to the environment sensors, such as wall markers (bartarget point and at the same time avoid collisions. codes, laser reflectors, etc.) that would help to[1] localize the robot with the odometry system existingSwarm Intelligence (SI) is an Artificial Intelligence in the mobile robot platform. The significanttechnique involving a new computational and features of the warehouse robot are:behavioral paradigm for solving distributed 1- Obstacle avoidance to move safely in theproblems based on self-organization. Examples of like this can be found in natural systems 2- Object and pattern recognition, to identifyconsisting of many individuals, such as ant colonies specified packages and locate them in the rightand flocks of birds, animal herding, honey bees, shelves.bacteria, and many more. Swarm-like algorithms, 3- Path optimization, to minimize time and distancesuch as Particle Swarm Optimization (PSO) and Ant during goods transportation.Colony Optimization (ACO), have appliedsuccessfully for real world Optimization problems 2.2 ENVIRONMENT MODELin engineering and telecommunication. Environment model is an important link of Ant Colony Optimization (ACO) is one of the robot path Planning. The essence of environmentthe prevailing algorithms in SI at present, which was model is according to the known environmental 1698 | P a g e
  2. 2. Ansari Muqueet Husain, Shaikh Mohammad Sohail, V. S. Narwane / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 5, September- October 2012, pp.1698-1701information, through extracting and analyzing algorithm, the initial search space of the artificialrelated characteristics, converts them into space that ants can be narrowed dramatically. Hence, in thisrobot can understand. Environment model selection paper, we use visibility graph as both the roadmapand specific methods to set up the path planning and construction graph to simplify the ARPPalgorithm related and rational selection of model problem, Fig.2 illustrates visibility graphmethod can reduce the troubles in the process of thepath searching. Visibility graph method is used inthis paper as a robots roadmap.III. PROBLEM REPRESENTATION In this section, we give the assumptions ofthe GPP problem. When we applied to anoptimization problem, the ACO usually involvessolution construction on a graph, hence, we presentvisibility graph model.3.1. Problem Assumptions The following assumptions need to bemade:1) We assume the environment is a two-dimensionalspace; 2) both the environment and obstacles have apolygonalshape; Fig no2.visibility graph3) in order to avoid a moving path too close to theobstacles, the boundaries of every obstacle can be 3.3 ARPP Algorithm for path planning (PP)expanded by an amount that is equal to half of the 3.3.1BACKGROUNDgreater size in the length and width of the robot’s Informally, the ACS works as follows: mbody plus the minimum measuring distance of the ants are initially positioned on n cities chosenrelevant sensors. In this case, the size of the robot according to some initialization rule (e.g.,can be ignored and we can see the robot as a point randomly). Each ant builds a tour (i.e., a feasiblerobot ℜ. Fig. 1 illustrates the moving environment solution) by repeatedly applying a stochastic greedyof ℜ, rule (the state transition rule). While constructing itsWhich is a 27 by 19 meters including six, obstacles? tour, an ant also modifies the amount of pheromone on the visited edges by applying the local updating rule. Once all ants have terminated their tour, the amount of pheromone on edges is modified again (by applying the global updating rule). As was the case in ant system, ants are guided, in building their tours, by both heuristic information (they prefer to choose short edges) and by pheromone information. An edge with a high amount of pheromone is a very desirable choice. The pheromone updating rules are designed so that they tend to give more pheromone to edges which should be visited by ants. 3.3.2 Active State Transition Rule In the ACS the state transition rule is asFigure 1. Moving Environment of ℜ follows: an ant positioned on node (i) chooses theFig.1. the (x, y) coordinates of vertices are (0, 0), city (j) to move by applying the rule given by (1)(07, 04), (11, 04), (11, 08), (07,08), (02, 11), (02,14), (07, 14), (07,11), (10, 11),(15, 11), (15, 14), 𝛼 𝛽 𝜏 𝑖𝑗 𝑡 𝜂 𝑖𝑗 𝑡(10, 14), (15, 03), (15,06), (20,06), (20,03), (24,07), , 𝑖𝑓𝑗𝜖𝑁 𝑖 𝑘 𝑡 ,(27,07), (27,12), (24,12), (20,14), (20,18), (25,18), 𝑃𝑖𝑗𝑘 𝑡 = 𝛽 𝑢 𝜖𝑁 𝑖 𝑘 𝑡 𝜏 𝑖𝑢 𝑡 𝜂 𝑖𝑢 𝑡 𝛼 (1)(25,14), and (27,19) respectively. The (x,y) 𝑘 0, 𝑖𝑓𝑗 ∉ 𝑁 𝑖 𝑡 ,coordinates of the starting point s (st) and goal point Where Nik is the set of nodes not yet visited by ant k(fn) are (0,0) and (27,19 ), respectively and connected to node i, τij represents the3.2. Visibility Graph Model pheromone for the link between nodes i and j, and α We consider visibility graph as the and β are the positive constant parameters used toroadmap of path planning problem. If we allowvisibility graph to be construction graph of the 1699 | P a g e
  3. 3. Ansari Muqueet Husain, Shaikh Mohammad Sohail, V. S. Narwane / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 5, September- October 2012, pp.1698-1701amplify the relative importance of the pheromone Where ξ is the local evaporation rate and τ0versus the heuristic function ηij given by (2) is the initial amount of pheromone. Actually, Local 1 Updating Rule is a kind of negative feedback whichηij = d (2) ij aims to encourage the exploration of new areas of Where dij is the distance of the link the search space. By evaporating pheromone trailsbetween node i and j Once all the ants have from visited arcs, the exploration of alternativeconstructed a path from the source node to the paths is increased. On the contrary, pheromonedestination node, each ant deposits pheromone in global updating encourages the exploitation ofthe amount of Δτk (t) at the links of the path using previously good paths. It is the common efforts of ijthe pheromone updating rule (3). the positive feedback and negative feedback that nk make the whole process of path optimization evolve. τij (t+1)=1(−ρo). τij (t) + k=1 Δτk t , (3) ij Where ρ0 is the evaporation rate parameter IV. Performance Analysis of ARPPdistributed in (0, 1), nk is the number of ants, and In order to examine the performance of ourΔτk t is the amount of pheromone deposited by ant ij proposed ARPP algorithm, it constructs an artificialk at link (i, j) at time step t, as given by (4) ant colony system Ω by using m ants. Simulation Q 𝑖𝑓 𝑎𝑛𝑡 𝑘 𝑢𝑠𝑒𝑑 𝑙𝑖𝑛𝑘 i, j , experiments executed up to 100 trials for three times Δτk = L k , ij (4) on the visibility graph which have been given in 0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒, Fig.2. The parameters we set as Table 1 shown: Where Q is a constant (normally set to one)and Lk, is the total distance of the path constructed Parameter α β ξ ρ ρ0 mby ant k. The main characteristic of the pheromone Parameter 0.15 2.0 0.15 0.2 0.8 8updating rule in AS is that the pheromone values are Valueupdated by all the ants involved in constructing apath from the source node to the destination node. Table no 1: Initial Parameters in ARPP 4.1 Experimental results on environment. The following graphs show the number of iteration3.3.3Global Updating Rule The ARPP has a pheromone trail updating V/S average distance travel by the 8 ants.approach that exploits the best solutions found. Bothevaporation and new pheromone deposit appliedonly to the arcs contained in the current bestsolution. Actually, the deposit of pheromone trail isa kind of positive feedback to reinforce the ant’sability in finding the good solutions. Through thepositive feedback, pheromone trails accumulate ongood paths, which can guide the whole ant colony intheir moving to search good paths directly.Ultimately, pheromone trails steer the ant colonysystem to evolve towards acquiring the optimalsolution. Global Updating Rule is given by (5) Graph no 1: number of iteration V/S averageτij ←(1−ρ )τij + ρ Δτijbs, ∀lij∈Tbs (5) distance travel by the 8 ants for first 100 trials. Where Δτijbs=1/Cbs is additionalpheromone, added to the edge lij that belongs tobest-so-far solution Tbs .ρ (with 0 < ρ < 1) is theglobal evaporation rate , for avoiding the excessiveincrease in pheromone.3.3.4Local Updating Rule The ARPP includes a local pheromoneupdate to reduce emphasis on exploitation ofexisting solutions, which is applied on only theedges that have been visited by ants. Immediatelyafter an ant adds an arc to its current path theamount of pheromone on the arc is decreased .This Graph no 2: number of iteration V/S averageis called Local Updating Rule and given by (6) distance travel by the 8 ants for second 100 trials.τ ij ←(1−ξ)τij +ξτ0 (6) 1700 | P a g e
  4. 4. Ansari Muqueet Husain, Shaikh Mohammad Sohail, V. S. Narwane / International Journal of Engineering Research and Applications (IJERA) ISSN: 2248-9622 Vol. 2, Issue 5, September- October 2012, pp.1698-1701 avoid human error or economically not viable. In this paper, ACS is used to find the shortest navigational path of an autonomous mobile robot avoiding obstacles to reach the target station from the source station. The output is found to be optimal and satisfying for the given problem. In future with increase in complexity of the problem, i.e. with increase in number of obstacles or in dynamic environment, the Ant Colony Optimization Algorithm can be applied effectively giving optimalGraph no 3: number of iteration V/S average path in lesser timedistance travel by the 8 ants for third 100 trials. By observing the graphs The global-bestpath of the minimum distance travels by the ants ACKNOWLEDGEMENTS The authors gratefully acknowledge andappears in the path shown in Table2 with length of thank to almighty (Allah) who created the beautiful33 also Statistic results are shown in Table2 ant system to find their optimal path from their nest First 100 Second 100 third 100 to food. And by observing this today we are able to trials best trials best trials best apply this logic to various optimal path findingpath 0→2→24→ 0→2→24→ 0→2→24→ areas. 25 25 25Path 33 33 33 REFERENCESvalu [1] Buniyamin N., Sariff N., Wan Ngahe W.A.J., Mohamad Z. Robot global pathTable no 2 Statistic Results of Performance of planning overview and a variation of antARPP colony system algorithm. international From the table no 2 we can say that the journal of mathematics and computers inoptimal path for the robot is 0→2→24→25 which is simulation Issue 1, Volume 5, 2011,9-16indicated in the visibility graph by the red line as [2] Marco Dorigo, Senior Member, IEEE, andshown in the fig given below. Luca Maria Gambardella,Member,IEEE, Ant Colony System: A Cooperative Learning Approach to the Traveling Salesman Problem, IEEE transactions on evolutionary computation, vol. 1, no. 1, APRIL 1997,53-66 [3] Jing Zhou, Guan-Zhong Dai, De-Quan He, Jun Ma, Xiao-Yan Cai, Swarm Intelligence: Ant-based Robot Path Planning 2009 Fifth International Conference on Information Assurance and Security2009IEEE,.459-463 [4] QingyaoHan, Qiang Wang ,Xiaoguang Zhu Fig no3.visibility graph of the optimal path and Jin Xu Path Planning of Mobile Robot Based on Improved Ant ColonyAlgorithm,V. CONCLUSION 2009IEEE,531-533 The proposed algorithm apply visibility [5]graph as the roadmap and construction graph. ARPP DayalR.Parhi,JayantaKumarPothal,Mukeshalgorithm has the advantage of real-time KumrSingh, Navigation of Multiple Mobileoptimization; ACS for ARPP cases within a given Robots Usingmap of feasible nodes was proven able to find the SwarmIntelligence,2009IEEE,1145-1149optimal path effectively. The optimal path found [6] Jagadish Chandra Mohanta, Dayalsatisfied the optimization criteria for ARPP Ramakrushna Parhi, Path planning strategypurposes which are to reduce the path cost and for autonomous mobile robot navigationshorter computation time with smaller number of using Petri-GA optimization, Computersgenerations. and Electrical Engineering(2011),1-13. Finally we can conclude that to find the Book:navigational path of robots is currently among the [7] Marco Dorigo and Thomas Stützle, Antmost intensively studied and promising area of Colony Optimization(A Bradford Bookresearch which has a significant role in robotics. It The MIT Press Cambridge, Massachusettshas varied application in different field of works, London, England)especially where human presence is dangerous, to 1701 | P a g e