Presentation is loading. Please wait.

Presentation is loading. Please wait.

1 Robust conflict-free routing of bi- directional Automated Guided Vehicles (AGVs) Institut de Recherche en Communication et Cybernétique de Nantes Samia.

Similar presentations


Presentation on theme: "1 Robust conflict-free routing of bi- directional Automated Guided Vehicles (AGVs) Institut de Recherche en Communication et Cybernétique de Nantes Samia."— Presentation transcript:

1

2 1 Robust conflict-free routing of bi- directional Automated Guided Vehicles (AGVs) Institut de Recherche en Communication et Cybernétique de Nantes Samia MAZA Pierre Castagna

3 2 Plan : Introduction to the AGV routing problem Classification of the AGVs routing methods The conflict-free shortest time path planning The robust conflict-free routing (2 algorithms) Some results & Conclusion

4 3 Definitions: Automated guided vehicles (AGVs) are used to transport materials and goods in manufacturing systems. They follow guidance circuits connecting various workstations in the warehouse. The guidance circuit is a physical track, which can be materialized with different manners, such as a colored bandage stuck on the ground, or an electrical conductor buried in the ground.

5 4 Kinds of guidance networks unidirectional Circuits A D C B E (8) (4) (5) (6) (1) (3)(2) (7) Bi-directional Circuit

6 5 The advantages of Bi-directionnal Circuits Reduction of the total traveled distances Reduction of flow times Reduction of the space requirement Best network reachability More complex control due to the conflicts between AGVs Egbelu et al, potentials for bi-directional guide-path for AGV based systems, 1986. nm V1V1 V2V2 collision

7 6 Classification of the AGVs routing methods Not robust. The Predictive methods * Find optimal routes for vehicles; off-line conflicts Prediction; Planning of the AGVs path. Good performances in the theory. H.Thomas, Optimisation des trajectoires dune flotte de chariots mobiles, Thèse de Doctorat, Nantes 1994. N.N.Krishnamurthy et al, Developing conflict-free routes for automated guided vehicles, 1993 Tanchoco et al, Conflict-free-shortest-time bidirectionnal AGV routeing, 1991 Tanchoco et al, Operational control of bidirectional automated guided vehicle system, 1993 *

8 7 Classification of the AGVs routing methods The performances are not optimized a priori. The reactive methods * The AGVs path is not planned; The decisions are taken in a real time manner Robust Methods Ying-Chin Ho, A dynamic zone strategy for vehicle collision prevention and load balancing in an AGV system with a single loop guide path, 2000. Spyros Reveliotis, Conflict resolution in AGV system, 2000. Qiu Ling & Hsu Wen –Jing, Conflict free AGV routing in a bidirectionnal path layout, 2001. *

9 8 Our objective Make one predictive control method more reactive to real time changes

10 9 The conflict-free shortest time AGV path planning Tanchoco et al, Conflict-free-shortest-time bidirectionnal AGV routeing, 1991 Tanchoco et al, Operational control of bidirectional automated guided vehicle system, 1993 The description of the method 10 2 1 5 4 3 6 (9)(10) (2) (3) (4) (7) (6) (5) (8) (1) (11) 7 9 8 1 3 2

11 10 f 6 1 6 f51f51 5 f41f41 4 f31f31 3 f 2 0 = r 2 1 2 f11f11 1 f52f52 f42f42 f43f43 f32f32 r51r51 r42r42 r41r41 r31r31 r11r11 f12f12 V1 V2 V3 0 10 20 30 40 50 60 70 Time Nodes The nodes reservation table A free time window A reserved time window 10 2 1 5 4 3 6 7 9 8

12 11 f 6 1 6 f51f51 5 f41f41 4 f31f31 3 f 2 0 = r 2 1 2 f11f11 1 f52f52 f42f42 f43f43 f32f32 r51r51 r42r42 r41r41 r31r31 r11r11 f12f12 V1 V2 V3 0 10 20 30 40 50 60 70 Time Nodes Principle of the method f12f12 f32f32 f43f43 f52f52 f61f61 f51f51 f41f41 f42f42 f31f31 f20f20 f11f11 Remark : A mission can appear to be impossible if such a path doesnt exist 10 2 1 5 4 3 6 7 9 8

13 12 cfstp The routing of the AGV V 3 by the cfstp algorithm f53f53 f 6 1 6 f51f51 5 f41f41 4 f31f31 3 r21r21 2 f11f11 1 f52f52 f42f42 f43f43 f32f32 r51r51 r43r43 r41r41 r31r31 r12r12 f13f13 V1 0 10 20 30 40 50 60 70 Time The time windows after the routing of V3 Nodes r11r11 r42r42 r44r44 f32f32 f12f12 r52r52 r61r61 V2 V3

14 13 The schedule of a new displacement Each AGV has an ordered list of missions A mission consists in going to visit a node N The guide path contains garages, their number is at least equal to AGVs fleet size. The garages nodes can not be destination nodes of the AGVs. The missions order can not be inverted A new mission of a vehicle is planned only if this one becomes free Assumptions MAZA & Castagna, Routage robuste sans conflits de chariots autoguidés bidirectionnels, CIFA 2002

15 14 Drawback of this predictive method This method is effective. It gives an optimal conflict free path by considering the previously established plans Open loop control method not robust: The disturbances * which can appear in a real system are not taken into account * ex: an accident, a slowing down in front of obstacles…etc Introduction of a shift between the predicted time windows and the realized one

16 15 cfstp The routing of the AGV V 3 by the cfstp algorithm f53f53 f 6 1 6 f51f51 5 f41f41 4 f31f31 3 r21r21 2 f11f11 1 f52f52 f42f42 f43f43 f32f32 r51r51 r43r43 r41r41 r31r31 r12r12 f13f13 V1 0 10 20 30 40 50 60 70 Time The time windows after the routing of V3 Nodes r11r11 r42r42 r44r44 f32f32 f12f12 r52r52 r61r61 V2 V3 r42r42 Conflict

17 16 Conclusion This method can not be applied directly on a real system.

18 17 The conflict free shortest time procedure (CFSTP) Predictive level Nodes crossing order controller Real time control level O i = An ordered list of AGVs having to cross the node i Collision avoidance in Real time Maza & Castagna, Conflict-free AGV Routing in Bi-directional Network, ETFA 2001

19 18 Real time collision avoidance The conflict free shortest time procedure (cfstp) Task for checking the nodes crossing order of AGV V 1 Task for checking the nodes crossing order of AGV V i Task for checking the nodes crossing order of AGV V n The central controller (predictive level) Decentrali- zed controllers (real time level)

20 19 4 V2 V3 V1 Arrival of the vehicle V x to the node n Is the vehicle V x the first vehicle in the List O n ? V x must wait for the crossing of another vehicle V x can cross n Yes No End V2 V3 V1 V3 O4=O4=

21 20 Consequence Consequence A robust closed loop control: the system state is taken into account at any moment, and the conflicts can be avoided in a real time only by respecting the established crossing order Forgetting time, the realized system behavior is as predicted in the planning level

22 21 Criticism of the method Criticism of the method MAZA & Castagna, Routage robuste sans conflits de chariots autoguidés bidirectionnels, CIFA 2002 N V2 V1 V2 V1 If a vehicle undergoes a significant delay, some other vehicles having to cross some common nodes will be delayed Will undergo a significant delay too

23 22 The improved robust AGV routing How to improve the robust routing control, by modifying the nodes crossing order, without causing conflicts ?

24 23 Example V1V1 V2V2 V3V3 {1}{1,3} {1,3,2} {1,2} {2} {3} V 1 is the late AGV Can V 2 cross the node i and continue its trip without colliding with V 1 its predecessor on that node ? i

25 24 i j V k m O i ={U, V}O m ={U, V} O k ={U, V} O j ={U, V}{U} {V} {U} A. Approach by delaying the late AGV U MAZA & Castagna, Routage robuste sans conflits de chariots autoguidés bidirectionnels, CIFA 2002

26 25 i j V k m O i ={U, V}O m ={U, V} O k ={U, V} O j ={U, V}{U} {V} {U} A. Approach by delaying the late AGV U MAZA & Castagna, Routage robuste sans conflits de chariots autoguidés bidirectionnels, CIFA 2002

27 26 i j V k m O i ={U, V}O m ={U, V} O k ={U, V} O j ={U, V}{U} {V} {U} U U is outside the blue zone the re- ordering is possible A. Approach by delaying the late AGV U MAZA & Castagna, Routage robuste sans conflits de chariots autoguidés bidirectionnels, CIFA 2002

28 27 i j V k m O i ={V, U}O m ={V, U} O k ={V, U} O j ={V, U}{U} {V} {U} U Delay-action of the vehicle throughout common way A. Approach by delaying the late AGV U MAZA & Castagna, Routage robuste sans conflits de chariots autoguidés bidirectionnels, CIFA 2002

29 28 i j V k m O i ={U, V}O m ={U, V} O k ={U, V} O j ={U, V}{U} {V} {U} U U is on the common path the vehicle V must wait A. Approach by delaying the late AGV U MAZA & Castagna, Routage robuste sans conflits de chariots autoguidés bidirectionnels, CIFA 2002

30 29 B. Approach by advancing the AGV V V2V2 {1}{1,3} {1,3,2} {1,2} {2} {3} NjNj V 2 is the first AGV in the list ? LiLi P={V 1 }P={V 1,V 3 } The re-ordering is possible update nodes associated lists for each node belonging to the path [N j, M] M MAZA & Castagna, Robust conflict-free routing of bi-directional automated guided vehicles, SMC02

31 30 B. Approach by advancing the AGV V V2V2 NjNj M V1V1 V2V2 V1V1 V3V3 V2V2 V1V1 V3V3 V2V2 V2V2 V3V3 V2V2 V1V1 V3V3 V1V1 V2V2 V3V3 V2V2 V1V1 V1V1 V2V2 V2V2 V1V1 V3V3 V2V2 V1V1 V3V3 V3V3 V1V1 MAZA & Castagna, Robust conflict-free routing of bi-directional automated guided vehicles, SMC02

32 31 Summary The simulation control scheme Reading of missions The CFSTP predictive algorithm Other Missions to be planned? Is the mission possible ? The blocked AGV is sent to the garage yes No yes A list of nodes d i to be visited AGVs move to node d i The cross of the node d i yes Call one of the robust routing algorithm No d i = destination node ? yes ?

33 32 Improved robust control Robust Control Gain of optimisation 0 2000 4000 6000 8000 10000 12000 0%7%13%20% AGVs Failure Rate Total duration of missions realization AGVs Failure Rates Robust AGV Routing (Time units) Robust AGV Routing (Time units) Gain (Time units) % of Gain 0,00% 6,66% 13,33% 20,00% 6825 7370 8731 10357 6825 7087 8154 8799 0 283 577 1558 0,00% 3,84% 6,60% 15% The simulation results A manufacturing system example 17 1 23421 18567822 19910111223 201314151624 G G G G G G Lancement de la simulation informatique Lancement de la simulation informatique Lancement de la simulation informatique Lancement de la simulation informatique

34 33 Conclusion Perspectives - Study other routing algorithms - Study the sensitivity of these methods to the AGVs fleet size and other systems parameters - Implementation on a real system. We have proposed a routing method which combines the efficiency of a predictive method to the robustness of a reactive method. Our method is generic and can be applied to any network configuration.

35 34

36 35 Proof A B B A A B B A n m Before BA After AB m n B A Before After A B Catching-up conflict Head-on conflict n m

37 36 If the reserved time windows are arranged as follows, there will be no conflicts n m n m time A shift due to a contingency The predited time reserved windows The realized time reserved widows n m n m


Download ppt "1 Robust conflict-free routing of bi- directional Automated Guided Vehicles (AGVs) Institut de Recherche en Communication et Cybernétique de Nantes Samia."

Similar presentations


Ads by Google