Elevator control method, control device, and program

Abstract

Enable the predicted arrival time of the elevator car to be calculated with high prediction accuracy even in an environment where both users and robots use the elevator. 【Solution】Among the stop times of the elevator car at each floor, the time required when the robot boards (including the fully open holding time) is defined as the first target stop time. Also, the boarding and alighting patterns that may occur when the robot boards are defined as the first patterns, and prediction values of the first target stop time are associated one by one with the first patterns. Then, each time the elevator car stops at the floor where the robot boards, a process is performed to identify which pattern among the first patterns the boarding and alighting pattern that occurred at that floor matches. The prediction value of the first target stop time corresponding to the first pattern identified in that process is learned using the time actually required at that boarding floor.

Description

Technical Field

[0001] The present invention relates to an elevator control technology used by both users and robots.

Background Art

[0002] In recent years, robots have been increasingly used in various tasks (such as cleaning, monitoring, and transportation) within buildings (see, for example, Patent Document 1). Along with this, the use of elevators for the inter-floor movement of robots within buildings has been increasing, and the number of cases where both users and robots use elevators has been increasing.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] On the other hand, robots take more time to board and alight compared to users. For this reason, in an environment where both users and robots use elevators, the following problems can occur.

[0005] Generally, in the group management control of an elevator with multiple cars, when extracting candidates for landing calls from among the multiple cars, the predicted arrival time of each car at the departure floor indicated by the landing call is used as one of the judgment indicators (an indicator for judging for each car whether it should be a candidate for assignment). When calculating the predicted arrival time, while considering the scheduled stop floors where stops are scheduled until arriving at the departure floor indicated by the landing call, the predicted value of the time required for movement (running time) between two different scheduled stop floors, the predicted value of the time required from when the door starts to open until it fully opens on each floor (door opening operation time), the predicted value of the time required to hold the door fully open (fully open holding time), the predicted value of the time required from when the door starts to close until it fully closes (door closing operation time), etc. are used. Further, for these predicted values, those learned using the time actually required in the previous operations (hereinafter, this time will be referred to as the "measured value") are used.

[0006] Here, the running time is a time determined by a series of mechanical movements of a preset elevator mechanism, and the door opening operation time and the door closing operation time are times determined by a series of mechanical movements of a preset opening and closing mechanism. In other words, these times are hardly affected even if there are changes in the boarding and alighting situations at the scheduled stop floors. And for each of these times determined by a series of mechanical movements, even if the measured value may deviate from the theoretical value, the deviation itself often has reproducibility. Therefore, for these times, by performing learning using the measured value and reflecting the deviation in the predicted value, the predicted value can be converged to a value with high prediction accuracy.

[0007] On the one hand, the fully open holding time is a time that can change each time according to the boarding and alighting situation at the floor where stopping is scheduled. Therefore, even if learning using measured values is performed for this time, it is difficult to converge the predicted value, and there is a risk that the measured value will deviate from the predicted value each time. However, when only users use the elevator, since boarding and alighting are performed smoothly, even if the measured value deviates from the predicted value, it hardly becomes a significant deviation that causes problems.

[0008] However, in an environment where both users and robots use the elevator, because the robot takes time for boarding and alighting, at the floor where the robot boards and alights, the fully open holding time becomes significantly longer compared to the case where only users board and alight. Also, assuming there is a user boarding and alighting on the same floor as the robot, interference that hinders each other's boarding and alighting is likely to occur between the robot and the user. Due to such interference, at the floor where the robot boards and alights, the length of the fully open holding time varies greatly depending on the presence or absence of the user boarding and alighting on that floor, and also varies greatly depending on the difference in the boarding and alighting situation of the user.

[0009] Therefore, in an environment where both users and robots use the elevator, if learning of the predicted value for the fully open holding time is performed in the same way as when only users board and alight, there is a risk that the predicted value will change greatly each time of learning and become unstable. Such a predicted value may cause a large deviation from the measured value, thus reducing the prediction accuracy of the arrival prediction time and, by extension, potentially reducing the transportation efficiency of the elevator.

[0010] Therefore, an object of the present invention is to enable the arrival prediction time of the car to be calculated with high prediction accuracy even in an environment where both users and robots use the elevator.

Means for Solving the Problem

[0011] The first control method according to the present invention is a control method in an elevator, which is a part or all of the time including at least the fully open holding time, which is the time during which the door of the car is held fully open among the stop times of the car at each floor, and uses, as a first target stop time, the time required when a robot gets on the car, and performs learning of a predicted value for the first target stop time, and has the following configuration (Aspect 1). All or some of the boarding and alighting patterns that may occur for the user when the robot gets on the car are defined as a first pattern, and the predicted value of the first target stop time is associated with each of the first patterns one by one. Then, each time the car stops at the floor where the robot gets on, a process is performed to identify which pattern among the first patterns the boarding and alighting pattern that occurred on that floor matches, and the predicted value of the first target stop time corresponding to the first pattern identified by that process is learned using the time actually required on that floor.

[0012] According to the above Aspect 1, the presence or absence of users and the boarding and alighting situations that may occur when the robot gets on the car are patterned and classified, and for each of the first patterns (boarding and alighting patterns) obtained thereby, only the situations that can be represented by that first pattern are extracted, and then the predicted value of the first target stop time required when the robot gets on the car in that situation can be learned. Therefore, it becomes possible to obtain a stable value for each first pattern as the predicted value of the first target stop time.

[0013] The control method according to the above Aspect 1 may have the following configuration (Aspect 2). For each floor of the elevator, the predicted value of the first target stop time may be associated with each of the first patterns one by one. Then, each time the car stops at the floor where the robot gets on, a process is performed to identify which pattern among the first patterns the boarding and alighting pattern that occurred on that floor matches, and the one corresponding to the first pattern identified by that process among the predicted values of the first target stop time associated with the above floor may be learned using the time actually required on that floor.

[0014] According to the above-described aspect 2, the predicted value of the first target stop time corresponding to each first pattern can be learned while distinguishing it from other floors for each floor of the elevator. Therefore, it becomes possible to reflect the changes that may occur due to the difference in the boarding floors of the robots in the predicted value of the first target stop time.

[0015] The control method according to the above-described aspect 1 or 2 may have the following configuration (aspect 3). For each type of robot, the predicted value of the first target stop time may be associated with each first pattern one by one. Then, each time the car stops at the boarding floor of the robot, a process of identifying which pattern among the first patterns the boarding and alighting pattern that occurred at that boarding floor matches is performed, and among the predicted values of the first target stop time associated with the type of robot that boarded at the above-mentioned boarding floor, the one corresponding to the first pattern identified in that process is learned using the time actually required at that boarding floor.

[0016] According to the above-described aspect 3, by learning the predicted value of the first target stop time corresponding to each first pattern while distinguishing it by the type of robot, it becomes possible to reflect the changes that may occur due to the difference in the type of robot in the predicted value of the first target stop time. Also, when there are many types of robots, if an appropriate value is to be set manually for each type of robot as the predicted value of the first target stop time corresponding to each first pattern, cumbersome work will be imposed on the operator. However, according to the above-described aspect 3, due to learning, the predicted value of the first target stop time corresponding to each first pattern is updated to an appropriate value for each type of robot at any time, so such cumbersome work becomes unnecessary.

[0017] The control method according to any one of the above aspects 1 to 3 may have the following configuration (Aspect 4). For each state that may occur in the robot, a predicted value of the first target stop time may be associated one by one with a first pattern. Then, each time the car stops at the boarding floor of the robot, a process is performed to identify which pattern among the first patterns the boarding and alighting pattern that occurred on that boarding floor matches, and among the predicted values of the first target stop time associated with the state at the time of boarding of the robot that boarded on the above boarding floor, the one corresponding to the first pattern that could be identified in that process is learned using the time that actually became necessary on that boarding floor.

[0018] According to the above Aspect 4, by learning the predicted values of the first target stop time corresponding to each first pattern while distinguishing them by the states (types and situations of work, etc.) that may occur in the robot, it becomes possible to reflect changes that may occur due to differences in the state of the robot in the predicted values of the first target stop time. Further, when there are a wide variety of states of the robot, if an appropriate value is to be set manually for each state of the robot as the predicted value of the first target stop time corresponding to each first pattern, cumbersome work will be forced on the operator. However, according to the above Aspect 4, through learning, the predicted values of the first target stop time corresponding to each first pattern are updated to appropriate values for each state of the robot at any time, so such cumbersome work becomes unnecessary.

[0019] The second control method according to the present invention is a control method in an elevator, which is part or all of the time including at least the fully open holding time, which is the time during which the door of the car is held fully open among the stop times of the car at each floor, and uses, as the second target stop time, the time required when the robot gets off the elevator, and performs learning of a predicted value for the second target stop time, and has the following configuration (Aspect 5). All or part of the boarding and alighting patterns that may occur for the user when the robot gets off the elevator are defined as the second patterns, and the predicted value of the second target stop time is associated with each of the second patterns one by one. Then, each time the car stops at the floor where the robot gets off, a process is performed to identify which pattern among the second patterns the boarding and alighting pattern that occurred at that floor matches, and the predicted value of the second target stop time corresponding to the second pattern identified in that process is learned using the time actually required at that floor where the robot gets off.

[0020] According to the above Aspect 5, the presence or absence and boarding and alighting status of the user that may occur when the robot gets off the elevator are patterned and classified, and for each of the second patterns (boarding and alighting patterns) obtained thereby, only the situations that can be represented by the second pattern are extracted, and then the predicted value of the second target stop time required when the robot gets off the elevator under that situation can be learned. Therefore, it becomes possible to obtain a stable value for each second pattern as the predicted value of the second target stop time.

[0021] The control method according to any one of the above Aspects 1 to 5 may have the following configuration (Aspect 6). Among the boarding and alighting patterns that may occur for the user when the robot boards or gets off the elevator, (1) the pattern that there is no user other than the robot, (2) the pattern that a user waits inside the car, (3) the pattern that a user gets off the car, (4) the pattern that a user waits inside the car and another user gets off the car from that car, (5) the pattern that a user boards the car, (6) The pattern in which the user waits inside the car and another user boards the car. (7) The pattern in which the user gets off the car and another user boards the car. (8) The pattern in which the user waits inside the car, another user gets off the car, and yet another user boards the car. These may be set as the first pattern or the second pattern.

[0022] The first control device according to the present invention is a control device that, in an elevator, learns a predicted value of a first target stop time, which is at least a part or all of the time including the fully open holding time, which is the time during which the door of the car is held fully open, among the stop times of the car at each floor, and is the time required when the robot boards the car, and has the following configuration (Aspect 7). The control device sets all or some of the boarding and alighting patterns that may occur for the user when the robot boards as the first pattern, and associates a predicted value of the first target stop time with each of the first patterns one by one. Then, each time the car stops at the boarding floor of the robot, the control device performs a process of identifying which pattern among the first patterns the boarding and alighting pattern that occurred at that boarding floor matches, and learns the predicted value of the first target stop time corresponding to the first pattern identified by the process using the time actually required at that boarding floor.

[0023] The second control device according to the present invention is a control device that, in an elevator, learns a predicted value of a second target stop time, which is part or all of the time including at least the fully open holding time, which is the time during which the car door is held fully open, among the stop times of the car at each floor, and is the time required when a robot gets off the elevator, and has the following configuration (Aspect 8). The control device sets all or a part of the boarding and alighting patterns that may occur for the user when the robot gets off the elevator as the second patterns, and associates a predicted value of the second target stop time with each of the second patterns one by one. Then, each time the car stops at the floor where the robot gets off, the control device performs a process of identifying which pattern among the second patterns the boarding and alighting pattern that occurred at that floor matches, and learns the predicted value of the second target stop time corresponding to the second pattern identified by that process using the time actually required at that floor.

[0024] The first program according to the present invention is a program that causes a control device to learn a predicted value of a first target stop time, which is part or all of the time including at least the fully open holding time, which is the time during which the car door is held fully open, among the stop times of the car at each floor, and is the time required when a robot gets on the elevator, and causes the control device to perform the following processes (Aspect 9). The control device is caused to set all or a part of the boarding and alighting patterns that may occur for the user when the robot gets on the elevator as the first patterns, and associate a predicted value of the first target stop time with each of the first patterns one by one. Then, each time the car stops at the floor where the robot gets on, the control device is caused to perform a process of identifying which pattern among the first patterns the boarding and alighting pattern that occurred at that floor matches, and learn the predicted value of the first target stop time corresponding to the first pattern identified by that process using the time actually required at that floor.

[0025] The second program according to the present invention is a program that causes a control device to execute the following processing (Aspect 10) in an elevator. The second program includes at least a part or all of the time including the fully opened holding time, which is the time during which the door of the car is held fully open, among the stop times of the car at each floor, and sets the time required when the robot gets off as the second target stop time, and learns a predicted value for the second target stop time. The control device is caused to execute the following: Regarding all or some of the boarding and alighting patterns that may occur for the user when the robot gets off as the second patterns, the control device is caused to associate a predicted value of the second target stop time with each of the second patterns one by one. Then, each time the car stops at the floor where the robot gets off, the control device is caused to perform a process of identifying which pattern among the second patterns the boarding and alighting pattern that occurred at that floor matches, and to learn the predicted value of the second target stop time corresponding to the second pattern identified by the process, using the time actually required at that floor.

Effects of the Invention

[0026] According to the present invention, the predicted arrival time of the car can be calculated with high prediction accuracy even in an environment where both the user and the robot use the elevator.

Brief Description of the Drawings

[0027]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Mode for Carrying Out the Invention

[0028] [1] Embodiment [1-1] Overall Configuration of Elevator FIG. 1 is a conceptual diagram showing the overall configuration of an elevator according to an embodiment. In this embodiment, the elevator includes a plurality of carriages G, and these carriages G are used not only by users but also by robots H that perform various operations (such as cleaning, monitoring, and transportation) within the building where the elevator is installed. Also, in this elevator, a first operation unit 1 for a user to specify the destination direction Kc is installed at the landing on each floor, and a second operation unit 2 for a user to specify the destination floor Fd is installed inside each carriage G. And in addition to these configurations, the elevator of this embodiment further includes a robot management device 3 and a group management control device 4. Hereinafter, the configuration of each part will be specifically described.

[0029] <First Operation Unit> The first operation unit 1 includes an upward button for specifying the upward direction as the destination direction Kc and a downward button for specifying the downward direction as the destination direction Kc on floors other than the topmost or bottommost end floors. On the other hand, on the topmost floor, the first operation unit 1 includes only the downward button, and on the bottommost floor, the first operation unit 1 includes only the upward button.

[0030] When a user operates the first operation unit 1 at the landing (presses the upward or downward button) to specify the destination direction Kc, that destination direction Kc is transmitted to the group management control device 4. Thereby, the assignment of the landing call X for the user (assignment to the carriage G) is requested to the group management control device 4 (assignment request from the user). At this time, in order to make the group management control device 4 recognize which floor's operation unit the operated first operation unit 1 is, the device information Pd1 for identifying the operation unit from other operation units, devices, etc. is also transmitted to the group management control device 4. Hereinafter, the landing call X of the user will be referred to as "landing call Xg".

[0031] <Second Operation Unit> The second operation unit 2 includes a plurality of destination floor buttons, and each of the destination floor buttons is provided corresponding to one of the plurality of floors that can be guided by the elevator of the present embodiment. Then, when each destination floor button is pressed at the second operation unit 2, the floor corresponding to the button is registered as the destination floor Fd.

[0032] When the user operates the second operation unit 2 (presses the destination floor button) in the car G to register the destination floor Fd, the destination floor Fd is transmitted to the group management control device 4. Thereby, the registration of the car call Y for the user (registration to the car G) is requested to the group management control device 4. At this time, in order to make the group management control device 4 recognize which car G the operated second operation unit 2 is in, the device information Pd2 for identifying the operation unit from other operation units and devices is also transmitted to the group management control device 4. Hereinafter, the car call Y of the user will be referred to as "car call Yg".

[0033] <Robot management device> The robot management device 3 is a device that centrally manages the robots H used in the building where the elevator of the present embodiment is installed.

[0034] In the present embodiment, when each robot H needs to move between floors, it transmits the disembarkation floor Fy that is the destination to the robot management device 3. At this time, in order to make the robot management device 3 recognize which robot H the transmission source of the disembarkation floor Fy is, the robot information Ph for identifying itself from other robots H is also transmitted to the robot management device 3.

[0035] When the robot management device 3 receives the alighting floor Fy and robot information Ph from each robot H, it makes a request to the group management control device 4 for allocating the landing call X for that robot H (allocation request process; see Fig. 9). Hereinafter, the landing call X for the robot H will be referred to as the "landing call Xh". Thereafter, the robot management device 3 causes the robot H to board and alight from the car G to which the landing call Xh has been allocated at an appropriate timing (boarding / alighting command process; see Fig. 18). Details of these processes will be described later.

[0036] As a specific configuration, the robot management device 3 includes a storage unit 31 and a control unit 32 (see Fig. 1).

[0037] The storage unit 31 is a part composed of storage devices such as ROM and RAM, and information necessary for the control processes performed by the robot management device 3 is stored in the storage unit 31. In the present embodiment, as such information, robot management data Dp and allocation request management data Ds are stored in the storage unit 31.

[0038] Here, the robot management data Dp is a database for managing, for each robot H, a plurality of pieces of information related to that robot H by associating them with each other. The allocation request management data Ds is data for managing information on allocation requests for the robot H. Specifically, it is as follows.

[0039] Fig. 2(A) is a conceptual diagram illustrating the robot management data Dp used in the present embodiment. In the robot management data Dp, for each robot H, the robot information Ph and boarding floor Fx of that robot H, and the destination when the robot H moves between floors are recorded in a state where they are associated with each other. Here, the boarding floor Fx associated with each robot H is the floor on which that robot H is arranged, and is updated each time the robot H moves between floors. Further, in the destination associated with each robot H, the alighting floor Fy transmitted by that robot H for moving between floors is recorded, and the alighting floor Fy is deleted when the alighting of that robot H on that floor is completed.

[0040] Accordingly, when the robot management device 3 receives the robot information Ph together with the getting-off floor Fy from each robot H, it becomes possible to specify the boarding floor Fx of the robot H from the robot information Ph. In the present embodiment, the boarding floor Fx of the robot H is used as the departure floor Fc when the robot H moves between floors using the elevator car G. Further, the robot management device 3 refers to the destination associated with the robot information Ph of each robot H. When the getting-off floor Fy is recorded at the destination, it can be determined that the robot H is in the process of moving between floors. On the other hand, when the getting-off floor Fy is not recorded at the destination, it can be determined that the robot H is deployed at the boarding floor Fx.

[0041] FIG. 2(B) is a conceptual diagram illustrating the allocation request management data Ds used in the present embodiment. In the allocation request management data Ds, each time an allocation request for each robot H is made to the group management control device 4, the robot information Ph of the robot H, the departure floor Fc and the destination floor Fd transmitted to the group management control device 4 in the allocation request are recorded in a state of being associated with each other. In the present embodiment, the information of each allocation request is deleted from the allocation request management data Ds when the car G arrives at the destination floor Fd associated with the allocation request (specifically, after the getting-off completion signal Sy described later is transmitted to the group management control device 4).

[0042] The control unit 32 is a part responsible for executing the control processes (including the allocation request process and the boarding / alighting command process) performed by the robot management device 3. Specifically, the control unit 32 is composed of processing devices such as a CPU and an MPU, and realizes the execution of the control processes it is responsible for in software by executing the control program installed in the robot management device 3. Incidentally, this control program may be stored in a portable storage medium (e.g., a flash memory, etc.) in a readable state before being installed in the robot management device 3, or may be stored in a downloadable state in another server or the like. Further, the control processes performed by the robot management device 3 are not limited to being realized in software by program execution, and may also be realized in hardware by a processing circuit (control unit 32) constructed in the robot management device 3.

[0043] <Group management control device> The group management control device 4 is a device that centrally manages a plurality of carriages G provided in the elevator of the present embodiment through the elevator control device provided for each carriage G.

[0044] In this embodiment, when a allocation request is received from either a user present at a landing or the robot management device 3, the group management control device 4 extracts an allocation candidate Gm from among a plurality of carriages G, and allocates a landing call X (Xg in the case of an allocation request from a user, and Xh in the case of an allocation request from the robot management device 3) to the allocation candidate Gm (if there are multiple candidates, one of the allocation candidates Gm) (allocation process; see FIG. 10). At this time, as one of the judgment indicators for extracting the allocation candidate Gm (an indicator for judging whether each carriage G should be the allocation candidate Gm), the group management control device 4 calculates the estimated arrival time Tc of each carriage G at the departure floor Fc transmitted in the allocation request (time calculation process; see FIG. 11). Then, the group management control device 4 controls the operation of each carriage G through the elevator control device, causing each carriage G to execute a response operation to the landing call X allocated to that carriage (response process; see FIGS. 13 to 17). Although the details of these processes will be described later, the time calculation process will be explained in a little more detail here.

[0045] When the group management control device 4 receives an allocation request from a user and extracts an allocation candidate Gm, it designates the boarding floor and destination direction Kc (the direction specified by the first operation unit 1) of the user as the "target floor Fw" and the "target direction Kw", and calculates the estimated arrival time Tc of each carriage G at the target floor Fw in the target direction Kw as follows. Further, when the group management control device 4 receives an allocation request for the robot H from the robot management device 3 and extracts an allocation candidate Gm, it designates the boarding floor Fx and destination direction Kc (the direction from the boarding floor Fx to the disembarking floor Fy) of the robot H as the "target floor Fw" and the "target direction Kw", and calculates the estimated arrival time Tc of each carriage G at the target floor Fw in the target direction Kw as follows.

[0046] The group management control device 4 sets one of the carriages G (in other words, the carriage G that can become the allocation candidate Gm) provided in the elevator of the present embodiment as the "target carriage Gk", and among the floors that the target carriage Gk will pass through before arriving at the target floor Fw in the arrival direction Kw, if there is a scheduled stop floor Fk where the stop of the target carriage Gk is scheduled, the group management control device 4 calculates the predicted arrival time Tc by appropriately combining the following four types of predicted values Te1 to Te4 required for the target carriage Gk while considering the scheduled stop floor Fk. Then, the group management control device 4 executes such time calculation processing for each of all the carriages G provided in the elevator one by one.

[0047] (1) The predicted value Te1 of the time (travel time) required for the target carriage Gk to move from one scheduled stop floor Fk to the next scheduled stop floor Fk (here, it also includes the movement of the target carriage Gk from the current floor Fp to the first scheduled stop floor Fk and the movement of the target carriage Gk from the last scheduled stop floor Fk to the target floor Fw). (2) The predicted value Te2 of the time (door opening operation time) required for the target carriage Gk to stop and the door to start opening until it fully opens. (3) The predicted value Te3 of the time (fully open holding time) required to keep the door of the target carriage Gk fully open. (4) The predicted value Te4 of the time (door closing operation time) required for the door of the target carriage Gk to start closing until it fully closes.

[0048] Furthermore, in the present embodiment, the group management control device 4 learns the predicted values Te1 to Te4 of the above times (travel time, door opening operation time, fully open holding time, door closing operation time) at any time using the time (hereinafter, this time will be referred to as the "actual measured value Tp") that was actually required in the previous operation (learning process. Refer to FIG. 13). Then, the group management control device 4 calculates the predicted arrival time Tc of each carriage G using the learned predicted values Te1 to Te4.

[0049] As a specific configuration, the group management control device 4 includes a storage unit 41 and a control unit 42 (refer to FIG. 1).

[0050] The storage unit 41 is a part composed of storage devices such as ROM and RAM. In the storage unit 41, information necessary for the control processing performed by the group management control device 4 is stored. In the present embodiment, as such information, device management data Dq, landing call management data Dx, car call management data Dy, and predicted value management data Dt are stored in the storage unit 41.

[0051] Here, the device management data Dq includes device management data Dq1 for the first operation unit 1 and device management data Dq2 for the second operation unit 2. The landing call management data Dx includes landing call management data DxG for the user and landing call management data DxH for the robot H. The car call management data Dy includes car call management data DyG for the user and car call management data DyH for the robot H. The predicted value management data Dt includes four pieces of predicted value management data Dt1 to Dt4.

[0052] The device management data Dq1 is a database for associating and managing a plurality of pieces of information related to the first operation unit 1 for each first operation unit 1. The device management data Dq2 is a database for associating and managing a plurality of pieces of information related to the second operation unit 2 for each second operation unit 2. The landing call management data DxG and the car call management data DyG are data for managing information on the landing call Xg and the car call Yg for the user, respectively. The landing call management data DxH and the car call management data DyH are data for managing information on the landing call Xh and the car call Yh for the robot H, respectively. The predicted value management data Dt1 to Dt4 are data for managing the predicted values Te1 to Te4 of the above-described times (travel time, door opening operation time, fully open holding time, door closing operation time), respectively. Specifically, it is as follows.

[0053] FIG. 3(A) is a conceptual diagram illustrating the device management data Dq1 for the first operation unit 1 used in the present embodiment. In the device management data Dq1, for each first operation unit 1, the device information Pd1 and the installation floor Fs of the first operation unit 1 are recorded in a state where they are associated with each other.

[0054] Accordingly, when the group management control device 4 receives the device information Pd1 from each first operation unit 1 together with the destination direction Kc, it becomes possible to specify the installation floor Fs of the first operation unit 1 (the first operation unit 1 where the user has specified the destination direction Kc) from the device information Pd1. In this embodiment, the installation floor Fs of the first operation unit 1 is used as the departure floor Fc (boarding floor) of the user who specified the destination direction Kc at the first operation unit 1.

[0055] FIG. 3(B) is a conceptual diagram illustrating the device management data Dq2 for the second operation unit 2 used in this embodiment. In the device management data Dq2, for each second operation unit 2, the device information Pd2 of the second operation unit 2 and the car information Pg of the car G in which the second operation unit 2 is installed are recorded in a state of being associated with each other.

[0056] Accordingly, when the group management control device 4 receives the device information Pd2 from each second operation unit 2 together with the destination floor Fd, it becomes possible to specify the car G (the car G where the user has registered the destination floor Fd) in which the second operation unit 2 is installed from the device information Pd2. Therefore, when the group management control device 4 registers the destination floor Fd received from each second operation unit 2 as the car call Yg, it becomes possible to specify the car G to be the registration destination.

[0057] FIG. 3(C) is a conceptual diagram illustrating the landing call management data DxG and the car call management data DyG for the users used in this embodiment, respectively.

[0058] In the landing call management data DxG (refer to the left diagram in Fig. 3(C)), for each car information Pg of the car G, the allocation status of the landing call Xg for the users of that car G is associated. Specifically, for each floor of the elevator and further for each direction in which the car G can move from that floor, the status indicating whether the allocation of the landing call Xg with a set of that floor and direction as the departure floor Fc and the destination direction Kc respectively is performed is associated as the allocation status. And in the example of Fig. 3(C), the allocation status for each direction from each floor is updated to "ON" when the allocation of the landing call Xg with a set of that floor and direction as the departure floor Fc and the destination direction Kc respectively is performed, and is shown to be updated to "OFF" when that landing call Xg is deleted.

[0059] Also, in the car call management data DyG (refer to the right diagram in Fig. 3(C)), for each car information Pg of the car G, the registration status of the car call Yg for the users of that car G is associated. Specifically, for each floor of the elevator, the status indicating whether the car call Yg of the user with that floor as the destination floor Fd is registered (in other words, whether the destination floor button for registering that floor as the destination floor Fd is pressed at the second operation unit 2 in the car G) is associated as the registration status. And in the example of Fig. 3(C), the registration status for each floor is updated to "ON" when the car call Yg with that floor as the destination floor Fd is registered, and is shown to be updated to "OFF" when that car call Yg is deleted.

[0060] Fig. 3(D) is a conceptual diagram exemplifying the landing call management data DxH and the car call management data DyH for the robot H used in this embodiment respectively.

[0061] In the landing call management data DxH (refer to the left diagram in Fig. 3(D)), for each car information Pg of the car G, the allocation status of the landing call Xh for the robot H with respect to the car G is associated. Specifically, each time the allocation of the landing call Xh for each robot H is performed, the robot information Ph of the robot H, the departure floor Fc and the destination floor Fd indicated by the landing call Xh, are recorded in a state where they are associated with each other. And each landing call Xh is deleted by deleting the information in the landing call management data DxH corresponding to the landing call Xh when it has completed its role.

[0062] Also, in the car call management data DyH (refer to the right diagram in Fig. 3(D)), for each car information Pg of the car G, the registration status of the car call Yh for the robot H with respect to the car G is associated. Specifically, each time the registration of the car call Yh for each robot H is performed, the robot information Ph of the robot H and the destination floor Fd indicated by the car call Yh are recorded in a state where they are associated with each other. And each car call Yh is deleted by deleting the information in the car call management data DyH corresponding to the car call Yh when it has completed its role.

[0063] Fig. 4(A) is a conceptual diagram illustrating the predicted value management data Dt1 used in this embodiment. In the predicted value management data Dt1, for each car information Pg of the car G, the predicted value Te1 of the time (travel time) required for the movement of the car G between two different floors is associated. Specifically, assuming that two different floors are sequentially selected from all the floors of the elevator, when the first selected floor is the first stop floor Fk1 and the second selected floor is the second stop floor Fk2, for each combination (Fk1, Fk2), the predicted value Te1(Fk1, Fk2) of the travel time of the car G from the first stop floor Fk1 to arrive at the second stop floor Fk2 is associated.

[0064] FIG. 4(B) is a conceptual diagram exemplifying prediction value management data Dt2 and Dt4 used in the present embodiment. In the prediction value management data Dt2 (see the left diagram in FIG. 4(B)), for each car information Pg of the car G, a predicted value Te2(Fz) (Fz: variable representing the floor number) of the door opening operation time at each floor of the elevator is associated. Further, in the prediction value management data Dt4 (see the right diagram in FIG. 4(B)), for each car information Pg of the car G, a predicted value Te4(Fz) (Fz: variable representing the floor number) of the door closing operation time at each floor of the elevator is associated.

[0065] In the present embodiment, the prediction value management data Dt3 includes first management data DtA that manages a predicted value TeA of the fully open holding time required when the robot H boards, second management data DtB that manages a predicted value TeB of the fully open holding time required when the robot H gets off, and third management data DtC that manages a predicted value TeC of the fully open holding time when neither the boarding nor the alighting of the robot H occurs.

[0066] FIG. 5(A) is a conceptual diagram exemplifying the first management data DtA used in the present embodiment. In the first management data DtA, eight boarding and alighting patterns that may occur for the user when the robot H boards are set as the first pattern Pa(M) (M is a number from 1 to 8 for distinguishing the patterns; see FIGS. 6(A) to 6(H)). Then, for each floor of the elevator and further for each first pattern Pa(M), a predicted value TeA(Fz, M) (Fz: variable representing the floor number) of the fully open holding time required when the first pattern Pa(M) occurs on that floor is associated.

[0067] FIG. 5(B) is a conceptual diagram exemplifying the second management data DtB used in the present embodiment. In the second management data DtB, eight boarding and alighting patterns that may occur for the user when the robot H gets off are set as the second pattern Pb(N) (N is a number from 1 to 8 for distinguishing the patterns; see FIGS. 7(A) to 7(H)). Then, for each floor of the elevator and further for each second pattern Pb(N), a predicted value TeB(Fz, N) (Fz: variable representing the floor number) of the fully open holding time required when the second pattern Pb(N) occurs on that floor is associated.

[0068] Figs. 6(A) to 6(H) are conceptual diagrams each illustrating eight boarding and alighting patterns used as the first pattern Pa(M) in the present embodiment. Also, Figs. 7(A) to 7(H) are conceptual diagrams each illustrating eight boarding and alighting patterns used as the second pattern Pb(N) in the present embodiment. Specifically, for values of M and N indicating the pattern numbers, which are 1 to 8, the following eight boarding and alighting patterns are associated as follows.

[0069] 1: A pattern where there are no users other than the robot H (Figs. 6(A) and 7(A)), 2: A pattern where a user waits inside the car G (Figs. 6(B) and 7(B)), 3: A pattern where a user gets off from the car G (Figs. 6(C) and 7(C)), 4: A pattern where a user waits inside the car G and another user gets off from the same car G (Figs. 6(D) and 7(D)), 5: A pattern where a user boards the car G (Figs. 6(E) and 7(E)), 6: A pattern where a user waits inside the car G and another user boards the same car G (Figs. 6(F) and 7(F)), 7: A pattern where a user gets off from the car G and another user boards the same car G (Figs. 6(G) and 7(G)), 8: A pattern where a user waits inside the car G, another user gets off from the same car G, and yet another user boards the same car G (Figs. 6(H) and 7(H)).

[0070] According to the data management for the predicted value TeA(Fz, M) of the fully open holding time like this, the presence or absence of users and the boarding and alighting situations that may occur when the robot H boards are patterned and classified, and for each first pattern Pa(M) obtained thereby, an appropriate value (in this embodiment, the learned value) is associated as the predicted value TeA(Fz, M) of the fully open holding time. As a result, when calculating the predicted arrival time Tc, the boarding and alighting pattern occurring on the boarding floor Fx of the robot H is specified, and by using the predicted value TeA(Fz, M) of the fully open holding time corresponding to the boarding and alighting pattern, it becomes possible to calculate the predicted arrival time Tc using the predicted value TeA that is most likely to be actually required.

[0071] Also, according to the data management for the predicted value TeB(Fz, N) of the fully open holding time, the presence or absence of users and the boarding and alighting situations that may occur when the robot H alights are also patterned and classified, and for each second pattern Pb(N) obtained thereby, an appropriate value is associated as the predicted value TeB(Fz, N) of the fully open holding time. As a result, when calculating the predicted arrival time Tc, not only the boarding and alighting pattern occurring on the boarding floor Fx of the robot H but also the boarding and alighting pattern occurring on the alighting floor Fy of the robot H is specified, and by using the predicted value TeB(Fz, N) of the fully open holding time corresponding to the boarding and alighting pattern (the predicted value TeB that is most likely to be actually required), it becomes possible to calculate the predicted arrival time Tc.

[0072] FIG. 8 is a conceptual diagram illustrating the third management data DtC used in this embodiment. In the third management data DtC, for each floor of the elevator, the predicted value TeC(Fz) (Fz: a variable representing the floor number) of the fully open holding time required when there is neither boarding nor alighting of the robot H on that floor is associated.

[0073] The control unit 42 is a part responsible for executing the control processes (including the allocation process, response process, and learning process) performed by the group management control device 4. Specifically, the control unit 42 is composed of processing devices such as a CPU and an MPU, and realizes the execution of the control processes it is responsible for through software by executing the control program installed in the group management control device 4. Note that this control program may be stored in a portable storage medium (e.g., flash memory, etc.) in a readable state before being installed in the group management control device 4, or may be stored in a downloadable state in another server or the like. Also, the control processes performed by the group management control device 4 are not limited to being realized by software through program execution, and may also be realized by hardware by a processing circuit (control unit 42) constructed in the group management control device 4.

[0074] [1-2] Control Processes Executed by the Elevator [1-2-1] Allocation Request Process Performed by the Robot Management Device FIG. 9 is a flowchart showing the allocation request process executed in this embodiment. This allocation request process is started each time the robot management device 3 receives the getting-off floor Fy and robot information Ph from each robot H. In the allocation request process, the robot H that transmitted those pieces of information is referred to as the "target robot Hk". Also, the information (including the getting-off floor Fy and robot information Ph) received by the robot management device 3 at that time is collectively referred to as "received information Pr1".

[0075] When the allocation request process is started, the robot management device 3 searches the robot management data Dp (see FIG. 2(A)) for the robot information Ph in the received information Pr1 that matches the robot information Ph among the robot information Ph managed therein, and then records the getting-off floor Fy in the received information Pr1 at the destination associated with that robot information Ph (step S100). Thereby, the robot management device 3 can grasp the destination of the target robot Hk also in the boarding / alighting command process described later.

[0076] After that, the robot management device 3 makes a request to the group management control device 4 for allocating the landing call Xh for the target robot Hk, using the boarding floor Fx and the alighting floor Fy of the target robot Hk as the departure floor Fc and the destination floor Fd, respectively (step S110). Specifically, the robot management device 3 uses the boarding floor Fx and the alighting floor Fy of the target robot Hk as the departure floor Fc and the destination floor Fd of the target robot Hk, respectively, and then transmits this information together with the robot information Ph of the target robot Hk to the group management control device 4. Also, the robot management device 3 records the information (robot information Ph, departure floor Fc, destination floor Fd) transmitted to the group management control device 4 in the allocation request management data Ds in a state where they are associated with each other as the information of the allocation request (see Fig. 2(B)). After step S110, the robot management device 3 terminates the allocation request process.

[0077] [1-2-2] Allocation process performed by the group management control device Fig. 10 is a flowchart showing the allocation process executed in this embodiment. This allocation process starts when an allocation request for a landing call is made to the group management control device 4 from the user (the first operation unit 1) or the robot management device 3.

[0078] Hereinafter, the information received by the group management control device 4 each time an allocation request is made will be collectively referred to as "received information Pr2". Specifically, when the allocation request is a request from the user (the first operation unit 1) (an allocation request for the landing call Xg for the user), the received information Pr2 is a set of information including the destination direction Kc and the device information Pd1, and when the allocation request is a request from the robot management device 3 (an allocation request for the landing call Xh for the robot H), the received information Pr2 is a set of information including the departure floor Fc, the destination floor Fd, and the robot information Ph.

[0079] When the allocation process starts, the group management control device 4 determines which of the device information Pd1 and the robot information Ph is included in the received information Pr2 in order to determine whether the received allocation request is from the user (the first operation unit 1) or the robot management device 3 (step S200).

[0080] When the group management control device 4 determines in step S200 that the "device information Pd1" is included, based on this determination, it can be determined that the received allocation request is a request from the user (the first operation unit 1). In this case, first, the group management control device 4 uses the device management data Dq1 to search for the device information Pd1 in the device information Pd1 managed therein that matches the device information Pd1 in the received information Pr2, and then extracts the installation floor Fs associated with the device information Pd1, and uses it as the departure floor Fc of the user. Then, the group management control device 4 determines the landing call Xg of the user for which the allocation should be executed (step S210). Specifically, the group management control device 4 sets the departure floor Fc (= installation floor Fs) of the user and the destination direction Kc (the user's destination direction Kc) in the received information Pr2 as the landing call Xg of the user.

[0081] After step S210, the group management control device 4 extracts the allocation candidate Gm for the landing call Xg determined in step S210 from among the plurality of carriages G (steps S211 and S212). Specifically, it is as follows.

[0082] The group management control device 4 sets the departure floor Fc and the destination direction Kc indicated by the landing call Xg determined in step S210 as the "target floor Fw" and the "target direction Kw", and first calculates the estimated arrival time Tc of each carriage G at the target floor Fw in the target direction Kw by executing the time calculation process (see FIG. 11) described later (step S211). Here, the estimated arrival time Tc is the time from the current point in time expected to be required until the carriage G arrives at the target floor Fw in the target direction Kw. Note that in step S211, the time when the carriage G is expected to arrive at the target floor Fw in the target direction Kw may be calculated as the estimated arrival time Tc.

[0083] Next, the group management control device 4 extracts an allocation candidate Gm from among the plurality of cars G using the arrival prediction time Tc of each car G calculated in step S211 (step S212). More specifically, for each car G, the group management control device 4 uses the arrival prediction time Tc of the car G calculated in step S211 as one of the judgment criteria to determine whether or not the car G should be an allocation candidate Gm for the landing call Xg. Then, the group management control device 4 extracts, as the allocation candidate Gm, those cars G for which it can be determined that they "should be the allocation candidate Gm (Yes)".

[0084] Then, the group management control device 4 executes the allocation of the landing call Xg to the allocation candidate Gm extracted in step S212 (if there are multiple candidates, one of the allocation candidates Gm) (step S213. See the left figure in Fig. 3(C)). After that, the group management control device 4 ends the allocation process.

[0085] On the other hand, when the group management control device 4 determines in step S200 that the "robot information Ph" is included, based on this determination, it can be determined that the received allocation request is a request from the robot management device 3. In this case, the group management control device 4 first determines the landing call Xh for the robot H for which the allocation is to be executed (step S220). Specifically, the group management control device 4 sets the departure floor Fc and the destination floor Fd in the received information Pr2 as the landing call Xh for the robot H.

[0086] After step S220, the group management control device 4 extracts the allocation candidate Gm for the landing call Xh determined in step S220 from among the plurality of cars G (steps S221 and S222). Specifically, it is as follows.

[0087] The group management control device 4 sets the departure floor Fc indicated by the landing call Xh determined in step S220 as the "target floor Fw", and sets the destination direction Kc of the robot H going from that floor to the destination floor Fd (the destination floor Fd indicated by the landing call Xh) as the "target direction Kw". First, the estimated arrival time Tc of each car G at the target floor Fw in the target direction Kw is calculated by executing a time calculation process (see FIG. 11), which is the same as in step S211 (step S221).

[0088] Next, the group management control device 4 extracts an allocation candidate Gm from among the plurality of cars G using the estimated arrival time Tc of each car G calculated in step S221 (step S222). More specifically, for each car G, the group management control device 4 uses the estimated arrival time Tc of the car G calculated in step S221 as one of the judgment criteria to determine whether or not the car G should be an allocation candidate Gm for the landing call Xg. Then, the group management control device 4 extracts those cars G that can be determined as "should be (Yes) an allocation candidate Gm" among the cars G as the allocation candidate Gm.

[0089] Then, the group management control device 4 executes the allocation of the landing call Xh to the allocation candidate Gm extracted in step S222 (if there are multiple candidates, one of the allocation candidates Gm) (step S223. See the left figure in FIG. 3(D)). After that, the group management control device 4 terminates the allocation process.

[0090] [1-2-3] Time calculation process performed by the group management control device FIG. 11 is a flowchart showing the time calculation process executed in this embodiment. This time calculation process is executed one by one for each of all the cars G (in other words, the cars G that can become allocation candidates Gm) provided in the elevator of this embodiment in steps S211 and S221 in the above-described allocation process (see FIG. 10), with each being regarded as the "target car Gk".

[0091] In the time calculation process, the group management control device 4 calculates the predicted arrival time Tc of the target car Gk as follows using the elevator information Pe it currently holds. At this time, as the elevator information Pe, the group management control device 4 uses information on the current position and moving direction indicating the operating status of the target car Gk, and further, information in the landing call management data Dx and car call management data Dy indicating the usage status of the target car Gk.

[0092] When the time calculation process starts, the group management control device 4 first performs initial settings for five variables Fz1, Fz2, Kz1, Kz2, and Tz used for calculating the predicted arrival time Tc (step S300). Here, the variable Fz1 is a variable for indicating the scheduled stop floor Fk (including the current floor Fp here) at which the target car Gk is scheduled to stop until it arrives at the target floor Fw in the target direction Kw during the process of the target car Gk making a circular movement within the operation section (see Fig. 12). The variable Kz1 is a variable for indicating the departure direction Kx of the target car Gk from the scheduled stop floor Fk indicated by the variable Fz1. The variables Fz2 and Kz2 are variables for sequentially indicating the floors that the target car Gk will pass through and the directions at that time when the target car Gk is virtually moved until it can arrive at the target floor Fw in the target direction Kw (see Fig. 12). The variable Tz is a variable for accumulating the predicted values Te1 to Te4 of the time required each time for moving the target car Gk until it can arrive at the target floor Fw in the target direction Kw among the four types of time (travel time, door opening operation time, fully open holding time, door closing operation time).

[0093] And in step S300, the group management control device 4 sets the values of both the variables Fz1 and Fz2 to the current floor Fp of the target car Gk as initial values, sets the values of both the variables Kz1 and Kz2 to the departure direction Kx of the target car Gk from the current floor Fp, and sets the value of the variable Tz to 0 (zero). Hereinafter, the departure direction Kx from the current floor Fp will be referred to as the "departure direction Kxp".

[0094] Next, the group management control device 4 determines whether the target car Gk is stopped and scheduled to depart in the direction Kw towards the target floor Fw. Specifically, it determines whether the value of the variable Fz1 (= current floor Fp) matches the target floor Fw and whether the value of the variable Kz1 (= departure direction Kxp) matches the target direction Kw (step S301).

[0095] If the group management control device 4 determines in step S301 that "both the floor and the direction match (Yes)", based on this determination, it can be determined that the target car Gk is stopped and scheduled to depart in the direction Kw towards the target floor Fw. In this case, the group management control device 4 sets the arrival prediction time Tc of the target car Gk to Tz (= 0) (step S302). Thereafter, the group management control device 4 terminates the time calculation process being executed for the target car Gk.

[0096] If the group management control device 4 determines in step S301 that "at least one of the floor and the direction does not match (No)", based on this determination, it can be determined that the target car Gk needs to pass through other floors before arriving at the target floor Fw in the target direction Kw.

[0097] In this case, the group management control device 4 extracts one by one in the order of stop the scheduled stop floors Fk where the target car Gk is scheduled to stop from among the floors passed through when the target car Gk moves within the predetermined section Rz (Fp, Kxp; Fw, Kw) (steps S310 to S312), calculates the predicted time required for the target car Gk to depart from the current floor Fp until it departs from the scheduled stop floor Fk (steps S313 to S360), and finally calculates the predicted time required for the target car Gk to arrive at the target floor Fw in the target direction Kw (step S370), thereby calculating the arrival prediction time Tc of the target car Gk. Here, the predetermined section Rz (Fp, Kxp; Fw, Kw) is the section passed through by the target car Gk from the current floor Fp in the departure direction Kxp until it arrives at the target floor Fw in the target direction Kw among the circular routes (see Fig. 12) passed through when the target car Gk moves in a circular motion within the operation section. This will be specifically described below.

[0098] The group management control device 4 uses route management data Dr (see FIG. 12) to virtually move the target car Gk from the current floor Fp to the target floor Fw within a predetermined section Rz (Fp, Kxp; Fw, Kw), thereby identifying the floor that the target car Gk will pass through next after passing through the floor of variable Fz2 in the direction of variable Kz2 and the direction at that time, and setting them as the new values of variables Fz2 and Kz2 (step S310). For example, when "7F" and "downward" are assigned to variables Fz2 and Kz2, in step S310, "6F" and "downward" are identified as the next floor and direction (see FIG. 12) to virtually move the target car Gk to the next floor from there, and they are set as the new values of variables Fz2 and Kz2.

[0099] Next, the group management control device 4 determines whether the target car Gk virtually moved in step S310 has arrived at the target floor Fw in the target direction Kw by determining whether the value of variable Fz2 matches the target floor Fw and whether the value of variable Kz2 matches the target direction Kw (step S311).

[0100] If the group management control device 4 determines in step S311 that "at least one of the floor and the direction does not match (No)", based on this determination, it can be determined that the target car Gk virtually moved in step S310 has not yet arrived at the target floor Fw in the target direction Kw. In this case, the group management control device 4 determines whether the target car Gk is scheduled to stop at the floor of variable Fz2 in the direction of variable Kz2 (step S312). In other words, the group management control device 4 determines whether the floor of variable Fz2 is the scheduled stop floor Fk of the target car Gk during the period until the target car Gk arrives at the target floor Fw in the target direction Kw. Specifically, the group management control device 4 determines whether an assignment of a landing call X with the floor and direction indicated by variables Fz2 and Kz2 as the departure floor Fc and the destination direction Kc (in the case of a landing call Xh, the direction from the departure floor Fc to the destination floor Fd), or a registration of a car call Y with the floor indicated by variable Fz2 as the destination floor Fd, has been made for the target car Gk.

[0101] When the group management control device 4 determines "No (not scheduled to stop)" in step S312, it returns to step S310, virtually moves the target car Gk to the next floor, and then makes the determinations in steps S311 and S312 again. Then, the group management control device 4 repeatedly executes the processes of steps S310 to S312 until it can determine "Yes (both the floor and the direction match)" in step S311 or "Yes (scheduled to stop)" in step S312.

[0102] When the group management control device 4 determines "Yes (scheduled to stop)" in step S312, it uses the predicted value management data Dt1 (see Fig. 4(A)) to obtain the predicted value Te1(Fz1, Fz2) of the running time corresponding to the two floors of the variables Fz1 and Fz2. Also, it uses the predicted value management data Dt2 (see the left figure in Fig. 4(B)) to obtain the predicted value Te2(Fz2) of the door opening operation time corresponding to the floor of the variable Fz2. After that, it adds those predicted values Te1 and Te2 to the value of the variable Tz, and sets the obtained value as the new value of the variable Tz (step S313). Here, the predicted value Te1(Fz1, Fz2) is the expected time required for the target car Gk to move from the floor of the variable Fz1 to arrive at the floor of the variable Fz2, and the predicted value Te2(Fz2) is the expected time required for the door of the target car Gk to start opening until it fully opens on the floor of the variable Fz2. Thereby, the expected time required for the target car Gk to start from the current floor Fp, stop at the floor of the variable Fz2, and the door to fully open is calculated.

[0103] After step S313, the group management control device 4 executes the following process to determine the predicted value Te3 of the full - open holding time required on the floor of the variable Fz2.

[0104] The group management control device 4 first determines whether there is boarding of the robot H at the floor of the variable Fz2 (step S314). Specifically, the group management control device 4 uses the information corresponding to the car information Pg of the target car Gk in the car call management data DxH for the robot H (see the left diagram in Fig. 3(D)) to check if there is a car call Xh in that information where the departure floor Fc (the boarding floor Fx of the robot H) matches the floor of the variable Fz2 and the direction from the departure floor Fc to the destination floor Fd (the alighting floor Fy from the boarding floor Fx of the robot H) matches the direction of the variable Kz2. In other words, the group management control device 4 determines whether such a car call Xh has been assigned to the target car Gk.

[0105] Furthermore, in both cases where the group management control device 4 determines "boarding exists (Yes)" in step S314 and where it determines "boarding does not exist (No)", the group management control device 4 further determines whether there is alighting of the robot H at the floor of the variable Fz2 (steps S315A and S315B). Specifically, the group management control device 4 uses the information corresponding to the car information Pg of the target car Gk in the car call management data DyH for the robot H (see the right diagram in Fig. 3(D)) to check if there is a car call Yh in that information where the destination floor Fd (the alighting floor Fy of the robot H) matches the floor of the variable Fz2. In other words, the group management control device 4 determines whether such a car call Yh has been registered for the target car Gk.

[0106] When the group management control device 4 determines "boarding exists (Yes)" in step S314 and determines "no alighting (No)" in step S315A, based on these determinations, it can be determined that there is "only boarding (boarding only)" of the robot H at the floor of the variable Fz2. In this case, the group management control device 4 uses the elevator information Pe it holds to identify which pattern among the first patterns Pa(M) (see Figs. 6(A) - 6(H)) the boarding and alighting pattern that occurs when the robot H boards at the floor of the variable Fz2 matches (pattern identification process. step S320).

[0107] While the group management control device 4 determines "No boarding" in step S314, if it determines "Yes alighting" in step S315B, based on these determinations, it can be determined that there is "only alighting (only alighting)" of the robot H at the level of the variable Fz2. In this case, the group management control device 4 uses the elevator information Pe it possesses to identify which pattern among the second patterns Pb(N) (see FIGS. 7(A) to 7(H)) the boarding / alighting pattern that occurs when the robot H alights at the level of the variable Fz2 matches (pattern identification process, step S330).

[0108] Here, the pattern identification process executed in steps S320 and S330 will be specifically described. First, the group management control device 4 makes the following three determinations [J1] to [J3] using the information corresponding to the car information Pg of the target car Gk among the landing call management data DxG and car call management data DyG (see FIG. 3(C)) for the user.

[0109] J1: Determination of whether there is a user boarding the target car Gk at the level of the variable Fz2 J2: Determination of whether there is a user alighting from the target car Gk at the level of the variable Fz2 J3: Determination of whether there is a user waiting inside the target car Gk at the level of the variable Fz2

[0110] Specifically, the following determinations are made. In determination [J1], the group management control device 4 determines whether there is a landing call Xg with the level of the variable Fz2 as the departure floor Fc and the variable Kz2 as the destination direction Kc in the above information. In determination [J2], the group management control device 4 determines whether there is a car call Yg with the level of the variable Fz2 as the destination floor Fd in the above information. In determination [J3], the group management control device 4 determines whether there is a car call Yg with a floor after the level of the variable Fz2 in the route management data Dr (see FIG. 12) as the destination floor Fd in the above information.

[0111] Then, based on the results of judgments [J1] to [J3], the group management control device 4 identifies one pattern (M = Mt) from the first patterns Pa(M) as follows (see FIGS. 6(A) to 6(H)), and also identifies one pattern (N = Nt) from the second patterns Pb(N) in step S330 (see FIGS. 7(A) to 7(H)). Note that "Yes" shown below indicates "boarding Yes" in judgment [J1], "alighting Yes" in judgment [J2], and "waiting Yes" in judgment [J3]. Also, "No" shown below indicates "boarding No" in judgment [J1], "alighting No" in judgment [J2], and "waiting No" in judgment [J3].

[0112] J1 = "No", J2 = "No", J3 = "No" ⇒ Mt = 1 or Nt = 1, J1 = "No", J2 = "No", J3 = "Yes" ⇒ Mt = 2 or Nt = 2, J1 = "No", J2 = "Yes", J3 = "No" ⇒ Mt = 3 or Nt = 3, J1 = "No", J2 = "Yes", J3 = "Yes" ⇒ Mt = 4 or Nt = 4, J1 = "Yes", J2 = "No", J3 = "No" ⇒ Mt = 5 or Nt = 5, J1 = "Yes", J2 = "No", J3 = "Yes" ⇒ Mt = 6 or Nt = 6, J1 = "Yes", J2 = "Yes", J3 = "No" ⇒ Mt = 7 or Nt = 7, J1 = "Yes", J2 = "Yes", J3 = "Yes" ⇒ Mt = 8 or Nt = 8.

[0113] After step S320, the group management control device 4 uses the first management data DtA (see FIG. 5(A)) to obtain a predicted value TeA(Fz2, Mt) of the fully open holding time corresponding to the level of the variable Fz2 and the first pattern Pa (M = Mt) specified in step S320. Then, the group management control device 4 adds the predicted value TeA to the value of the variable Tz, and sets the obtained value as the new value of the variable Tz (step S321). Here, the predicted value TeA(Fz2, Mt) is the predicted value of the fully open holding time required when the first pattern Pa (M = Mt) specified in step S320 occurs at the level of the variable Fz2, and is the predicted time required to keep the door of the target car Gk fully open when the robot H gets on at the level of the variable Fz2 (the planned stop level Fk).

[0114] After step S330, the group management control device 4 uses the second management data DtB (see FIG. 5(B)) to obtain a predicted value TeB(Fz2, Nt) of the fully open holding time corresponding to the level of the variable Fz2 and the second pattern Pb (N = Nt) specified in step S330. Then, the group management control device 4 adds the predicted value TeB to the value of the variable Tz, and sets the obtained value as the new value of the variable Tz (step S331). Here, the predicted value TeB(Fz2, Nt) is the predicted value of the fully open holding time required when the second pattern Pb (N = Nt) specified in step S330 occurs at the level of the variable Fz2, and is the predicted time required to keep the door of the target car Gk fully open when the robot H gets off at the level of the variable Fz2 (the planned stop level Fk).

[0115] When the group management control device 4 determines "Yes (there is boarding)" in step S314 and further determines "Yes (there is alighting)" in step S315A, based on these determinations, it can be determined that "both boarding and alighting occur (boarding and alighting both)" for the robot H at the level of the variable Fz2. In the present embodiment, when two robots H perform boarding and alighting respectively on the same level in this way, in order to ensure that the boarding and alighting of these two robots H are performed smoothly, first, the robot H in the target car Gk is commanded to alight, and then the robot H at the boarding area is commanded to board (see Fig. 18). Therefore, at the level of the variable Fz2, the robot H and the user in the target car Gk alight first, and then the robot H and the user at the boarding area board the target car Gk.

[0116] Therefore, when the group management control device 4 determines that "both boarding and alighting occur (boarding and alighting both)", first, it focuses only on the robot H that alights among the two robots H (in other words, assuming that there is no robot H that boards), and specifies which pattern among the second patterns Pb(N) (see Figs. 7(A) to 7(H)) the boarding and alighting pattern that occurs when that robot H alights at the level of the variable Fz2 matches (pattern specifying process. Step S340).

[0117] After that, the group management control device 4 focuses only on the robot H that boards among the two robots H (in other words, assuming that there is no robot H that alights), and specifies which pattern among the first patterns Pa(M) (see Figs. 6(A) to 6(H)) the boarding and alighting pattern that occurs when that robot H boards at the level of the variable Fz2 matches (pattern specifying process. Step S341).

[0118] Specifically, the group management control device 4 executes the following processes in steps S340 and S341.

[0119] When the robot H in the target car Gk gets off at the level of the variable Fz2, since the user who is scheduled to board at that level is at the boarding area, interference is likely to occur between the robot H getting off and the user at the boarding area. Therefore, in step S340, the group management control device 4 considers all the users in the target car Gk (including the user scheduled to get off at the level of the variable Fz2 and the user waiting in the target car Gk without getting off at the level of the variable Fz2) and the user scheduled to board at the level of the variable Fz2, makes the above three judgments [J1] - [J3], and based on the result of the judgment, specifies one pattern (N = Nt) from the second pattern Pb(N).

[0120] On the other hand, when the robot H at the boarding area boards the target car Gk, since the user scheduled to get off at the level of the variable Fz2 has already gotten off the target car Gk first, interference is unlikely to occur between the robot H boarding and the user scheduled to get off. Therefore, in step S341, the group management control device 4 considers the user waiting in the target car Gk without getting off at the level of the variable Fz2 and the user scheduled to board at the level of the variable Fz2, while assuming that there is no user scheduled to get off at the level of the variable Fz2, makes the above three judgments [J1] - [J3], and based on the result of the judgment, specifies one pattern (M = Mt) from the first pattern Pa(M).

[0121] In addition, when specifying the pattern in step S341 assuming that there is no user scheduled to get off at the level of the variable Fz2 in this way, the group management control device 4 can specify the usage status of all users at the level of the variable Fz2 at the time when the second pattern Pb(N = Nt) is specified in step S340 even without making the three judgments [J1] - [J3]. Therefore, by excluding the user scheduled to get off from it, one pattern (M = Mt) can be uniquely specified from the first pattern Pa(M). Specifically, the first pattern Pa(M = Mt) can be uniquely specified from the second pattern Pb(N = Nt) according to the following relationship.

[0122] Nt = 1 or 3 ⇒ Mt = 1, Nt = 2 or 4 ⇒ Mt = 2, Nt = 5 or 7 ⇒ Mt = 5, Nt = 6 or 8 ⇒ Mt = 6.

[0123] After steps S340 and S341, the group management control device 4 uses the first management data DtA (see FIG. 5(A)) to obtain a predicted value TeA(Fz2, Mt) of the fully open holding time corresponding to the level of the variable Fz2 and the first pattern Pa (M = Mt) specified in step S341, and uses the second management data DtB (see FIG. 5(B)) to obtain a predicted value TeB(Fz2, Nt) of the fully open holding time corresponding to the level of the variable Fz2 and the second pattern Pb (N = Nt) specified in step S340. Then, the predicted values TeA and TeB are added to the value of the variable Tz, and the value obtained thereby is set as the new value of the variable Tz (step S342).

[0124] According to the processing of steps S340 to S342, even when two robots H get on and off at the same level respectively, considering the situation where the robot H and the user in the target car Gk get off first, and then the robot H and the user at the boarding area get on, it is possible to predict the fully open holding time required at that level, so that the prediction accuracy of the fully open holding time can be improved.

[0125] When the group management control device 4 determines "No boarding" in step S314 and further determines "No alighting" in step S315B, based on these determinations, it can be determined that there is "neither boarding nor alighting (no boarding / alighting)" of the robot H at the level of the variable Fz2. In this case, the group management control device 4 uses the third management data DtC (see FIG. 8) to obtain a predicted value TeC(Fz2) of the fully open holding time corresponding to the level of the variable Fz2, and then adds the predicted value TeC to the value of the variable Tz, and the value obtained thereby is set as the new value of the variable Tz (step S350). Here, the predicted value TeC(Fz2) is the predicted value of the fully open holding time required when there is neither boarding nor alighting of the robot H at the level of the variable Fz2 (in other words, when only the user boards and alights), and it is the expected time required to keep the door of the target car Gk fully open when only the user boards and alights at that level.

[0126] After the group management control device 4 executes any one of steps S321, S331, S342, and S350, it uses the predicted value management data Dt4 (see the right diagram in FIG. 4(B)) to obtain the predicted value Te4(Fz2) of the door closing operation time corresponding to the level of the variable Fz2. Then, it adds the predicted value Te4 to the value of the variable Tz, and sets the value obtained thereby as the new value of the variable Tz (step S360). Here, the predicted value Te4(Fz2) is the expected time required from when the door of the target car Gk starts to close until it is fully closed at the level of the variable Fz2. Thereby, the expected time required from when the target car Gk departs from the current floor Fp until the target car Gk stops at the level of the variable Fz2 and then the door is fully closed after the opening and closing of the door is calculated.

[0127] Thereafter, the group management control device 4 sets the values of the variables Fz2 and Kz2 as the new values of the variables Fz1 and Kz1 respectively (step S361), and returns to step S310. Then, the group management control device 4 repeatedly executes the processes of steps S310 to S361 until it can determine in step S311 that "both the floor and the direction match (Yes)". Thereby, from among the floors passed through when the target car Gk moves within the predetermined section Rz(Fp, Kxp; Fw, Kw), the planned stop floors Fk of the target car Gk are extracted one by one in the stop order (steps S310 to S312), and the expected time required from when the target car Gk departs from the current floor Fp until it departs from the last planned stop floor Fk is calculated. And that expected time will be substituted into the variable Fz.

[0128] And when the group management control device 4 determines in step S311 that "both the floor and the direction match (Yes)", based on this determination, it can be determined that the target car Gk virtually moved in step S310 has arrived at the target floor Fw in the target direction Kw. In this case, the group management control device 4 calculates the predicted time required from when the target car Gk departs from the last scheduled stop floor Fk until the target car Gk arrives at the target floor Fw in the target direction Kw and then until the door fully opens. To do this, using the prediction value management data Dt1 (see Fig. 4(A)), it obtains the predicted value Te1(Fz1, Fw) of the running time corresponding to the floor of the variable Fz1 and the target floor Fw. Also, using the prediction value management data Dt2 (see the left figure in Fig. 4(B)), it obtains the predicted value Te2(Fw) of the door opening operation time corresponding to the target floor Fw. Then, the group management control device 4 calculates the predicted arrival time Tc of the target car Gk by adding those predicted values Te1 and Te2 to the value of the variable Tz (step S370). After that, the group management control device 4 ends the time calculation process being executed for the target car Gk.

[0129] [1-2-4] Response processing (including learning processing) performed by the group management control device Fig. 13 is a flowchart showing the response processing (including learning processing) executed in this embodiment. This response processing is a process for causing each car G to execute a response operation, and it is started at the timing when the next stop floor of the car G (here, this car G will be referred to as the "target car Gn") is determined for each car G.

[0130] Specifically, when the group management control device 4 virtually moves the target car Gn from the current floor Fp in the departure direction Kxp using the route management data Dr (see FIG. 12), for each floor that the target car Gn will pass through in order, the group management control device 4 determines whether at least one of the following operations is performed for the target car Gn: assignment of a landing call X with the floor as the departure floor Fc, and registration of a car call Y with the floor as the destination floor Fd. The group management control device 4 makes this determination using the information in the portion corresponding to the car information Pg of the target car Gn in the landing call management data Dx and the car call management data Dy (see FIGS. 3(C) and 3(D)). Then, after determining the floor where it was first determined that the operation was "performed" as the next stop floor of the target car Gn, the group management control device 4 starts the response process in FIG. 13. Here, the next stop floor of the target car Gn is referred to as the "target floor Fn", and the departure direction Kx of the target car Gn from that floor is referred to as the "target direction Kn".

[0131] Also, in the learning process, the group management control device 4 learns the predicted values Te1 to Te4 of the four types of time (travel time, door opening operation time, fully open holding time, door closing operation time) required for the target car Gn using the time (measured value Tp) that actually became necessary by causing the target car Gn to execute a response operation. Hereinafter, the learning of the predicted value Te3 of the fully open holding time will be described in detail.

[0132] When the response process is started, the group management control device 4 first determines what calls are included in the response target in that process (step S400). Specifically, the group management control device 4 extracts a landing call X with the target floor Fn as the departure floor Fc and a car call Y with the target floor Fn as the destination floor Fd from all the calls for which assignment or registration to the target car Gn has been performed. More specifically, the group management control device 4 performs the following extractions [E1] to [E4].[[]]

[0133] E1: Extraction using the information corresponding to the car information Pg of the target car Gn among the landing call management data DxG for users (refer to the left figure in Fig. 3(C)), and extraction of the landing call Xg in which the departure floor Fc and the destination direction Kc (the floor and direction where "ON" is set) respectively match the target floor Fn and the target direction Kn from among that information. E2: Extraction using the information corresponding to the car information Pg of the target car Gn among the car call management data DyG for users (refer to the right figure in Fig. 3(C)), and extraction of the car call Yg in which the destination floor Fd (the floor where "ON" is set) matches the target floor Fn from among that information. E3: Extraction using the information corresponding to the car information Pg of the target car Gn among the landing call management data DxH for the robot H (refer to the left figure in Fig. 3(D)), and extraction of the landing call Xh in which the departure floor Fc (the boarding floor Fx of the robot H) matches the target floor Fn, and the direction from the departure floor Fc to the destination floor Fd (the alighting floor Fy from the boarding floor Fx of the robot H) matches the target direction Kn from among that information. E4: Extraction using the information corresponding to the car information Pg of the target car Gn among the car call management data DyH for the robot H (refer to the right figure in Fig. 3(D)), and extraction of the car call Yh in which the destination floor Fd (the alighting floor Fy of the robot H) matches the target floor Fn from among that information.

[0134] In step S400, the group management control device 4 performs such extractions [E1] to [E4] for the response target, and then determines whether only the landing call X (either one or both of the user's landing call Xg and the robot H's landing call Xh) is included in the response target, or whether only the car call Y (either one or both of the user's car call Yg and the robot H's car call Yh) is included in the response target, or whether both the landing call X and the car call Y are included in the response target.

[0135] Furthermore, based on response processing (including the first to third response processes described later) and various determinations in the learning process (including the determination in step S400 above), the group management control device 4 also appropriately makes the following five determinations [Ja1], [Ja2], and [Jb1] to [Jb3] for identifying the boarding and alighting status of the robot H and the user.

[0136] Ja1: A determination as to whether the robot H has boarded the target car Gn at the target floor Fn. Ja2: A determination as to whether the robot H has alighted from the target car Gn at the target floor Fn. Jb1: A determination as to whether a user has boarded the target car Gn at the target floor Fn. Jb2: A determination as to whether a user has alighted from the target car Gn at the target floor Fn. Jb3: A determination as to whether a user is waiting inside the target car Gn at the target floor Fn.

[0137] When the group management control device 4 determines in step S400 that only "landing call X" is included, it executes the first response process (see FIG. 14). Furthermore, based on the determination result, the group management control device 4 determines that there is no alighting of the robot H or the user from the target car Gn at the target floor Fn (Ja2 = "no alighting", Jb2 = "no alighting").

[0138] When the group management control device 4 determines in step S400 that only "car call Y" is included, it executes the second response process (see FIG. 15). Furthermore, based on the determination result, the group management control device 4 determines that there is no boarding of the robot H or the user to the target car Gn at the target floor Fn (Ja1 = "no boarding", Jb1 = "no boarding").

[0139] When the group management control device 4 determines in step S400 that both "landing call X" and "car call Y" are included, it executes the third response process (see FIGS. 16 and 17).

[0140] Next, regarding the first to third response processes and the learning process, a specific explanation will be given including how the judgments [Ja1], [Ja2], and [Jb1] to [Jb3] are made therein.

[0141] <First Response Process (Part 1)> Figure 14 is a flowchart showing the first response process executed in this embodiment. In this first response process, the group management control device 4 first determines whether or not the boarding call Xh of the robot H is included in the response target (step S401).

[0142] When the group management control device 4 determines "not included (No)" in step S401, based on the determination result, it determines that only the user's boarding call Xg is included in the response target. In other words, the group management control device 4 determines that there is no boarding of the robot H on the target car Gn at the target floor Fn, while there is boarding of the user on the target car Gn at the target floor Fn (Ja1 = "no boarding", Jb1 = "boarding exists"). In this case, the group management control device 4 causes the response operation to the boarding call Xg to be executed on the target car Gn. Specifically, it is as follows.

[0143] The group management control device 4 transmits a command to stop the target car Gn in the destination direction Kc indicated by the boarding call Xg to the target car Gn at the departure floor Fc (the user's departure floor Fc) indicated by the boarding call Xg (step S410).

[0144] After step S410, the group management control device 4 determines whether or not the target car Gn has arrived at the departure floor Fc indicated by the boarding call Xg (step S411). Also, the group management control device 4 repeatedly executes step S411 until it can determine "arrived (Yes)" in step S411.

[0145] When the group management control device 4 determines "arrived (Yes)" in step S411, it deletes the boarding call Xg that has completed its role with this arrival (step S412).

[0146] After that, when the door of the target car Gn is fully opened, the group management control device 4 starts measuring (actually measuring) the fully opened holding time and continues the measurement until the start time of door closing. The group management control device 4 uses the actually measured value Tp obtained from this measurement in the learning (step S731) of the predicted value Te3 of the fully opened holding time described later.

[0147] When the door is fully opened, the user gets on the arrived target car Gn and registers their destination floor Fd by operating the second operation unit 2 in the target car Gn (by pressing the destination floor button). At this time, if the destination floor button corresponding to the user's destination floor Fd has already been pressed and registered by another user, the user only gets on the target car Gn.

[0148] Therefore, after step S412, the group management control device 4 determines whether an unregistered destination floor Fd has been pressed by the destination floor button in the target car Gn (step S413).

[0149] When the group management control device 4 determines "pressed (Yes)" in step S413, it registers the pressed destination floor Fd as the user's car call Yg in the target car Gn (step S414). As a result, it becomes possible to stop the target car Gn at the destination floor Fd registered by the user in the target car Gn. After that, the group management control device 4 proceeds to step S700 (see FIG. 13) and starts the learning process. Here, summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "no boarding", Ja2 = "no alighting", Jb1 = "boarding", Jb2 = "no alighting" (boarding and alighting status [C1]).

[0150] On the other hand, when the group management control device 4 determines "not pressed (No)" in step S413, it proceeds to step S700 without performing step S414. Also in this case, summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "no boarding", Ja2 = "no alighting", Jb1 = "boarding", Jb2 = "no alighting" (boarding and alighting status [C2]).

[0151] <Learning Process (Part 1)> In step S700 of FIG. 13, the group management control device 4 determines whether the measurement (actual measurement) of the fully open holding time has been completed due to the start of door closing. Also, the group management control device 4 repeatedly executes step S700 until it can determine "completed (Yes)" in step S700.

[0152] When the group management control device 4 determines "completed (Yes)" in step S700, in order to identify under what circumstances the actual measurement value Tp obtained by measurement was required based on the boarding and alighting status (presence / absence and pattern of boarding and alighting) of the robot H and the user at the target floor Fn, the following processing is executed using the previous judgment results.

[0153] First, the group management control device 4 executes the following processing to identify the boarding and alighting status of the robot H. The group management control device 4 discriminates whether the result of determination [Ja1] is "boarding" or "no boarding" (step S701). Further, in both cases where the group management control device 4 discriminates Ja1 = "boarding" and Ja1 = "no boarding" in step S701, the group management control device 4 discriminates whether the result of determination [Ja2] is "alighting" or "no alighting" (steps S702A and S702B).

[0154] Regarding the above-mentioned boarding and alighting statuses [C1] and [C2], the group management control device 4 discriminates Ja1 = "no boarding" in step S701, and further discriminates Ja2 = "no alighting" in step S702B. Then, based on these discrimination results, the group management control device 4 identifies that the boarding and alighting status of the robot H at the target floor Fn is a situation where neither "boarding" nor "alighting" exists ("no boarding and alighting") (C1 = robot "no boarding and alighting", C2 = robot "no boarding and alighting"). In this case, without identifying the boarding and alighting status of the user, the group management control device 4 learns the predicted value TeC(Fn) of the fully open holding time required when neither boarding nor alighting of the robot H occurs at the target floor Fn (in other words, when only the user boards and alights) (step S731).

[0155] Specifically, the group management control device 4 updates the predicted value TeC(Fn) of the fully open holding time recorded in the third management data DtC (see FIG. 8) after learning using the measured value Tp.

[0156] As an example, the group management control device 4 can calculate the average value (TeC×I + Tp) / (I + 1) using the predicted value TeC(Fn) (number of samples = I) recorded in the third management data DtC, and use the value obtained from this calculation as the new predicted value TeC(Fn) after learning. In other words, the group management control device 4 can use the average value of the previous measured value Tp (number of samples = I) measured only when the user gets on and off at the target floor Fn and the currently obtained measured value Tp as the new predicted value TeC(Fn) after learning.

[0157] <First Response Processing (Part 2)> When the group management control device 4 determines in step S401 of FIG. 14 that the landing call Xh of the robot H is "included (Yes)" in the response target, based on this determination, it determines that there is boarding of the robot H to the target car Gn at the target floor Fn (Ja1 = "boarding exists"). In this case, the group management control device 4 causes the target car Gn to execute a response operation to the landing call Xh, and also performs processing for the landing call Xg of the user if the landing call Xg of the user is further included in the response target. Specifically, it is as follows.

[0158] The group management control device 4 transmits a command to stop the target car Gn in the direction toward the destination floor Fd indicated by the landing call Xh to the departure floor Fc (the departure floor Fc of the robot H) indicated by the landing call Xh to the target car Gn (step S510).

[0159] After step S510, the group management control device 4 determines whether the target car Gn has arrived at the departure floor Fc indicated by the landing call Xh (step S511). Also, the group management control device 4 repeatedly executes step S511 until it can determine "arrived (Yes)" in step S511.

[0160] When the group management control device 4 determines "arrived (Yes)" in step S511, then, when the door of the target car Gn is fully opened, it starts measuring (actually measuring) the fully opened holding time and continues the measurement until the start time of closing the door. The group management control device 4 uses the actually measured value Tp obtained from this measurement in the learning (step S711) of the predicted value Te3 of the fully opened holding time described later.

[0161] Furthermore, the group management control device 4 determines whether it has received a boarding completion signal Sx for notifying the boarding completion of the robot H from the robot management device 3 in order to determine whether the boarding of the robot H has been completed at the departure floor Fc indicated by the landing call Xh (step S512). Also, the group management control device 4 repeatedly executes step S512 until it can determine "received (Yes)" in step S512.

[0162] When the group management control device 4 determines "received (Yes)" in step S512, it registers the destination floor Fd indicated by the landing call Xh (the destination floor Fd of the robot H) as the car call Yh of the robot H in the target car Gn (step S513). As a result, it becomes possible to stop the target car Gn at the destination floor Fd indicated by the landing call Xh. On the other hand, due to the registration of the car call Yh to such a target car Gn, the landing call Xh will finish its role. Therefore, the group management control device 4 deletes the landing call Xh that has finished its role.

[0163] Thereafter, the group management control device 4 determines whether the user's landing call Xg is further included in the response target in this first response process (step S514).

[0164] When the group management control device 4 determines "included (Yes)" in step S514, based on the determination result, it determines that there is a user boarding the target car Gn at the target floor Fn (Jb1 = "boarding"). In this case, the group management control device 4 executes the processes of steps S412 to S414 (deletion of the landing call Xg to registration of the car call Yg) as the process for the landing call Xg. Then, the group management control device 4 proceeds to step S700 (see FIG. 13) and starts the learning process. Here, summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "boarding", Ja2 = "no alighting", Jb1 = "boarding", Jb2 = "no alighting" (boarding and alighting status [C3]).

[0165] On the other hand, when the group management control device 4 determines "not included (No)" in step S514, based on the determination result, it determines that there is no user boarding the target car Gn at the target floor Fn (Jb1 = "no boarding"). In this case, the group management control device 4 proceeds to step S700 without performing the processes of steps S412 to S414. Here, summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "boarding", Ja2 = "no alighting", Jb1 = "no boarding", Jb2 = "no alighting" (boarding and alighting status [C4]).

[0166] <Learning process (Part 2)> Regarding the above-mentioned boarding and alighting status [C3] and [C4], the group management control device 4 determines Ja1 = "boarding" in step S701, and further determines Ja2 = "no alighting" in step S702A. Then, based on these determination results, the group management control device 4 specifies that the boarding and alighting status of the robot H at the target floor Fn was a "boarding only" status (C3 = robot "boarding only", C4 = robot "boarding only"). In this case, the group management control device 4 specifies which pattern among the first patterns Pa(M) (see FIGS. 6(A) to 6(H)) occurred at the target floor Fn as the boarding and alighting pattern generated when the robot H boards (step S710. Pattern specification process).

[0167] Specifically, the group management control device 4 first further determines whether there is a user waiting in the target car Gn at the target floor Fn. More specifically, the group management control device 4 uses the information corresponding to the car information Pg of the target car Gn in the car call management data DyG for users (see the right diagram in Fig. 3(C)) to determine whether there is a car call Yg in the information that uses a floor after the target floor Fn in the route management data Dr (see Fig. 12) as the destination floor Fd.

[0168] Then, based on the result of the determination [Jb3] and the results of the determinations [Jb1] and [Jb2] obtained in the previous processing, the group management control device 4 specifies one pattern (M = Mt) from the first patterns Pa(M) as follows. Note that "Yes" shown below means "boarding yes" in determination [Jb1], "alighting yes" in determination [Jb2], and "waiting yes" in determination [Jb3]. Also, "No" shown below means "boarding no" in determination [Jb1], "alighting no" in determination [Jb2], and "waiting no" in determination [Jb3].

[0169] Jb1 = "No", Jb2 = "No", Jb3 = "No" ⇒ Mt = 1, Jb1 = "No", Jb2 = "No", Jb3 = "Yes" ⇒ Mt = 2, Jb1 = "No", Jb2 = "Yes", Jb3 = "No" ⇒ Mt = 3, Jb1 = "No", Jb2 = "Yes", Jb3 = "Yes" ⇒ Mt = 4, Jb1 = "Yes", Jb2 = "No", Jb3 = "No" ⇒ Mt = 5, Jb1 = "Yes", Jb2 = "No", Jb3 = "Yes" ⇒ Mt = 6, Jb1 = "Yes", Jb2 = "Yes", Jb3 = "No" ⇒ Mt = 7, Jb1 = "Yes", Jb2 = "Yes", Jb3 = "Yes" ⇒ Mt = 8.

[0170] Therefore, when the group management control device 4 reaches step S710 after the determination of "included (Yes)" in step S514 of FIG. 14 (in the case of the boarding / alighting situation [C3]. Jb1 = "boarding", Jb2 = "no alighting"), in step S710, it is determined that a pattern of Mt = 5 or 6 has occurred according to the result of determination [Jb3].

[0171] Also, when the group management control device 4 reaches step S710 after the determination of "not included (No)" in step S514 of FIG. 14 (in the case of the boarding / alighting situation [C4]. Jb1 = "no boarding", Jb2 = "no alighting"), in step S710, it is determined that a pattern of Mt = 1 or 2 has occurred according to the result of determination [Jb3].

[0172] Thereafter, when the first pattern Pa(Mt) specified in step S710 occurs at the target floor Fn, the group management control device 4 learns the predicted value TeA(Fn, Mt) of the fully open holding time required at that time (step S711).

[0173] Specifically, the group management control device 4 updates the predicted value TeA(Fn, Mt) of the fully open holding time recorded in the first management data DtA after learning using the measured value Tp.

[0174] As an example, the group management control device 4 can calculate the average value (TeA×I + Tp) / (I + 1) using the predicted value TeA(Fn, Mt) (number of samples = I) recorded in the first management data DtA, and use the value obtained from this calculation as the new predicted value TeA(Fn, Mt) after learning. In other words, the group management control device 4 can use the average value of the previous measured value Tp (number of samples = I) measured when the first pattern Pa(Mt) occurred at the target floor Fn and the currently obtained measured value Tp as the new predicted value TeA(Fn, Mt) after learning.

[0175] <Second Response Processing (Part 1)> FIG. 15 is a flowchart showing the second response process executed in the present embodiment. In this second response process, the group management control device 4 first determines whether or not the car call Yh of robot H is included in the response target (step S402).

[0176] When the group management control device 4 determines "not included (No)" in step S402, based on the determination result, it determines that only the user's car call Yg is included in the response target. In other words, the group management control device 4 determines that there is no getting off of robot H at the target car Gn on the target floor Fn, while there is getting off of the user from the target car Gn on the target floor Fn (Ja2 = "no getting off", Jb2 = "getting off"). In this case, the group management control device 4 causes the target car Gn to execute a response operation to the car call Yg. Specifically, it is as follows.

[0177] The group management control device 4 transmits a command for stopping the target car Gn to the target floor Fd (the user's destination floor Fd) indicated by the car call Yg to the target car Gn (step S420).

[0178] After step S420, the group management control device 4 determines whether or not the target car Gn has arrived at the destination floor Fd indicated by the car call Yg (step S421). Also, the group management control device 4 repeatedly executes step S421 until it can determine "arrived (Yes)" in step S421.

[0179] When the group management control device 4 determines "arrived (Yes)" in step S421, it deletes the car call Yg that has completed its role with this arrival (step S422).

[0180] Thereafter, when the door of the target car Gn is fully opened, the group management control device 4 starts measuring (actual measurement) the fully open holding time and continues the measurement until the start time of door closing. The group management control device 4 uses the actually measured value Tp obtained from this measurement in the learning (step S731) of the predicted value Te3 of the fully open holding time described later.

[0181] After that, the group management control device 4 proceeds to step S700 (see FIG. 13) and starts the learning process. Here, when summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "No boarding", Ja2 = "No alighting", Jb1 = "No boarding", and Jb2 = "Alighting" (boarding and alighting status [C5]).

[0182] In the learning process, the group management control device 4, based on the discrimination from the results of Ja1 = "No boarding" and Ja2 = "No alighting" for the above boarding and alighting status [C5], specifies that the boarding and alighting status of the robot H at the target floor Fn was the status of "No boarding or alighting" (C5 = Robot "No boarding or alighting"). Therefore, the group management control device 4 proceeds to step S731 without specifying the user's boarding and alighting status, and when there is neither boarding nor alighting of the robot H at the target floor Fn (in other words, when there is only the user's boarding and alighting), the group management control device 4 performs learning of the predicted value TeC(Fn) of the fully open holding time required using the measured value Tp.

[0183] <Second Response Process (Part 2)> When the group management control device 4 determines in step S402 of FIG. 15 that the car call Yh of the robot H is "included (Yes)" in the response target, based on that determination result, it is determined that there is alighting of the robot H at the target floor Fn to the target car Gn (Ja2 = "Alighting"). In this case, the group management control device 4 causes the response operation to the car call Yh to be executed at the target car Gn, and when the user's car call Yg is further included in the response target, it also performs processing for the user's car call Yg. Specifically, it is as follows.

[0184] The group management control device 4 sends a command to the target car Gn to stop the target car Gn at the destination floor Fd indicated by the car call Yh (the destination floor Fd of the robot H) (step S520).

[0185] After step S520, the group management control device 4 determines whether or not the target car Gn has arrived at the destination floor Fd indicated by the car call Yh (step S521). Also, the group management control device 4 repeatedly executes step S521 until it can determine "arrived (Yes)" in step S521.

[0186] When the group management control device 4 determines "arrived (Yes)" in step S521, it deletes the car call Yh that has completed its role with this arrival (step S522).

[0187] After that, when the door of the target car Gn is fully opened, the group management control device 4 starts measuring (actual measurement) the fully opened holding time and continues the measurement until the start time of door closing. The group management control device 4 uses the actually measured value Tp obtained by this measurement in the learning (step S721) of the predicted value Te3 of the fully opened holding time.

[0188] Furthermore, the group management control device 4 determines whether or not it has received a getting-off completion signal Sy for notifying the completion of the getting-off of the robot H from the robot management device 3 in order to determine whether or not the getting-off of the robot H has been completed at the destination floor Fd indicated by the car call Yh (step S523). Also, the group management control device 4 repeatedly executes step S523 until it can determine "received (Yes)" in step S523.

[0189] When the group management control device 4 determines "received (Yes)" in step S523, it then determines whether or not the user's car call Yg is further included in the response target in this second response process (step S524).

[0190] When the group management control device 4 determines "included (Yes)" in step S524, based on the determination result, it determines that there is a user getting off from the target car Gn at the target floor Fn (Jb2 = "There is getting off"). In this case, the group management control device 4 executes the process of step S422 (deletion of the car call Yg) as the process for the car call Yg. After that, the group management control device 4 proceeds to step S700 (see FIG. 13) and starts the learning process. Here, summarizing the determination results of the boarding and alighting situations obtained up to this point, the results are Ja1 = "No boarding", Ja2 = "There is getting off", Jb1 = "No boarding", and Jb2 = "There is getting off" (boarding and alighting situation [C6]).

[0191] On the other hand, when the group management control device 4 determines "not included (No)" in step S524, based on the determination result, it determines that there is no user getting off from the target car Gn at the target floor Fn (Jb2 = "No getting off"). In this case, the group management control device 4 proceeds to step S700 without performing the process of step S422. Here, summarizing the determination results of the boarding and alighting situations obtained up to this point, the results are Ja1 = "No boarding", Ja2 = "There is getting off", Jb1 = "No boarding", and Jb2 = "No getting off" (boarding and alighting situation [C7]).

[0192] <Learning process (Part 3)> For the above-described boarding and alighting situations [C6] and [C7], the group management control device 4 discriminates Ja1 = "No boarding" in step S701, and further discriminates Ja2 = "There is getting off" in step S702B. Then, based on these discrimination results, the group management control device 4 specifies that the boarding and alighting situation of the robot H at the target floor Fn was a situation of "only getting off" (C6 = Robot "only getting off", C7 = Robot "only getting off"). In this case, the group management control device 4 specifies which pattern among the second patterns Pb(N) (see FIGS. 7(A) to 7(H)) occurred at the target floor Fn as the boarding and alighting pattern generated when the robot H gets off (step S720. Pattern specification process).

[0193] Specifically, the group management control device 4 first further determines whether there is a user waiting in the target car Gn at the target floor Fn [Jb3]. Then, based on the result of the determination [Jb3] and the results of the determinations [Jb1] and [Jb2] obtained in the previous processing, the group management control device 4 specifies one pattern (N = Nt) from the second patterns Pb(N) as follows. Note that "Yes" shown below means "boarding Yes" in determination [Jb1], "alighting Yes" in determination [Jb2], and "waiting Yes" in determination [Jb3]. Also, "No" shown below means "boarding No" in determination [Jb1], "alighting No" in determination [Jb2], and "waiting No" in determination [Jb3].

[0194] Jb1 = "No", Jb2 = "No", Jb3 = "No" ⇒ Nt = 1, Jb1 = "No", Jb2 = "No", Jb3 = "Yes" ⇒ Nt = 2, Jb1 = "No", Jb2 = "Yes", Jb3 = "No" ⇒ Nt = 3, Jb1 = "No", Jb2 = "Yes", Jb3 = "Yes" ⇒ Nt = 4, Jb1 = "Yes", Jb2 = "No", Jb3 = "No" ⇒ Nt = 5, Jb1 = "Yes", Jb2 = "No", Jb3 = "Yes" ⇒ Nt = 6, Jb1 = "Yes", Jb2 = "Yes", Jb3 = "No" ⇒ Nt = 7, Jb1 = "Yes", Jb2 = "Yes", Jb3 = "Yes" ⇒ Nt = 8.

[0195] Therefore, when the group management control device 4 reaches step S720 after the determination of "included (Yes)" in step S524 of FIG. 15 (in the case of the boarding / alighting situation [C6]. Jb1 = "boarding No", Jb2 = "alighting Yes"), in step S720, it is specified that a pattern with Nt = 3 or 4 has occurred according to the result of the determination [Jb3].

[0196] Further, when the group management control device 4 reaches step S720 after making a determination of "not included (No)" in step S524 of FIG. 15 (in the case of the boarding / alighting status [C7]. Jb1 = "not boarded", Jb2 = "not alighted"), in step S720, it is determined that a pattern of Nt = 1 or 2 has occurred according to the result of the determination [Jb3].

[0197] After that, when the second pattern Pb(Nt) specified in step S720 occurs at the target floor Fn, the group management control device 4 learns the predicted value TeB(Fn, Nt) of the fully open holding time required at that time (step S721).

[0198] Specifically, the group management control device 4 updates the predicted value TeB(Fn, Nt) of the fully open holding time recorded in the second management data DtB after learning it using the measured value Tp.

[0199] As an example, the group management control device 4 can calculate the average value (TeB×I + Tp) / (I + 1) using the predicted value TeB(Fn, Nt) (number of samples = I) recorded in the second management data DtB, and use the value obtained from this calculation as the new predicted value TeB(Fn, Nt) after learning. In other words, the group management control device 4 can use the average value of the previous measured value Tp (number of samples = I) measured when the second pattern Pb(Nt) occurred at the target floor Fn and the currently obtained measured value Tp as the new predicted value TeB(Fn, Nt) after learning.

[0200] <Third Response Process (Part 1)> FIGS. 16 and 17 are flowcharts showing the third response process executed in this embodiment. In this third response process, the group management control device 4 first determines whether or not the landing call Xh of the robot H is included in the response target (step S403). Further, the group management control device 4 determines whether or not the car call Yh of the robot H is included in the response target in both cases where it is determined as "not included (No)" and "included (Yes)" in step S403 (steps S404A and S404B).

[0201] When the group management control device 4 determines "not included (No)" in step S403 and further determines "not included (No)" in step S404A as well, based on these determination results, it determines that none of the calls for the robot H (such as the landing call Xh or the car call Yh) are included in the response target, while on the other hand, the user's landing call Xg is included in the response target. In other words, the group management control device 4 determines that there is neither boarding of the robot H onto the target car Gn nor alighting of the robot H from the target car Gn at the target floor Fn, while on the other hand, it determines that there is boarding of the user onto the target car Gn at the target floor Fn (Ja1 = "no boarding", Ja2 = "no alighting", Jb1 = "boarding exists").

[0202] In this case, the group management control device 4 executes the same processing as steps S410 to S414 (including measurement of the fully open holding time) described in the first response processing (steps S430 to S434). Then, the group management control device 4 further determines whether the user's car call Yg is included in the response target in this third response processing (step S561).

[0203] When the group management control device 4 determines "included (Yes)" in step S561, based on this determination result, it determines that there is alighting of the user from the target car Gn at the target floor Fn (Jb2 = "alighting exists"). Here, when the user's car call Yg is included in the response target, with the arrival of the target car Gn, that car call Yg will also have completed its role. Therefore, when the group management control device 4 determines "included (Yes)" in step S561, it deletes the car call Yg that has completed its role (step S562).

[0204] After that, the group management control device 4 proceeds to step S700 (see FIG. 13) to start the learning process. Here, summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "no boarding", Ja2 = "no alighting", Jb1 = "boarding exists", Jb2 = "alighting exists" (boarding and alighting status [C8]).

[0205] On the other hand, when the group management control device 4 determines "not included (No)" in step S561, based on the determination result, it determines that there is no getting-off of the user from the target car Gn at the target floor Fn (Jb2 = "no getting-off"). In this case, the group management control device 4 skips the process of step S562 and proceeds to step S700. Here, summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "no boarding", Ja2 = "no getting-off", Jb1 = "boarding", and Jb2 = "no getting-off" (boarding and alighting status [C9]).

[0206] In the learning process, for both of the above boarding and alighting statuses [C8] and [C9], the group management control device 4 specifies that the boarding and alighting status of the robot H at the target floor Fn was a "no boarding or alighting" status based on the discrimination from the results of Ja1 = "no boarding" and Ja2 = "no getting-off" (C8 = robot "no boarding or alighting", C9 = "no boarding or alighting"). Therefore, the group management control device 4 proceeds to step S731 without specifying the boarding and alighting status of the user, and learns the predicted value TeC(Fn) of the fully open holding time required when there is neither boarding nor alighting of the robot H at the target floor Fn (in other words, when there is only boarding and alighting of the user) using the measured value Tp.

[0207] <Third Response Process (Part 2)> While the group management control device 4 determines "not included (No)" in step S403 of FIG. 16, if it determines "included (Yes)" in step S404A, based on those determination results, it determines that the landing call Xh of the robot H is not included in the response target, while both the landing call Xg of the user and the car call Yh of the robot H are included in the response target. In other words, the group management control device 4 determines that there is no boarding of the robot H to the target car Gn at the target floor Fn, while there is boarding of the user and alighting of the robot H to the target car Gn at that target floor Fn (Ja1 = "no boarding", Ja2 = "alighting", Jb1 = "boarding").

[0208] In this case, the group management control device 4 performs the same processes as in steps S430 and S431 (steps S440 and S441). When it is determined in step S441 that "arrived (Yes)", first, as processing for the car call Yh of the robot H, the same processes as steps S522 and S523 (including measurement of the fully open holding time) described in the second response process are executed (steps S530 and S531). Then, when the group management control device 4 determines in step S531 that the getting-off completion signal Sy has been "received (Yes)", as processing for the user's landing call Xg, the process from step S432 is executed.

[0209] After that, the group management control device 4 proceeds to step S561 and determines whether the user's car call Yg is further included in the response target in this third response process.

[0210] When the group management control device 4 determines in step S561 that "included (Yes)", based on the determination result, it is determined that there is a getting-off of the user from the target car Gn at the target floor Fn (Jb2 = "there is a getting-off"). In this case, the group management control device 4 deletes the car call Yg that has finished its role (step S562). After that, the group management control device 4 proceeds to step S700 (see FIG. 13) and starts the learning process. Here, summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "no boarding", Ja2 = "there is a getting-off", Jb1 = "there is a boarding", Jb2 = "there is a getting-off" (boarding and alighting status [C10]).

[0211] In the learning process, the group management control device 4 specifies that the boarding and alighting status of the robot H at the target floor Fn is the "only alighting" status (C10 = robot "only alighting") by discriminating from the results of Ja1 = "no boarding" and Ja2 = "alighting" for the above boarding and alighting status [C10]. Therefore, the group management control device 4 shifts to step S720 and specifies one pattern (N = Nt) from the second patterns Pb(N) based on the results of Jb1 = "boarding", Jb2 = "alighting", and the result of the determination [Jb3]. Specifically, the group management control device 4 specifies that a pattern with Nt = 7 or 8 has occurred according to the result of the determination [Jb3]. After that, the group management control device 4 performs learning of the predicted value TeB(Fn, Nt) using the measured value Tp in step S721.

[0212] On the other hand, when the group management control device 4 determines "not included (No)" in step S561, it determines that there is no alighting of the user from the target car Gn at the target floor Fn based on the determination result (Jb2 = "no alighting"). In this case, the group management control device 4 shifts to step S700 without performing the process of step S562. Here, summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "no boarding", Ja2 = "alighting", Jb1 = "boarding", and Jb2 = "no alighting" (boarding and alighting status [C11]).

[0213] Regarding this boarding and alighting status [C11] as well, in the learning process, the group management control device 4 specifies that the boarding and alighting status of the robot H at the target floor Fn is the "only alighting" status (C11 = robot "only alighting"). On the other hand, in step S720, the group management control device 4 specifies one pattern (N = Nt) from the second patterns Pb(N) based on the results of Jb1 = "boarding" and Jb2 = "no alighting" and the result of the determination [Jb3]. Therefore, the group management control device 4 specifies that a pattern with Nt = 5 or 6 has occurred according to the result of the determination [Jb3]. After that, the group management control device 4 performs learning of the predicted value TeB(Fn, Nt) using the measured value Tp in step S721.

[0214] <Third Response Processing (Part 3)> While the group management control device 4 determines "included (Yes)" in step S403 of FIG. 16, if it determines "not included (No)" in step S404B, based on these determination results, the response target includes the landing call Xh of robot H, while it determines that the car call Yh of robot H is not included in the response target. In other words, the group management control device 4 determines that there is boarding of robot H to the target car Gn at the target floor Fn, while there is no alighting of robot H from the target car Gn at the target floor Fn (Ja1 = "boarding exists", Ja2 = "no alighting").

[0215] In this case, the group management control device 4 shifts to process Z1 (refer to the left figure in FIG. 17) and executes the same processes as steps S510 to S513 (including measurement of the fully open holding time) described in the first response processing (steps S540 to S543). Then, the group management control device 4 further determines whether the user's landing call Xg is included in the response target in this third response processing (step S560 in FIG. 16).

[0216] If the group management control device 4 determines "included (Yes)" in step S560, based on this determination result, it determines that there is boarding of the user to the target car Gn at the target floor Fn (Jb1 = "boarding exists"). In this case, the group management control device 4 executes the processes of steps S432 to S434 (deletion of the landing call Xg to registration of the car call Yg) as the process for the landing call Xg.

[0217] After that, the group management control device 4 proceeds to step S561 and further determines whether or not the user's car call Yg is included in the response target in this third response process. Here, the third response process is a process executed when both the landing call X and the car call Y are included in the response target. Therefore, at step S561 at this time, which has reached through the determination that the car call Yh of the robot H is "not included (No)", the group management control device 4 will surely determine that it is "included (Yes)". In other words, the group management control device 4 will surely determine that there is a user getting off from the target car Gn at the target floor Fn (Jb2 = "there is a getting off"). Then, the group management control device 4 deletes the car call Yg that has finished its role (step S562).

[0218] After that, the group management control device 4 proceeds to step S700 (see FIG. 13) and starts the learning process. Here, summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "there is a boarding", Ja2 = "there is no alighting", Jb1 = "there is a boarding", Jb2 = "there is a getting off" (boarding and alighting status [C12]).

[0219] In the learning process, the group management control device 4 specifies that the boarding and alighting status of the robot H at the target floor Fn was a "boarding only" status (C12 = robot "boarding only") by discriminating from the results of Ja1 = "there is a boarding" and Ja2 = "there is no alighting" for the above boarding and alighting status [C12]. Therefore, the group management control device 4 proceeds to step S710 and specifies one pattern (M = Mt) from among the first patterns Pa(M) based on the results of Jb1 = "there is a boarding", Jb2 = "there is a getting off", and the result of the determination [Jb3]. Specifically, the group management control device 4 specifies that a pattern where Mt = 7 or 8 has occurred according to the result of the determination [Jb3]. After that, the group management control device 4 performs learning of the predicted value TeA(Fn, Mt) using the measured value Tp at step S711.

[0220] On the other hand, when the group management control device 4 determines "not included (No)" in step S560, based on the determination result, it determines that there is no user boarding the target car Gn at the target floor Fn (Jb1 = "no boarding"). In this case, the group management control device 4 skips the processes of steps S432 to S434 and proceeds to step S561. Even at step S561 at this time, the group management control device 4 always determines "included (Yes)" (Jb2 = "there is alighting"), and deletes the car call Yg that has finished its role (step S562).

[0221] After that, the group management control device 4 proceeds to step S700 (see FIG. 13) and starts the learning process. Here, summarizing the determination results regarding the boarding and alighting status obtained up to this point, the results are Ja1 = "there is boarding", Ja2 = "no alighting", Jb1 = "no boarding", and Jb2 = "there is alighting" (boarding and alighting status [C13]).

[0222] Also for this boarding and alighting status [C13], in the learning process, the group management control device 4 will specify that the boarding and alighting status of the robot H at the target floor Fn was a "boarding only" status (C13 = robot "boarding only"). On the other hand, at step S710, the group management control device 4 specifies one pattern (M = Mt) from among the patterns of the first pattern Pa(M) based on the results of Jb1 = "no boarding", Jb2 = "there is alighting", and the result of determination [Jb3]. Therefore, the group management control device 4 will specify that a pattern with Nt = 3 or 4 has occurred according to the result of determination [Jb3]. After that, the group management control device 4 performs learning of the predicted value TeA(Fn, Mt) using the measured value Tp at step S711.

[0223] <The Third Response Process (Part 4)> When the group management control device 4 determines "included (Yes)" in step S403 and further determines "included (Yes)" in step S404B as well, based on these determination results, it determines that the response targets include both the landing call Xh of robot H and another car call Yh of robot H. In other words, the group management control device 4 determines that there are both boarding and alighting of robot H at the target car Gn on the target floor Fn (Ja1 = "boarding exists", Ja2 = "alighting exists"). In this case, the group management control device 4 shifts to process Z2 (refer to the right figure in Fig. 17) and first executes the same processes as steps S540 and S541 of process Z1 (steps S550 and S551).

[0224] In this embodiment, when two robots H perform boarding and alighting on the same floor in this way, in order to smoothly perform the boarding and alighting of these two robots H, first, the robot H in the target car Gn is commanded to alight, and then the robot H at the landing is commanded to board (refer to Fig. 18).

[0225] Therefore, when the group management control device 4 determines "arrived (Yes)" in step S551 of process Z2, it deletes the car call Yh (the car call Yh of the robot H scheduled to alight on the target floor Fn) that has completed its role with this arrival (step S552). Then, when the door of the target car Gn is fully opened, it starts measuring (actual measurement) the full - open holding time.

[0226] Then, the group management control device 4 first determines whether the alighting of robot H has been completed on the target floor Fn by determining whether it has received an alighting completion signal Sy from the robot management device 3 to notify the completion of alighting of the robot H (step S553). Also, the group management control device 4 repeatedly executes step S553 until it can determine "received (Yes)" in step S553.

[0227] When the group management control device 4 determines "Received (Yes)" in step S553, next, in order to determine whether boarding of another robot H has been completed at the target floor Fn, it determines whether it has received a boarding completion signal Sx for notifying the boarding completion of the robot H from the robot management device 3 (step S554). Further, the group management control device 4 repeatedly executes step S554 until it can determine "Received (Yes)" in step S554.

[0228] When the group management control device 4 determines "Received (Yes)" in step S554, it registers the destination floor Fd indicated by the landing call Xh of the robot H that boarded at the target floor Fn as the car call Yh of that robot H in the target car Gn (step S555). Further, the group management control device 4 deletes the landing call Xh that has finished its role by registering the car call Yh.

[0229] Then, after executing process Z2, the group management control device 4 executes steps S560 to S562. Thereafter, the group management control device 4 shifts to step S700 (see FIG. 13). In this embodiment, when the results regarding the boarding and alighting status are Ja1 = "There is boarding" and Ja2 = "There is alighting", the group management control device 4 ends the process without learning the predicted value Te3 regarding the fully open holding time.

[0230] According to such a learning process of this embodiment, the presence or absence of users and the boarding and alighting status that may occur when the robot H boards are patterned and classified, and for each first pattern Pa(M) obtained thereby, only the situations that can be represented by that first pattern Pa(M) are extracted, and then the predicted value TeA regarding the fully open holding time required when the robot H boards under that situation can be learned. Therefore, it becomes possible to obtain a stable value for each first pattern Pa(M) as the predicted value TeA of the fully open holding time.

[0231] Furthermore, the presence or absence of users and boarding / alighting situations that may occur when the robot H gets off the elevator are also patterned and classified. For each second pattern Pb(N) obtained thereby, only the situations that can be represented by the second pattern Pb(N) are extracted, and then the predicted value TeB of the fully open holding time required when the robot H gets off the elevator under that situation can also be learned. Therefore, it becomes possible to obtain a stable value for each second pattern Pb(N) as the predicted value TeB of the fully open holding time.

[0232] Therefore, in the above-described time calculation process, the arrival prediction time Tc of each car G can be calculated with high prediction accuracy even in an environment where both the user and the robot H use the elevator.

[0233] Furthermore, in the present embodiment, for each floor of the elevator, learning of the predicted value TeA(Fz, M) of the fully open holding time (Fz: variable representing the floor number) is performed in a state where one is corresponded to each first pattern Pa(M), and learning of the predicted value TeB(Fz, N) of the fully open holding time is performed in a state where one is corresponded to each second pattern Pb(N). Therefore, the predicted value TeA(Fz, M) of the fully open holding time corresponding to each first pattern Pa(M) can be learned separately from other floors for each floor of the elevator, and the predicted value TeB(Fz, N) of the fully open holding time corresponding to each second pattern Pb(N) can be learned separately from other floors for each floor of the elevator. As a result, it becomes possible to reflect the changes that may occur due to the difference in the boarding floor Fx of the robot H in the predicted value TeA(Fz, M) of the fully open holding time. Also, it becomes possible to reflect the changes that may occur due to the difference in the getting-off floor Fy of the robot H in the predicted value TeB(Fz, N) of the fully open holding time.

[0234] Therefore, in the present embodiment, the arrival prediction time Tc of each car G can be calculated with higher prediction accuracy.

[0235] [1-2-5] Boarding / Alighting Command Processing Performed by Robot Management Device FIG. 18 is a flowchart showing the boarding / alighting command process executed in this embodiment.

[0236] In this embodiment, the robot management device 3 acquires car operation information indicating the operation status of each car G from the group management control device 4 at any time. Specifically, the robot management device 3 requests the group management control device 4 to return the car operation information at that time at any time, and acquires the necessary information by receiving the car operation information returned from the group management control device 4 in response to the request. Then, based on the car operation information thus acquired, the robot management device 3 grasps the current position and moving direction of each car G (including the departure direction Kx from the stop floor and the arrival direction Ky to the stop floor). Therefore, the robot management device 3 can determine whether each car G has arrived at the stop floor, and when it can be determined that any car G has "arrived", it can specify the floor number of the arrival floor at that time, the arrival direction Ky to that arrival floor, and the next departure direction Kx from that arrival floor.

[0237] Therefore, the robot management device 3 determines for each car G whether the car G has arrived at the stop floor, and when it can be determined that it has "arrived", in order to appropriately execute boarding or alighting of the robot H in the car G (here, this car G will be referred to as the "target car Gn"), after specifying the floor number of the arrival floor at that time, the arrival direction Ky to that arrival floor, and the next departure direction Kx from that arrival floor, it starts the boarding / alighting command process in FIG. 18.

[0238] When the boarding / alighting command process is started, the robot management device 3 first determines whether the stop of the target car Gn at the arrival floor may correspond to the stop in response to the car call Yh of the robot H (the stop for getting the robot H off) by determining whether the arrival floor of the target car Gn matches any of the destination floors Fd recorded in the allocation request management data Ds (see FIG. 2(B)) (step S601).

[0239] When the robot management device 3 determines "match (Yes)" in step S601, it further makes a determination as to whether the stop of the target car Gn at the arrival floor is a stop in response to the car call Yh of the robot H, and performs processing (such as a getting-off command to the robot H) according to the determination result. To do so, it executes a getting-off command process (steps S610 to S616). Thereafter, the robot management device 3 proceeds to the process of step S602. The details of the getting-off command process will be described later.

[0240] On the other hand, when the robot management device 3 determines "does not match (No)" in step S601, it proceeds to step S602 without performing the getting-off command process.

[0241] In step S602, the robot management device 3 determines whether the stop of the target car Gn at the arrival floor may correspond to a stop in response to the boarding call Xh of the robot H (a stop for boarding the robot H). To do so, it determines whether the arrival floor of the target car Gn matches any of the departure floors Fc recorded in the allocation request management data Ds (Fig. 2(B)) (in other words, the departure floor Fc transmitted in the allocation request).

[0242] When the robot management device 3 determines "match (Yes)" in step S602, it makes a determination as to whether the stop of the target car Gn at the arrival floor is a stop in response to the boarding call Xh of the robot H, and performs processing (such as a boarding command to the robot H) according to the determination result. To do so, it executes a boarding command process (steps S620 to S624). Thereafter, the robot management device 3 terminates the boarding and alighting command process. The details of the boarding command process will be described later.

[0243] In the boarding / alighting command process of FIG. 18, by enabling the alighting command process to be executed before the boarding command process in this way, when it becomes necessary to have two robots H perform boarding and alighting respectively on the same floor, the robot H within the target car Gn can be made to alight first, and then the robot H at the landing can be made to board. Therefore, according to the boarding / alighting command process of FIG. 18, the boarding and alighting of those two robots H can be performed smoothly.

[0244] When the robot management device 3 determines "No (not matching)" in step S602, it ends the boarding / alighting command process without performing the boarding command process. Here, when the robot management device 3 determines "No (not matching)" in step S601 and also determines "No (not matching)" in step S602, based on those determinations, it can be determined that the stop of the target car Gn at the arrival floor is neither a stop in response to the landing call Xh of the robot H nor a stop in response to the car call Yh of the robot H. In other words, it is a stop in response to the call of the user (landing call Xg or car call Yg). In this case, since the robot management device 3 does not require any processing for the robot H, it ends the boarding / alighting command process without performing either the boarding command process or the alighting command process.

[0245] <Alighting command process> In the alighting command process, first, the robot management device 3 designates the allocation request within the allocation request management data Ds that could be determined as "Yes (matching)" in step S601 as the first target request, and uses the departure floor Fc and the destination floor Fd transmitted with that first target request (specifically, the departure floor Fc and the destination floor Fd corresponding to the first target request in the allocation request management data Ds) to specify the direction from the departure floor Fc to the destination floor Fd as the conveyance direction Kh of the robot H (step S610).

[0246] Next, the robot management device 3 determines whether the stop of the target car Gn at the arrival floor is a stop in response to the car call Yh of the robot H by determining whether the arrival direction Ky of the target car Gn at the arrival floor matches the conveyance direction Kh of the robot H specified in step S610 (step S611).

[0247] If the robot management device 3 determines "match (Yes)" in step S611, based on this determination, it can be specified that the stop of the target car Gn at the arrival floor is a stop in response to the car call Yh of the robot H.

[0248] On the other hand, if the robot management device 3 determines "not match (No)" in step S611, based on this determination, it can be determined that the stop of the target car Gn at the arrival floor is not a stop in response to the car call Yh of the robot H. In other words, it is a stop in response to the landing call Xg of the user, the car call Yg, or the landing call Xh of the robot H. In this case, since it is not necessary for the robot management device 3 to let the robot H get off at the arrival floor, the getting-off command process is terminated and the process proceeds to step S602.

[0249] In this way, by the robot management device 3 making the determinations in steps S601 and S611, the robot management device 3 itself can specify that the target car Gn has arrived at the destination floor Fd in response to the car call Yh of the robot H without notification from the group management control device 4.

[0250] If the robot management device 3 determines "match (Yes)" in step S611, to let the robot H get off from the target car Gn, it specifies the robot H to be made to get off with the robot information Ph in the allocation request management data Ds (see Fig. 2(B)) corresponding to the first target request, and commands the specified robot H (target robot Hk) to get off from the target car Gn (step S612).

[0251] As a result, the target robot Hk starts getting off the target car Gn in response to a command from the robot management device 3, and when the getting-off is completed, notifies the robot management device 3 of the completion of getting off. Therefore, after step S612, the robot management device 3 determines whether it has received a notification of the completion of getting off from the target robot Hk, thereby determining whether the getting off of the target robot Hk from the target car Gn has been completed (step S613). Also, the robot management device 3 repeatedly executes step S613 until it can determine "completed (Yes)" in step S613.

[0252] When the robot management device 3 determines "completed (Yes)" in step S613, it transmits a getting-off completion signal Sy for notifying the completion of getting off of the target robot Hk to the group management control device 4 together with the car information Pg of the target car Gn and the robot information Ph of the target robot Hk (step S614).

[0253] Then, the robot management device 3 updates the boarding floor Fx recorded in the robot management data Dp for the target robot Hk with the floor of the destination (getting-off floor Fy) associated with that boarding floor Fx, and erases the recorded getting-off floor Fy from that destination (step S615). Further, the robot management device 3 erases the information (robot information Ph, departure floor Fc, destination floor Fd) about the first target request that has finished its role due to the getting off of the robot H from the allocation request management data Ds (step S616). After that, the robot management device 3 ends the getting-off command process and proceeds to step S602.

[0254] <Boarding Command Process> In the boarding command process, the robot management device 3 first designates, as a second target request, the allocation request in the allocation request management data Ds that it was able to determine "match (Yes)" in step S602, and uses the departure floor Fc and destination floor Fd transmitted in that second target request (specifically, the departure floor Fc and destination floor Fd corresponding to that second target request in the allocation request management data Ds) to specify the conveyance direction Kh of the robot H as the direction from the departure floor Fc to the destination floor Fd (step S620).

[0255] Next, in order for the robot management device 3 to identify whether the stop of the target car Gn at the arrival floor is a stop in response to the landing call Xh of the robot H, the robot management device 3 determines whether the departure direction Kx of the target car Gn from the arrival floor matches the conveyance direction Kh of the robot H specified in step S620 (step S621).

[0256] When the robot management device 3 determines "match (Yes)" in step S621, based on this determination, it can be specified that the stop of the target car Gn at the arrival floor is a stop in response to the landing call Xh of the robot H.

[0257] On the other hand, when the robot management device 3 determines "not match (No)" in step S621, based on this determination, it can be determined that the stop of the target car Gn at the arrival floor is not a stop in response to the landing call Xh of the robot H, in other words, it is a stop in response to a call from a user (landing call Xg or car call Yg). In this case, since it is not necessary for the robot management device 3 to let the robot H board at the arrival floor, the boarding / alighting command process is terminated.

[0258] In this way, by the robot management device 3 making the determinations in steps S602 and S621, the robot management device 3 itself can identify, without notification from the group management control device 4, that the target car Gn has arrived at the departure floor Fc in response to the landing call Xh of the robot H.

[0259] When the robot management device 3 determines "match (Yes)" in step S621, in order to let the robot H board the target car Gn, the robot H to be boarded is specified by the robot information Ph in the allocation request management data Ds (see Fig. 2(B)) corresponding to the second target request, and the robot H (target robot Hk) is commanded to board the target car Gn (step S622).

[0260] As a result, the target robot Hk starts boarding the target car Gn in response to a command from the robot management device 3, and when the boarding is completed, notifies the robot management device 3 of the completion of boarding. Therefore, after step S622, the robot management device 3 determines whether the boarding of the target robot Hk on the target car Gn is completed by determining whether it has received a notification of the completion of boarding from the target robot Hk (step S623). Also, the robot management device 3 repeatedly executes step S623 until it can determine "completed (Yes)" in step S623.

[0261] When the robot management device 3 determines "completed (Yes)" in step S623, it transmits a boarding completion signal Sx for notifying the completion of boarding of the target robot Hk to the group management control device 4 together with the car information Pg of the target car Gn and the robot information Ph of the target robot Hk (step S624). After that, the robot management device 3 ends the boarding and alighting command process.

[0262] According to the control process of this embodiment as described above, the robot management device 3 itself can determine whether it is necessary to board and alight the robot H at the arrival floor of each car G without receiving a notification from the group management control device 4 (in other words, autonomously), and can also transmit commands and signals according to the determination to the robot H and the group management control device 4.

[0263] [2] Modification [2-1] First Modification There are various types of robots H corresponding to operations such as cleaning, monitoring, and transportation, and the time required for the robot H to board and alight may vary depending on the type of the robot H. In such a case, the fully open holding time required when the robot H boards and alights will vary depending on the type of the robot H.

[0264] Therefore, in the above-described embodiment, the first management data DtA (data for managing the predicted value TeA of the fully open holding time required when the robot H gets on the vehicle. Refer to FIG. 5(A)), and the second management data DtB (data for managing the predicted value TeB of the fully open holding time required when the robot H gets off the vehicle. Refer to FIG. 5(B)) may be set for each type of the robot H.

[0265] Then, in step S711 of the learning process (refer to FIG. 13), the group management control device 4 may perform learning of the predicted value TeA(Fn, Mt) using the measured value Tp for each type of the robot H that gets on at the target floor Fn. Further, in step S721 of the learning process, the group management control device 4 may perform learning of the predicted value TeB(Fn, Nt) using the measured value Tp for each type of the robot H that gets off at the target floor Fn.

[0266] According to the first modification example, by learning the predicted value TeA of the fully open holding time corresponding to each first pattern Pa(M) while distinguishing by the type of the robot H, it becomes possible to reflect the changes that may occur due to the difference in the type of the robot H in the predicted value TeA of the fully open holding time. Further, when there are many types of the robot H, if an appropriate value for each type of the robot H is to be set manually as the predicted value TeA of the fully open holding time corresponding to each first pattern Pa(M), troublesome work will be imposed on the operator. However, according to the first modification example, through learning, the predicted value TeA of the fully open holding time corresponding to each first pattern Pa(M) is updated to an appropriate value for each type of the robot H at any time, so such troublesome work becomes unnecessary. Similarly, for the predicted value TeB of the fully open holding time corresponding to each second pattern Pb(N), it also becomes possible to reflect the changes that may occur due to the difference in the type of the robot H.

[0267] [2-2] Second Modification Example Even if the types of robot H are the same, the time required for robot H to board and alight may vary depending on the state that occurs in the robot H when using the elevator (such as the type and situation of work. Specifically, during luggage transportation, cleaning, etc.). In such a case, the fully open holding time required when robot H boards and alights will vary depending on the state of the robot H.

[0268] Therefore, in any of the above-described embodiments and modifications, the first management data DtA (see FIG. 5(A)) and the second management data DtB (see FIG. 5(B)) may be set for each state that can occur in the robot H.

[0269] Then, in step S711 of the learning process (see FIG. 13), the group management control device 4 may perform learning of the predicted value TeA(Fn, Mt) using the measured value Tp for each boarding state of the robot H that boarded at the target floor Fn (further, it may also be for each type of the robot H). Also, in step S721 of the learning process, the group management control device 4 may perform learning of the predicted value TeB(Fn, Nt) using the measured value Tp for each alighting state of the robot H that alighted at the target floor Fn (further, it may also be for each type of the robot H).

[0270] According to the second modification example, by learning the predicted value TeA of the fully open holding time corresponding to each first pattern Pa(M) while distinguishing the states (such as the type and situation of the work) that can occur in the robot H, it becomes possible to reflect the changes that can occur due to differences in the states of the robot H in the predicted value TeA of the fully open holding time. Further, when there are various states of the robot H, if an appropriate value is to be set manually for each state of the robot H as the predicted value TeA of the fully open holding time corresponding to each first pattern Pa(M), cumbersome work will be imposed on the operator. However, according to the second modification example, through learning, the predicted value TeA of the fully open holding time corresponding to each first pattern Pa(M) is updated to an appropriate value for each state of the robot H as needed, so such cumbersome work becomes unnecessary. Similarly, it becomes possible to reflect the changes that can occur due to differences in the states of the robot H in the predicted value TeB of the fully open holding time corresponding to each second pattern Pb(N).

[0271] [2-3] Third Modification Example In any of the above-described embodiments and modification examples, instead of the predicted value Te3 of the fully open holding time, a predicted value of the time obtained by adding the door opening operation time and the door closing operation time to the fully open holding time may be used.

[0272] In other words, instead of the predicted value TeA of the fully open holding time required when the robot H gets on the elevator, a part or all of the stop time of the car G at each floor that includes at least the fully open holding time, and which is also the predicted value of the time (first target stop time) required when the robot H gets on the elevator, may be used. In this case, in the first management data DtA (see FIG. 5(A)), the predicted value of the first target stop time is associated with each first pattern Pa(M) one by one.

[0273] Alternatively, instead of the predicted value TeB of the fully open holding time required when the robot H gets off the elevator, a part or all of the stop time of the car G at each floor that includes at least the fully open holding time, and the time (second target stop time) required when the robot H gets off the elevator may be used. In this case, in the second management data DtB (see FIG. 5(B)), the predicted values of the second target stop time are associated one by one with the second patterns Pb(N).

[0274] [2-4] Fourth Modification Example In the above-described time calculation process (FIG. 11), when the group management control device 4 determines that "both boarding and alighting of the robot H occur (both boarding and alighting)" at the floor of the variable Fz2 based on the determinations in steps S314 and S315A, the first pattern Pa(Fz2, Mt) is specified in step S340, and the second pattern Pb(Fz2, Nt) is specified in step S341. Then, in step S342, the predicted values TeA(Fz2, Mt) and TeB(Fz2, Nt) of the fully open holding time corresponding thereto are both added to the value of the variable Tz, and the value obtained thereby is set as the new value of the variable Tz.

[0275] Alternatively, the group management control device 4 may add only the larger value of the predicted values TeA(Fz2, Mt) and TeB(Fz2, Nt) to the value of the variable Tz, and set the value obtained thereby as the new value of the variable Tz.

[0276] Or, without specifying the patterns in steps S340 and S341, the group management control device 4 may add both the longest one among the predicted values TeA(Fz2, M) corresponding to the eight first patterns Pa(Fz2, M) and the longest one among the predicted values TeB(Fz2, N) corresponding to the eight second patterns Pb(Fz2, N), or only the smaller value of them to the value of the variable Tz, and set the value obtained thereby as the new value of the variable Tz.

[0277] In this way, by overestimating the predicted value of the fully open holding time required when two robots H board and alight on the same floor respectively, the car G scheduled to stop on such a floor is likely to be excluded because it is extracted as an allocation candidate Gm in steps S211 and S212 of the allocation process (Fig. 10).

[0278] [2-5] Fifth Modification Example In any of the above-described embodiments and modification examples, the first management data DtA (see Fig. 5(A)) may be appropriately changed to a form in which the predicted value TeA of the fully open holding time is associated one by one with the first pattern Pa(M) without depending on the floor of the elevator. Also, the second management data DtB (see Fig. 5(B)) may be appropriately changed to a form in which the predicted value TeB of the fully open holding time is associated one by one with the second pattern Pb(N) without depending on the floor of the elevator.

[0279] [2-6] Sixth Modification Example In any of the above-described embodiments and modification examples, the first pattern Pa(M) is not limited to the eight boarding and alighting patterns shown in Figs. 6(A) to 6(H), and only a part of them may be used, or all or a part of the combinations of these eight boarding and alighting patterns with other boarding and alighting patterns may be used. Also, the second pattern Pb(N) is not limited to the eight boarding and alighting patterns shown in Figs. 7(A) to 7(H), and only a part of them may be used, or all or a part of the combinations of these eight boarding and alighting patterns with other boarding and alighting patterns may be used.

[0280] [2-7] Seventh Modification Example Any of the above-described embodiments and modification examples may be appropriately changed to those that do not execute the learning process (see Fig. 13), in other words, those that perform the time calculation process (see Fig. 11) using fixed values preset as the predicted values Te1 to Te4 of various times (travel time, door opening operation time, fully open holding time, door closing operation time).

[0281] [2-8] Eighth Modification Example In any of the above-described embodiments and modified examples, the request for allocating the landing call Xg for the user may be appropriately changed to be requested from the destination floor registration device installed on each floor when the user registers the destination floor Fd with the destination floor registration device. In this case, the destination floor registration device transmits the destination floor Fd registered by the user and its own device information Pd1 to the group management control device 4 as a request for allocating the landing call Xg for the user.

[0282] And when the group management control device 4 receives the allocation request from the destination floor registration device, it determines that it is "device information Pd1" in step S200 of FIG. 10, and in the subsequent step S210, with the installation floor Fs of the destination floor registration device (the destination floor registration device specified by the received device information Pd1; see FIG. 3(A)) as the departure floor Fc, it executes the allocation to the allocation candidate Gm with the departure floor Fc and the received destination floor Fd (the destination floor Fd registered by the user with the destination floor registration device) as one landing call Xg.

[0283] In this case, when responding to the user's landing call Xg in the first response process of FIG. 14, instead of steps S412 to 414, the group management control device 4 registers the destination floor Fd (the user's destination floor Fd) indicated by the landing call Xg as the user's car call Yg in the car G, and deletes the landing call Xg that has thus completed its role. Further, when responding to the user's landing call Xg in the third response process of FIG. 16, the group management control device 4 also performs the same process instead of steps S432 to S434.

[0284] [2-9] Ninth Modified Example In any of the above-described embodiments and modified examples, each robot H may be appropriately modified to execute the control processes (including the allocation request process and the boarding / alighting command process) performed by the robot management device 3 instead of the robot management device 3. In this case, each robot H communicates with the group management control device 4 without going through the robot management device 3. As a result, each robot H can autonomously use the elevator car G.

[0285] The descriptions of the above embodiments and modifications should be considered as illustrative in all respects and not restrictive. The scope of the present invention is indicated not by the above embodiments or modifications, but by the claims. Furthermore, it is intended that the scope of the present invention includes all changes within the meaning and scope equivalent to the claims.

[0286] Also, from the above embodiments and modifications, as the subject of the invention, it is not limited to the control device (the group management control device 4 in the above embodiments and modifications), but the control process (including the control method) and program executed by the control device may be individually extracted, or a part of them may be partially extracted. Also, as the subject of the invention, the robot management device 3, the robot H, and further, the control process (including the control method) and program executed by them may be individually extracted, or a part of them may be partially extracted. Furthermore, as the subject of the invention, a part or all of the elevator including the control device, the robot management device 3, and the robot H may be extracted.

Explanation of Reference Numerals

[0287] 1 First operation unit 2 Second operation unit 3 Robot management device 4 Group management control device G Car H Robot X Landing call Y Car call 31, 41 Storage unit 32, 42 Control unit Dp Robot management data Dq Device management data Dr Route management data Ds Assignment request management data Dt Prediction value management data Dx Landing call management data Dy Car call management data Fc Departure floor Fd Destination floor Fk Scheduled stop floor Fn, Fw Target floor Fp Current floor Fs setting floor Fx boarding floor Fy alighting floor Gk target car Gm allocation candidate Gn target car Hk target robot Kc destination direction Kh conveyance direction Kn, Kw target direction Kx, Kxp departure direction Ky arrival direction Pa first pattern Pb second pattern Pe elevator information Pg car information Ph robot information Rz predetermined section Sx boarding completion signal Sy alighting completion signal Tc estimated arrival time Tp measured value Xg, Xh landing call Yg, Yh car call Dq1, Dq2 device management data Dt1, Dt2, Dt3, Dt4 predicted value management data DtA first management data DtB second management data DtC third management data DxG, DxH landing call management data DyG, DyH car call management data Fk1 first stop floor Fk2 second stop floor Pd1, Pd2 device information Pr1, Pr2 received information Te1, Te2, Te3, Te4 predicted values TeA, TeB, TeC predicted values

Claims

1. The control method for an elevator is characterized in that a first target stop time is a part or all of a stop time of a car at each floor, the part including at least a fully open hold time during which the door of the car is held fully open, and is a time required for a robot to board the car, and the method learns a predicted value for the first target stop time, all or a part of possible boarding and alighting patterns of users when the robot is boarded are defined as first patterns, and the predicted values ​​of the first target stopping times are made to correspond to the first patterns one by one; The elevator control method includes, each time the car stops at a boarding floor of the robot, a process for identifying which of the first patterns the boarding / alighting pattern occurring at that boarding floor matches, and learning a predicted value of the first target stop time corresponding to the first pattern identified by that process, using the time actually required at that boarding floor.

2. The predicted value of the first target stop time is associated with the first pattern one by one for each elevator floor; 2. The elevator control method according to claim 1, further comprising: performing a process for identifying which of the first patterns the boarding / alighting pattern occurring at the boarding floor matches each time the car stops at the boarding floor of the robot; and learning, using the time actually required at the boarding floor, the predicted value of the first target stop time associated with the boarding floor that corresponds to the first pattern identified by the process.

3. the predicted value of the first target stopping time is associated with the first pattern one by one for each type of the robot; 2. The elevator control method of claim 1, further comprising: each time the car stops at a floor where the robot is to board, a process is performed to identify which of the first patterns the boarding / alighting pattern occurring at that floor matches; and learning, using the time actually required at that floor, the predicted values ​​of the first target stop times associated with the type of robot boarded at that floor that correspond to the first pattern identified by the process.

4. a predicted value of the first target stopping time is associated with each of the first patterns for each state that may occur in the robot; 2. The elevator control method of claim 1, further comprising: each time the car stops at a boarding floor for the robot, a process is performed to identify which of the first patterns the boarding / alighting pattern occurring at that boarding floor matches; and learning, using the time actually required at that boarding floor, the predicted values ​​of the first target stop times associated with the boarding state of the robot that boarded at that boarding floor that correspond to the first pattern identified by the process.

5. Among the boarding and alighting patterns that may occur for a user when the robot is boarded, (1) There are no users other than the robot. (2) A pattern in which the user waits in the elevator; (3) A pattern in which a passenger gets off the car; (4) A pattern in which a user waits in the elevator and another user gets off from the elevator; (5) A pattern in which a user gets on the elevator; (6) A pattern in which a user waits in the elevator and another user boards the elevator; (7) A pattern in which a user gets off the elevator and another user gets on the elevator; (8) A pattern in which a user waits in the elevator, another user gets off the elevator, and yet another user gets on the elevator; The elevator control method according to any one of claims 1 to 4, wherein the first pattern is:

6. The control method for an elevator is characterized in that a second target stop time is a part or all of a stop time of a car at each floor, the part including at least a fully open hold time during which the door of the car is held fully open, and is a time required for a robot to disembark, and the method learns a predicted value for the second target stop time, all or a part of the boarding and alighting patterns that may occur for the user when the robot alights are defined as second patterns, and the predicted values ​​of the second target stopping times are made to correspond to the second patterns one by one; The elevator control method includes, each time the car stops at the robot's disembarking floor, a process for identifying which of the second patterns the boarding and alighting pattern occurring at that disembarking floor matches, and learning a predicted value of the second target stop time corresponding to the second pattern identified by that process using the time actually required at that disembarking floor.

7. Among the boarding and alighting patterns that may occur for a user when the robot gets off, (1) There are no users other than the robot. (2) A pattern in which the user waits in the elevator; (3) A pattern in which a passenger gets off the car; (4) A pattern in which a user waits in the elevator and another user gets off from the elevator; (5) A pattern in which a user gets on the elevator; (6) A pattern in which a user waits in the elevator and another user boards the elevator; (7) A pattern in which a user gets off the elevator and another user gets on the elevator; (8) A pattern in which a user waits in the elevator, another user gets off the elevator, and yet another user gets on the elevator; The elevator control method according to claim 6, wherein the second pattern is:

8. a control device for learning a predicted value of a first target stop time, the first target stop time being a part or all of a stop time of a car at each floor, the part including at least a fully open hold time during which the door of the car is held fully open, and being a time required for a robot to board the car; all or a part of possible boarding and alighting patterns of users when the robot is boarded are defined as first patterns, and the predicted values ​​of the first target stopping times are made to correspond to the first patterns one by one; The elevator control device performs a process to identify which of the first patterns the boarding / alighting pattern occurring at the robot's boarding floor matches each time the car stops at that boarding floor, and learns a predicted value of the first target stop time corresponding to the first pattern identified by the process using the time actually required at that boarding floor.

9. a control device for learning a predicted value of a second target stop time, the second target stop time being a part or all of a stop time of a car at each floor that includes at least a fully open hold time during which the door of the car is held fully open, and being a time required for a robot to disembark; all or a part of the boarding and alighting patterns that may occur for the user when the robot alights are defined as second patterns, and the predicted values ​​of the second target stopping times are made to correspond to the second patterns one by one; The elevator control device performs a process to identify which of the second patterns the boarding and alighting pattern occurring at the robot's disembarking floor matches each time the car stops at the robot's disembarking floor, and learns a predicted value of the second target stop time corresponding to the second pattern identified by the process using the time actually required at the disembarking floor.

10. a program for causing a control device to execute the following: a first target stop time is a part or all of a stop time of a car at each floor, the part including at least a fully open hold time during which the door of the car is held fully open, and is a time required for a robot to board the car; and learning a predicted value for the first target stop time, defining all or some of the boarding and alighting patterns that may occur for a user when the robot is boarded as first patterns, and corresponding the predicted values ​​of the first target stopping times to the first patterns one by one; The program executes the following: each time the elevator stops at the robot's boarding floor, a process is performed to identify which of the first patterns the boarding / alighting pattern occurring at that boarding floor matches, and a predicted value of the first target stop time corresponding to the first pattern identified by that process is learned using the time actually required at that boarding floor.

11. a program for causing a control device to execute the following: a second target stop time is a part or all of a stop time of a car at each floor, the part including at least a fully open hold time during which the door of the car is held fully open, and is a time required for a robot to get off the car; and learning a predicted value of the second target stop time, defining all or some of the boarding and alighting patterns that may occur for the user when the robot alights as second patterns, and corresponding the predicted values ​​of the second target stopping times to the second patterns one by one; The program executes the following: each time the elevator stops at the robot's disembarking floor, a process is performed to identify which of the second patterns the boarding and alighting pattern occurring at that disembarking floor matches, and a predicted value of the second target stop time corresponding to the second pattern identified by that process is learned using the time actually required at that disembarking floor.

Images