Target-seeking of method of AGV robot much representative direction of 4 old research

  • Time:
  • Click:99
  • source:KOMANG CNC Machining
Introduction: The issue that the article discussed to appear when popularizing MAPF method actual setting and the 4 research direction that solve these problems. What we emphasize is the importance that solves these problems, is not the method with the standard model rapidder development that is MAPF problem. Much representative method is searched (Multi - MAPF of ／ of Agent Path Finding) already was in science of artificial intelligence, robot, academic computer and real operation research get much research. The issue that the article discussed to appear when popularizing MAPF method actual setting and the 4 research direction that solve these problems. What we emphasize is the importance that solves these problems, is not the method with the standard model rapidder development that is MAPF problem. Method of 1 foreword much representative is searched (MAPF, also call much representative to find way) mix in science of artificial intelligence, robot, academic computer much research gets in handling research actually. (standard) the task of MAPF is to be many representatives (Agent) find in give a picture (Graph) in from its current acme (Vertices) the to its target method that does not produce collision with other representative, optimize cost function at the same time (Cost Function) . The method that existing MAPF uses includes: From can satisfy a gender to reduce a problem (Reductions To Problems From Satisfiability) , integral linear programming (Integer Linear Programming) , answer collect process designing (Answer Set Programming) [Yu And LaValle, 2013b; Erdem Et Al. , 2013; Surynek, 2015] , best ／ is finite second actor (Optimal, bounded - Suboptimal) or second actor searchs a method (Suboptimal Search Method) [Silver, 2005; Sturtevant And Buro, 2006; Ryan, 2008; Wang And Botea, 2008; Standley, 2010; Standley And Korf, 2011; Wang And Botea, 2011; Luna And Bekris, 2011; Sharon Et Al. , 2013; De Wilde Et Al. , 2013; Barer Et Al. , 2014; Goldenberg Et Al. , 2014; Wagner And Choset, 2015; Boyarski Et Al. , 2015; Sharon Et Al. , 2015] . All sorts of problems that when we considered to popularize MAPF actual setting recently, appear, include Kiva (Amazon Robotics) warehouse system [Wurman Et Al. , 2008] (graph 1) with automatic aircraft tractor [Morris Et Al. , 2016] . These problems can divide for two average issues: 1, the method with the standard model rapidder development that is MAPF problem is insufficient, because be below a lot of actual conditions, can use new structure or need new problem model. 2, optimizing MAPF or its new model the problem to undertake study as combination only is insufficient, the MAPF solution that arises because of place also needs to carry out. We discussed to settle the 4 research way of these two problems from different point of view: 1. In a lot of effective much representative systems, before finding optimal way for all representatives, the representative is differentiated first group (Team) , allocate specific target to every group next, every representative need is appointed from inside the group of the place a target. We had made combination for the representative of different group target allocation and method are searched (TAPF ／ Target Assignment And Path Finding) it is difficult that the issue will solve this. We still developed method of an optimal TAPF, it can expand a few groups act as agent with hundreds [Ma And Koenig, 2016] . 2. In a lot of effective much representative systems, the representative is faceless (exchangeable) , but their effective load dispute is faceless (not exchangeable) , and need be delivered given target. The representative can exchange his in such system normally effective load. Regard first time as the attempt, we designed wrap up to exchange robot road by (Package - PERR of ／ of Exchange Robot Routing) problem, in order to solve more generalization (load of effect of be patient of transfers) carriage problem [Ma Et Al. , 2016] . In this article, the difficult sex that we still proved to approximate and best MAPF is solved (complex degree) . 3. In a lot of effective much representative systems, the representative moves (Agent Motions) it is important that consistency and the result that the representative moves can forecast a gender (it is especially in the working space that by support of the people the representative shares) , but existing MAPF method did not consider this. We had divided a phase to explore the problem structure that gives MAPF example: In the first phase, we developed a kind of plan that seeks way to act as agent, included what offer by the user among them a lot of freeways that carry the margin (Highways) , the consistency that this plan reached a representative to move and can forecast a gender [Cohen Et Al. , 2015] ; In the 2nd phase, we developed the method that builds a highway automatically [Cohen Et Al. , 2016] . 4. The inspiration of MAPF basically originates the navigation of much robot system or athletic program module. However, the best sex of MAPF solution or finite the rash club sex that second actor sex does not mean them certainly, the program with the faulty robot in considering reality especially is carried out (Plan - Execution) ability. We developed a framework, it can be effective ground later period is handled (Postprocesses) the output of MAPF method, the program that is used at founding to be able to be carried out by effective much robot system carries out arrangement. Graph 1: (Zun Tu) drive automatically unit and can be driven the memory cabin of the memory product of unit shift (Storage Pod) ; (right graph) the layout of system of typical Kiva warehouse (Wurman Et Al. , 2008) to popularize MAPF method actual setting, what we reveal these to study way now is practical, the standard model method that solves the MAPF problem with these two problems and rapidder development in order to prove is euqally important (more important even) . The target of 2 representatives group allocates and method is searched (TAPF) combination generally speaking, it is every constituent that is in according to the representative matchs a target. The representative is differentiated first in the group, next every representative need is appointed from inside the group of the place a target, so that get a representative,from current acme to its the method of the target optimizes cost function. For example, in Kiva warehouse system, drive those who store cabin carries new memory position from storehouse unit (Drive Unit) can form a group, because each need in them is allocated,a practicable stores the position. The MAPF method previously assumes existence target distributes a program, make every representative is allocated beforehand a target, but to realize best sex, we built TAPF model, it is integrated target allocation and method seek an issue and defined a common cause for them. In TAPF, in acting as agent to be divided to each groups, the target amount of every group and this groups of medium representative amounts are same. The task of TAPF is give a representative target allocation, plan to act as agent to arrive from current acme the won't produce collision method of its target, make every representative apropos shift is in a target of the group to its, thereby all targets in the group are visited, and the greatest time that finish (the condition of the earliest time when had reached its goal when all representatives and stopping shift is chief) by the smallest change. Every representative in the group can be allocated the target of place team, and the representative in same group is exchangeable accordingly. However, the representative in different group is not exchangeable. TAPF can be regarded as (standard) the faceless and aberrant generalization of MAPF and MAPF: · comes from TAPF (standard) MAPF result, if every group is comprised by a representative only, criterion the amount that the amount of the group is equal to a representative. The target goes to acting allocation is predefine, acting dispute is accordingly faceless (not exchangeable) . If · has a group only (include all representatives) , criterion the faceless variable of MAPF (also call the MAPF with constant target) come from TAPF. The representative can be allocated any targets, because this is exchangeable. It can be used inside polynomial time be based on drifting MAPF method (Flow - Based MAPF Method) receive optimum solution [Yu And LaValle, 2013a; Turpin Et Al. , 2014] . Current the most advanced best TAPF method -- call those who be based on collision the smallest cost flows (Conflict - Based Min - Cost Flow) [Ma And Koenig, 2016] -- was united in wedlock search and be based on drifting MAPF method. It can popularize a few groups and hundreds representative. The wrap up of 3 MAPF exchanges robot road by (PERR) and new complex spending computation to act as agent as a result is faceless normally, but the effective load that carries end be allocationed (wrap up) , dispute is accordingly faceless. For example, in Kiva warehouse system, drive unit is faceless, but the memory cabin that they carry was allocated memory position, dispute is accordingly faceless. If every act as agent to carry a package, criterion this problem is equivalent to (standard) MAPF. Actually, wrap up can be delivered between the representative normally, this causes commonner carriage problem, for example, the passenger that contains change is shared by the car [Coltin And Veloso, 2014] send with the package that uses a robot in the office [Veloso Et Al. , 2015] . The first pace that we had regarded understanding as these problems PERR [Ma Et Al. , 2016] . In PERR, every representative is locomotive a wrap up, two any representatives in photograph adjacent acme can exchange his to lap, and every bag needs wrap should be sent give a cause. PERR can be regarded as accordingly (standard) of MAPF improve edition: The wrap up in · PERR can be regarded as to be in (standard) the oneself are mobile representative in MAPF. · allows two wrap up in photograph adjacent acme to exchange their acme in PERR, but two representatives in photograph adjacent acme do not allow to be in (standard) the acme that they exchange in MAPF. K - the generalization that PERR is PERR, lap among them the K that be become by cent type, and the wrap up of same kind is exchangeable. Because be in TAPF, the representative is divided in the group, and be the same as the representative in posse team be exchangeable, so K - PERR can be regarded as pair of K the correcting of the TAPF of the group, same principle, PERR can be regarded as (standard) the correcting of MAPF. We had proved approximate and optimal PERR and K - the difficult sex that PERR understands (to K ≥ 2) . An inference of our research is: 4 ／ are less than in any factor the 3 the greatest finishing time inside are the least change, approximate MAPF and TAPF are NP - of Hard, even if the TAPF that has two groups only. We still proved to add exchange to MAPF the operation won't be reasonable theory on reduce its complex degree, but make PERR easier than MAPF solve. The in different actual setting successive problem that produces from this: " a representative that has a lot of package " the rural postman problem with classic generation (Rural Postman Problem) ; " representative and wrap up are euqally much " generation MAPF, TAPF or PERR. Understand problem of these two extremes to conduce to us solving common problem, other of no less than the requirement of a lot of real world tasks is same. Graph 2: The high speed way that in storehouse of Kiva of a simulative the user in the system offers (Highway) of the structure of 4 exploration problem and motion can forecast sexual representative and person to share working space, of the consistency that they move and its motion result can forecast a gender the safety to the mankind is important, because this takes no account of existing MAPF method. This makes us exploration is given the problem structure of MAPF example, the brim that designs an incentive representative to be offerred along the user (Edge) gather (call a freeway) mobile plan [Cohen Et Al. , 2015] . We are expanding simply plan (Inflation Scheme) use below setting be based on experience to pursue (Experience Graph) freeway [Phillips Et Al. , 2012] idea, in order to derive new inspiration is worth (Heuristic Values) , this value uses incentive MAPF method to return the method that includes freeway margin, this kind of method can avoid to act as agent between collide head on (Head - To - Head Collisions) , the consistency that achieves its exercise and can forecast a gender. For example, in Kiva warehouse system, we can design a freeway along the narrow channel between memory position, if pursue,2 medium arrowhead are shown. We had proved in system of simulative Kiva warehouse, such freeway can quicken MAPF method significantly, what hold the MAPF solution cost of expectation at the same time is finite second actor sex. The problem structure of TAPF and PERR example also can use same method. In feasibility research, we still developed the builds road automatically means that provides highway photograph to rival with the user. It is OK that 5 solutions plan faultlessly to implement MAPF or TAPF method anything but the won't produce collision method that the second actor sex that is hundreds representative to find first-rate to perhaps be offerred in the user inside reasonable computational time assures to fall. They also can work normally in mixed and disorderly and compact environment even, be like Kiva warehouse system. However, the representative has faulty program to carry out ability normally, and cannot faultlessly synchronism their motion, this can bring about frequent new program to waste time. Accordingly, we offerred a framework, use a simple time network to effectively later period handles MAPF solution and found a program to carry out arrangement, this is applied to be not complete robot (Non - Holonomic Robot) , the biggest translation that considers them is spent with rotate speed coming back, offerred safety is apart from between a robot and flabby border (the definition is newest the difference with the place that enters time the earliest) assure, the new program with the concentrated time that faulty program is carried out and avoids in order to alleviate to fall in a lot of circumstances [Et Al of Honig ¨ . , 2016] . This frame has been in emulate and get assessment in true robot. TAPF and PERR method also can be in same framework application. The safe space that the problem that should solve in prospective job includes to increase an user to offer, additional motion obligation, uncertainty plans and plan afresh. We discussed 6 conclusion 4 research direction, should popularize MAPF method in order to solve actual setting to counteract exploration problem structure or the problem that when having MAPF method, appear. Our target is to be the researcher that works in MAPF domain to point out interesting research direction. CNC Milling CNC Machining