Group control system and method
The group control system addresses positional misalignment issues by using a deep reinforcement learning model to integrate acceleration data, ensuring accurate position information and preventing collisions among AMRs, thus enhancing space efficiency and reducing costs.
Patent Information
- Application Number
- JP2023087172
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-26
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2043-05-26
Smart Images

Figure 2024170162000001_ABST
Abstract
Description
[Technical field]
[0001] The present disclosure relates to group control systems and methods. [Background technology]
[0002] In a group control system for mobile objects that issues control commands to multiple mobile objects capable of traveling autonomously within a specified area, the position information of each mobile object is estimated, and a route course for each mobile object is created based on the estimated position information so that the mobile objects do not interfere with each other (Patent Document 1). [Prior art documents] [Patent documents]
[0003] [Patent Document 1] JP 2017-142778 A Summary of the Invention [Problem to be solved by the invention]
[0004] The above prior art does not take into account the acceleration and deceleration of the moving body. Therefore, due to the acceleration and deceleration of the moving body, the estimated position of the moving body and the actual position of the moving body are misaligned due to the motion behavior of the moving body, and group control based on correct position information cannot be performed. If multiple AMRs (Autonomous Mobile Robots) are introduced in a narrow space where the effects of acceleration and deceleration are significant, there is a risk that they will come into contact with each other, and layout changes may be required, which may ultimately lead to increased costs.
[0005] The present disclosure has been made to solve such problems, and aims to provide a group control system and a group control method that perform machine learning that takes into account acceleration and deceleration data, and is capable of performing group control based on correct position information. [Means for solving the problem]
[0006] A group control system according to one aspect of the present disclosure includes: In a group control system for controlling a plurality of autonomously moving objects, at least one memory having instructions stored therein; at least one controller configured to execute the instructions; An acceleration sensor and a position sensor attached to each moving object; Equipped with The control unit is A position information estimation unit that estimates position information of each moving object; a route planning unit that creates a route plan for each moving object based on the estimated position information; an acceleration data acquisition unit that acquires acceleration data acquired from the acceleration sensor; a moving object position acquisition unit that acquires an actual position of each moving object using the position sensor; Equipped with The deep reinforcement learning model is configured to train so as to correct the amount of deviation between the actual position of the moving body acquired using the position sensor and the estimated position of the moving body based on the acquired acceleration data.
[0007] A group control method according to one aspect of the present disclosure includes: A group control method for controlling a plurality of autonomously traveling mobile objects, comprising: Estimating location information of each moving object; creating a route plan for each moving object based on the estimated position information; Obtaining acceleration data obtained from an acceleration sensor; Obtaining an actual position of each moving object using a position sensor; Including, A deep reinforcement learning model is trained so as to correct the amount of deviation between the actual position of the moving body obtained using the position sensor and the estimated position of the moving body based on the obtained acceleration data. Effect of the Invention
[0008] According to the present disclosure, it is possible to provide a group control system and a group control method capable of performing group control based on correct position information by taking into account acceleration / deceleration data. [Brief description of the drawings]
[0009] [Figure 1] FIG. 1 is a diagram for explaining the state of AMR during a trial run. [Diagram 2] This shows the position of the AMR recognized by the group control system and the actual position of the AMR. [Diagram 3] 1 is a graph showing velocity versus time in several cases; [Figure 4] 1 is a schematic diagram showing a configuration of a group control system according to an embodiment. [Diagram 5] 4 is a flowchart showing a group control method according to the embodiment. [Figure 6] FIG. 2 is a block diagram illustrating an example of a hardware configuration of the group control system. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0010] Hereinafter, specific embodiments to which the present invention is applied will be described in detail with reference to the drawings. However, the present invention is not limited to the following embodiments. In addition, the following description and drawings are appropriately simplified for clarity of explanation.
[0011] Systems have been developed in which multiple AMRs (Autonomous Mobile Robots) travel autonomously and transport goods to assist and replace human work in factories and other locations. Basically, each AMR can estimate its own position individually and travel autonomously within a given area according to a route plan. However, as shown in Figure 1, when a running trial was conducted with an actual AMR, if multiple AMRs travel in a narrower area, the AMRs may come into contact with each other. Therefore, a group control device for multiple AMRs has been introduced to prevent the AMRs from colliding or interfering with each other.
[0012] Figure 2 shows the position of the AMR recognized by the group control system (RP_10) and the actual position of the AMR (AP_10). The position of the AMR recognized by the group control system is the position where the AMR is estimated to be at a certain time when a destination (goal) is set for the AMR and the AMR is moved according to a route plan to the goal. On the other hand, the actual position of the AMR is the actual position of the AMR determined using a position sensor (e.g., LiDAR).
[0013] As shown in Figure 2, there is a slight deviation between the AMR's position (RP_10) recognized by the group control system and the AMR's actual position (AP_10). This is thought to be because the effect of acceleration becomes significant when moving an AMR in a small area. Similar position deviations occurred when similar trials were performed using actual AMRs with various different shapes and sizes.
[0014] Figure 3 is a graph showing the relationship between speed and time for several AMR driving scenarios. Acceleration is the rate of change of velocity per unit time, so Figure 3 also shows acceleration. In the case of A1, the moving body accelerates from a stopped state, then moves at a constant speed, then decelerates, and stops. In a relatively wide area, the moving body accelerates and decelerates as shown in A1. In the case of A2, the moving body accelerates from a stopped state, then decelerates and stops before reaching a constant speed state. The above-mentioned deviation between the position of the AMR recognized by the group control system and the actual position of the AMR is more noticeable in the case of A2, which requires acceleration and deceleration in a narrow area. Even in the case of A1, if the constant speed state is relatively short, position deviation may occur due to the influence of acceleration and deceleration. In general, position deviation due to the influence of acceleration and deceleration is more noticeable in a narrow layout than in a wide layout.
[0015] In this embodiment, a group control system and a group control method are provided that can perform group control based on correct position information by taking into account the acceleration of the moving object. More specifically, a deep reinforcement learning model is used to correct the position deviation of the AMR based on the acquired acceleration of the AMR.
[0016] FIG. 4 is a schematic diagram showing the configuration of a group control system according to this embodiment. Each AMR managed by the group control system is equipped with an acceleration sensor (e.g., a gyro sensor, an IMU (Inertial Measurement Unit)). Acceleration and deceleration data at each time is acquired from the acceleration sensor equipped in each AMR. The acquired acceleration and deceleration data is input to an input layer 201 of the deep reinforcement learning model 20. The input layer 201 is also called an acceleration data acquisition unit. In the example of FIG. 3, for example, acceleration when accelerating from a stopped state, acceleration in a constant speed state, deceleration from deceleration to a stopped state, etc. may be input at predetermined intervals. In addition, the position of the AMR recognized by the group control system and the actual position of the AMR are also input to the input layer 201 of the deep reinforcement learning model 20. In this example, the deep reinforcement learning model uses Deep Q-Network, but is not limited to this, and various neural network models that can be understood by a person skilled in the art can be used.
[0017] The deep reinforcement learning model 20 learns to correct the amount of deviation between the actual position of the AMR and the position of the AMR recognized by the group control system (i.e., to make the amount of deviation zero). In addition, the deep reinforcement learning model learns to prevent the actual positions of the AMRs from overlapping (i.e., to prevent the AMRs from contacting each other) as a reward. In this way, the deep reinforcement learning model is trained based on a predetermined amount of input data by giving several rewards, and a trained model is generated.
[0018] An autonomous driving control device 100 is provided corresponding to each moving body 10. The autonomous driving control device 100 is composed of a computer having a processor such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) and a memory. The autonomous driving control device 100 controls the autonomous driving of the moving body 10 to a specified destination. Each autonomous driving control device 100 includes a self-position estimation unit 101 and a route planning unit 102. The self-position estimation unit 101 estimates the position of the moving body for each time when the moving body moves to the destination according to the route plan. The route planning unit 102 sets a destination for each moving body and creates a route plan from the current position to the destination. In FIG. 4, two autonomous driving control devices 100 are provided corresponding to two moving bodies, respectively, but three or more autonomous driving control devices 100 may be provided corresponding to three or more moving bodies, respectively.
[0019] A group control device 300 is provided behind the autonomous driving control device 100 to control the operation plan so that each moving lane does not collide or interfere with each other. The group control device 300 may be, for example, a server computer equipped with a processor such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) and a memory. The group control device 300 includes an optimization unit 301, a task management allocation unit 302, and an operation management unit 303. The optimization unit 301 optimizes the route plan and destination from each autonomous driving control device 100 using the above-mentioned deep reinforcement learning model that takes into account the acceleration data. The task management allocation unit 302 receives the output from the above-mentioned deep reinforcement learning model 20 and allocates tasks such as the optimized route plan and destination to each moving body. The operation management unit 303 manages the operation so that each moving body does not collide.
[0020] The task management allocation unit 302 assigns tasks and route information to each moving body (i.e., AMR) so that the AMRs do not collide with each other. Each AMR moves according to the assigned task and route information, and at that time, the acceleration and deceleration values of each AMR are obtained from each acceleration sensor. In this way, the deep reinforcement learning model is re-learned using the acquired acceleration and deceleration values. When a change in the environment occurs (for example, a change in the layout, an increase in the number of AMRs, the occurrence of an obstacle, etc.), the AMRs may collide with each other. Therefore, such re-learning can be effective in that it can self-correct without human intervention. Instructions are given to each AMR based on the results of the re-learning.
[0021] According to the embodiment described above, it is possible to instruct each AMR on tasks and route information that take into account the actual AMR's positional deviation due to acceleration and deceleration and the AMR's positional deviation recognized by the group control system. This makes it possible to utilize multiple AMRs in a small space, and avoids costs associated with changing the layout.
[0022] FIG. 5 is a flowchart showing a group control method according to the embodiment. The acceleration data acquisition unit acquires acceleration and deceleration data from the acceleration sensor 11 (e.g., gyro sensor) mounted on each AMR (step S11). The acquired acceleration and deceleration data is input to the input of the deep reinforcement learning model (step S12). The deep reinforcement learning model is trained to correct the amount of deviation between the actual position of the AMR acquired by the mobile object position acquisition unit using a position sensor (e.g., LiDAR) and the position of the AMR recognized by the group control system (position information estimation unit) (step S13) (to prevent the AMRs from colliding with each other). The group control device assigns task and route information to each AMR so that the AMRs do not collide with each other (step S14). The acceleration and deceleration values of each AMR moving according to the assigned task and route information are acquired, and re-learning is performed from the acquired acceleration and deceleration values (step S15). Instructions are given to each AMR based on the re-learning results (step S16).
[0023] FIG. 6 is a block diagram showing a configuration example of a group control system including an autonomous driving control device 100 and a group control device 300. Referring to FIG. 6, the group control system includes a network interface 1201, a processor 1202, and a memory 1203. The network interface 1201 is used to communicate with other network node devices constituting a communication system. The network interface 1201 may be used to perform wireless communication. For example, the network interface 1201 may be used to perform wireless LAN communication defined in the IEEE 802.11 series, or mobile communication defined in 3GPP (registered trademark) (3rd Generation Partnership Project). Alternatively, the network interface 1201 may include, for example, a network interface card (NIC) conforming to the IEEE 802.3 series. The acceleration sensor 11 and the position sensor 12 described above are connected to the network interface 1201.
[0024] The processor 1202 reads and executes software (computer programs) from the memory 1203 to perform the processing of the group control system described using the flowcharts or sequences in the above-mentioned embodiments. The processor 1202 may be, for example, a microprocessor, an MPU (Micro Processing Unit), a CPU (Central Processing Unit), or a GPU (Graphics Processing Unit). The processor 1202 may include multiple processors.
[0025] The memory 1203 is configured by a combination of volatile memory and non-volatile memory. The memory 1203 may include storage located remotely from the processor 1202. In this case, the processor 1202 may access the memory 1203 via an I / O interface (not shown).
[0026] 6, the memory 1203 is used to store software modules. The processor 1202 reads out and executes these software modules from the memory 1203 to perform the processing of the group control system described in the above embodiment.
[0027] As explained with reference to FIG. 5, each of the processors in the group control system executes one or more programs containing instructions for causing a computer to perform the algorithms explained with reference to the figures.
[0028] In the above examples, the program includes instructions (or software code) that, when loaded into a computer, cause the computer to perform one or more functions described in the embodiments. The program may be stored in a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, the computer-readable medium or tangible storage medium includes random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technology, CD-ROM, digital versatile disc (DVD), Blu-ray® disk or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage device. The program may be transmitted on a transitory computer-readable medium or communication medium. By way of example and not limitation, the transitory computer-readable medium or communication medium includes electrical, optical, acoustic, or other forms of propagated signals.
[0029] The present invention is not limited to the above-described embodiment, and can be modified as appropriate without departing from the spirit and scope of the present invention. For example, in the above-described embodiment, an AMR is described as a moving object, but the present invention can also be applied to a moving object that can move vertically, such as a drone or an airplane. [Explanation of symbols]
[0030] 10 Mobile 11 Acceleration Sensor 12 Position Sensor 20 Deep Reinforcement Learning Models 100 Autonomous driving control device 101 Self-position estimation part 102 Route Planning Department 201 Input Layer 202 Hidden Layer 203 Output Layer 300 Group Control Device 301 Optimization Department 302 Task Management Allocation Department 303 Operation Management Department
Claims
1. In a group control system for controlling a plurality of autonomously moving objects, at least one memory storing instructions; at least one controller configured to execute the instructions; An acceleration sensor and a position sensor attached to each moving object; Equipped with The control unit is A location information estimation unit that estimates location information of each moving object; a route planning unit that creates a route plan for each moving object based on the estimated position information; an acceleration data acquisition unit that acquires acceleration data acquired from the acceleration sensor; a moving object position acquisition unit that acquires an actual position of each moving object using the position sensor; Equipped with A group control system configured to train a deep reinforcement learning model to correct the amount of deviation between the actual position of the moving body acquired using the position sensor and the estimated position of the moving body based on the acquired acceleration data.
2. The group control system according to claim 1 , configured to train the deep reinforcement learning model so that moving objects do not interfere with each other based on the estimated position information and the amount of deviation.
3. 2. The group control system of claim 1, configured to: assign to each moving body a route plan that has been planned using a trained deep reinforcement learning model; move each moving body according to the route plan; acquire the acceleration data obtained from the acceleration sensor during the movement; correct the amount of deviation between the actual position of the moving body acquired using the position sensor and an estimated position of the moving body based on the acquired acceleration data; and re-learn the deep reinforcement learning model so that the moving bodies do not interfere with each other.
4. The group control system according to claim 1 , configured to assign a route plan to each moving object and move each moving object based on the trained deep reinforcement learning model.
5. A group control method for controlling a plurality of autonomously traveling mobile objects, comprising: Estimating location information of each moving object; creating a route plan for each moving object based on the estimated position information; Obtaining acceleration data obtained from an acceleration sensor; Obtaining an actual position of each moving object using a position sensor; Including, A group control method, comprising: training a deep reinforcement learning model so as to correct the amount of deviation between the actual position of the moving body acquired using the position sensor and the estimated position of the moving body based on the acquired acceleration data.
Citation Information
Patent Citations
Vacuum cleaner
JP2017146775A
Operation plan generation device, learning device, operation plan generation method, learning method, and program
JP2022026419A
Group control method of mobile body
JP2017142778A
