A medical robot cross-floor elevator dispatching method, system and device

Abstract

The application discloses a medical robot cross-floor elevator scheduling method, system and device. The method comprises the following steps: receiving a domain application and dividing a physical coordinate grid; calculating a comprehensive traffic weight index to allocate a waiting grid; dynamically defining an asymmetric virtual channel in response to an in-out conflict when the elevator car door is opened; controlling the timing by issuing a token and dynamically interrupting and re-planning when the channel is covered by a repulsive potential field exceeding a set threshold. The application can eliminate physical deadlock when multiple agents cooperate to take the elevator, ensure absolute priority of priority tasks, and improve the flow efficiency of the cross-floor scheduling system and the robustness of dynamic obstacle avoidance.

Description

Technical Field

[0001] This application relates to the field of special traffic control and logistics scheduling, specifically belonging to the category of traffic control systems, and specifically refers to a scheduling method, system, and equipment for medical robots to travel across floors via elevators. This application solves the physical deadlock problem in multi-agent collaborative scenarios through asymmetric channel allocation and regional traffic control. Background Technology

[0002] In large medical institutions, the use of medical robots for tasks such as medication delivery, test sample delivery, and waste collection is becoming increasingly common. Medical robots must use vertical elevators when operating across floors. Elevator lobbies and elevator car entrances are typical traffic bottlenecks with limited space. When multiple robots carrying different priority tasks are waiting on the same floor, or when concurrent operations occur—such as robots inside the elevator car needing to exit while robots outside the lobby need to enter—complex local congestion problems arise.

[0003] Existing medical robot elevator solutions primarily focus on establishing communication handshake protocols between the robot control unit and the elevator's ground floor control board, with their scheduling logic largely confined to a mechanical, single-line mode. This traditional approach lacks a mechanism for right-of-way allocation and global collaborative management within the micro-traffic environment of the elevator lobby. In a physical space where multiple robots coexist, traditional solutions cannot accurately determine which robot should queue where or which should have priority when the doors open. When there is simultaneous movement demand inside and outside the elevator car, spatial physical deadlock can easily occur, leading to widespread robot stagnation. Without dynamic avoidance methods and right-of-way allocation, collisions between robots and medical personnel or medical gurneys are common.

[0004] Furthermore, hospital scenarios involve numerous time-sensitive tasks. Existing systems operate with extremely low overall efficiency when faced with high-frequency scheduling demands, failing to guarantee absolute priority for high-priority tasks such as emergency blood delivery. Due to the lack of refined virtual spatiotemporal grid partitioning and time-slot token control mechanisms, the site descends into a state of disorderly congestion, which not only reduces the efficiency of medical logistics but also poses a potential disruption to the normal medical order of the hospital. There is an urgent need for a scheduling solution that can resolve conflicts and provide dynamic traffic coordination. Summary of the Invention

[0005] The purpose of this application is to provide a scheduling method, system and equipment for medical robots to ride elevators across floors, which solves the problem that the existing robot elevator system lacks multi-agent traffic coordination, resulting in path conflicts, physical deadlocks and delays of high-priority tasks in the elevator lobby area.

[0006] This application provides a scheduling method for medical robots to ride elevators across floors, including: receiving an entry request from a medical robot; dividing a specific distance range of the waiting hall into multiple physical coordinate grids based on a pre-stored high-precision two-dimensional grid map; calculating the comprehensive passage weight index of the medical robot, sorting it in descending order, assigning it a designated physical coordinate grid as a waiting grid, and changing its system state to a locked waiting state; responding to entry and exit conflicts after the elevator car door opens, dynamically defining an asymmetrical one-way exit virtual channel and a one-way entry virtual channel based on the effective width of the car door and the central axis of the elevator door; first issuing an elevator token to the medical robot waiting to exit and retracting it after confirming that it has crossed the safety boundary, and then issuing an entry token to the medical robot waiting to enter; acquiring the locally occupied grid generated by the end-side sensor scan of the medical robot and marking it as the center of the repulsive potential field distribution; when the one-way entry virtual channel is covered by the repulsive potential field for more than a set threshold, withdrawing the current entry token and replanning the path.

[0007] In one optional implementation, the step of calculating the comprehensive access weight index of the medical robot, sorting it in descending order, assigning it a designated physical coordinate grid as a waiting grid, and changing its system state to a locked waiting state includes obtaining multi-dimensional parameters, including the inherent task priority of the medical robot, the waiting time, the distance to the elevator door, and the remaining capacity of the target elevator car; calculating the comprehensive access weight index by weighted summation of the multi-dimensional parameters based on preset weight coefficients; sorting multiple medical robots in descending order according to the comprehensive access weight index, and assigning the physical coordinate grid closest to the elevator door to the corresponding ranked medical robot according to the sorting result.

[0008] In one optional implementation, after the step of obtaining multidimensional parameters, the system is configured to, in response to detecting that the task of the medical robot is emergency rescue or blood sample transportation, set the inherent priority of the task to the highest level, adjust the preset weight coefficient corresponding to the inherent priority of the task to a first preset value, and set the weight coefficient corresponding to the waiting time and the distance from the elevator door to zero; in response to detecting a regular transportation task, determine a first dynamic coefficient based on the ratio of the distance from the elevator door to the standard distance, and determine a second dynamic coefficient based on the ratio of the waiting time to a preset tolerance time; and multiply the first dynamic coefficient and the second dynamic coefficient by their respective preset weight coefficients and then sum them to obtain the final comprehensive passage weight index.

[0009] In one optional implementation, the step of dynamically delineating an asymmetrical one-way exit virtual channel and a one-way entry virtual channel based on the effective width of the elevator door and using the elevator door centerline as a reference includes: acquiring the current opening width data of the elevator car door and identifying the physical coordinates of the elevator door centerline; dividing the current opening width data according to a first ratio and a second ratio; delineating the one-way exit virtual channel on the first side of the elevator door centerline based on the width corresponding to the first ratio; and delineating the one-way entry virtual channel on the second side of the elevator door centerline based on the width corresponding to the second ratio.

[0010] In one optional implementation, the step of acquiring a locally occupied grid generated by the end-side sensors of the medical robot and marking it as the center of the repulsive potential field distribution, and withdrawing the current entry token and replanning the path when the one-way elevator virtual channel is covered by the repulsive potential field for more than a set threshold, is configured as follows: in response to the medical robot moving within the one-way elevator virtual channel, a two-dimensional locally occupied grid map is generated by continuously scanning dynamic obstacles within a set height range using an end-side LiDAR; the coordinates of the obstacle contour center in the two-dimensional locally occupied grid map are extracted as the center of the repulsive potential field distribution, and a corresponding Gaussian repulsive potential field distribution range is generated according to the moving speed of the dynamic obstacle; the overlap area between the Gaussian repulsive potential field distribution range and the one-way elevator virtual channel in the physical coordinate grid is calculated; in response to the overlap area accounting for more than 50% of the total area of ​​the one-way elevator virtual channel, an interrupt command is triggered, the entry token is revoked, a guidance path is generated to allow the medical robot to return to the nearest safe avoidance grid and re-initiate the domain entry request.

[0011] In one alternative implementation, the step of issuing an exit token to the medical robot to exit the elevator and retracting it after confirming that it has crossed a safety boundary includes setting a safety boundary line at the end of the one-way exit virtual channel; generating a confirmation feedback signal in response to the physical centroid of the medical robot carrying the exit token crossing the safety boundary line; and canceling the exit token based on the confirmation feedback signal.

[0012] This application also provides a scheduling system for medical robots to travel across floors via elevators, comprising: an edge computing gateway, physically deployed inside the elevator lobby, for receiving entry requests from medical robots and dividing the lobby into multiple physical coordinate grids based on a pre-stored high-precision two-dimensional grid map; a medical robot, externally equipped with end-side sensors and with an internal main control computing unit connected to the edge computing gateway; the edge computing gateway calculating the comprehensive passage weight index of the medical robots, sorting them in descending order, assigning a designated physical coordinate grid as the elevator waiting grid for the medical robot, and instructing it to enter the locking waiting area. The edge computing gateway dynamically delineates an asymmetrical one-way exit virtual channel and a one-way entry virtual channel based on the effective width of the elevator door and the elevator door centerline after the elevator door opens. The edge computing gateway first issues an elevator token to the medical robot waiting to exit and retracts it after confirming that it has crossed the safety boundary. Then it issues an entry token to the medical robot waiting to enter. The edge computing gateway receives the local occupancy grid generated by the end-side sensor and marks it as the center of the repulsive potential field distribution. When the one-way entry virtual channel is covered by the repulsive potential field for more than a set threshold, the current entry token is withdrawn and the path is replanned.

[0013] In one optional implementation, the edge computing gateway includes an embedded processing unit and a near-field communication antenna array; the near-field communication antenna array is oriented above the elevator door frame to cover a specific distance range of the waiting hall; the embedded processing unit establishes a low-latency dedicated local area network connection with the medical robot located within the physical coordinate grid through the near-field communication antenna array, and receives the domain entry request and issues the elevator entry token through the low-latency dedicated local area network connection.

[0014] In one optional implementation, the edge sensor includes a multi-line LiDAR and a depth camera; the multi-line LiDAR is fixedly installed at the top center of the medical robot and configured to provide 360-degree two-dimensional point cloud data in the horizontal direction; the depth camera is installed at an angle downwards on the front end face of the medical robot and configured to acquire three-dimensional depth information of low dynamic obstacles within a preset height range above the ground; the data from the multi-line LiDAR and the depth camera are fused after time-stamp alignment and spatial registration to generate the local occupancy grid; the main control computing unit compresses and packages the local occupancy grid and continuously uploads it to the edge computing gateway via the low-latency dedicated local area network connection.

[0015] This application also provides an electronic device, including: a memory for storing computer instructions; and a processor communicatively connected to the memory; wherein the processor executes the computer instructions to implement the aforementioned scheduling method for a medical robot to travel across floors in an elevator.

[0016] This application achieves orderly management of the lobby area by introducing a virtual physical coordinate grid for the initial allocation of spatiotemporal right-of-way. At the moment the elevator doors open, asymmetrical entry and exit channels are dynamically defined based on the car door width, and a strict exit and entry time slot token mechanism is used to control the start-up sequence, eliminating congestion and physical deadlock caused by multiple robots moving concurrently at the elevator entrance. Real-time safety interruption and path replanning are performed in conjunction with environmental potential field feedback, improving the system's robustness in complex medical and patient environments. A comprehensive passage weight index based on multi-dimensional parameter considerations ensures the absolute priority passage of emergency medical supply transport robots, improving the efficiency of cross-floor movement of multiple robots without altering the original mechanical structure of the elevator. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall structure of the scheduling system for medical robots to ride elevators across floors, provided in an embodiment of this application.

[0018] Figure 2 This is a flowchart of a scheduling method for medical robots to ride elevators across floors, provided in an embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0020] In one embodiment of this application, a scheduling system and deployment architecture for medical robots to ride elevators across floors are provided. Figure 1 This is a schematic diagram of the overall structure of the medical robot elevator scheduling system provided in this application embodiment. Figure 1 As shown, the system in this application adopts a three-layer physical architecture, including a cloud-based dispatch center, an edge computing gateway deployed inside the elevator lobby, and several medical robots performing specific tasks. Low-latency data transmission links are established between nodes via local area wireless networks or 5G private networks.

[0021] Specifically, the cloud-based dispatch center includes high-performance server equipment, database hardware support environment, and a global order distribution system. The cloud-based dispatch center is responsible for maintaining the operational status mapping table of elevators on each floor, coordinating the global paths of all medical robots within the hospital, and distributing cross-floor tasks to the corresponding medical robots.

[0022] As the core traffic control node of this system, the edge computing gateway is physically deployed in the top or side wall of the elevator lobby on each floor. The edge computing gateway has a built-in high-performance embedded processing unit and a near-field communication antenna array. In this embodiment, the edge computing gateway pre-stores a high-precision two-dimensional grid map of the elevator lobby on that floor to provide an absolute spatial reference coordinate system. The medical robot, as the system's end-side actuator, carries end-side sensors such as multi-line LiDAR and depth cameras, possessing independent synchronous positioning and mapping navigation capabilities. The medical robot's internal main control computing unit establishes a network connection with the edge computing gateway, enabling it to report its current three-dimensional coordinates and task tags to the edge computing gateway in real time. It should be noted that the medical robot's chassis motion control and motor drive module are directly driven by its built-in main control computing unit, responsible for executing the underlying speed and steering control, thereby responding to the time-space scheduling commands from the edge computing gateway.

[0023] Based on the above system architecture, this embodiment provides a scheduling method for medical robots to ride elevators across floors. Figure 2 This is a flowchart of a scheduling method for medical robots to ride elevators across floors, provided in an embodiment of this application. The method uses an edge computing gateway to perform refined right-of-way allocation in the micro-traffic scenario of the elevator lobby. Its main execution process is as follows.

[0024] When the medical robot receives a cross-floor task and moves towards the elevator, it initiates an entry request to the edge computing gateway on the target floor to execute the division of physical coordinate grids. The edge computing gateway receives the entry request from the medical robot and, based on a pre-stored high-precision two-dimensional grid map, divides a specific distance range in the waiting hall into multiple physical coordinate grids. Specifically, the edge computing gateway divides a fan-shaped or rectangular area 1 to 3 meters from the elevator door into multiple physical coordinate grids with sides of 0.5 meters, and assigns a unique identifier to each grid. The 0.5-meter side length is physically determined because this size can precisely encompass the maximum outer diameter space of the medical robot's chassis. By pre-defining the physical grid features within a 1-3 meter range, orderly spatial queuing management of the waiting area is achieved, and a definite start and end point spatial reference is provided for the subsequent generation of dynamic asymmetric channels and the token mechanism.

[0025] After completing the grid division, the edge computing gateway needs to perform waiting grid allocation. This involves calculating the comprehensive access weight index of the medical robots, sorting them in descending order, assigning them designated physical coordinate grids as waiting grids, and changing their system state to locked waiting state. To evaluate the access priority of multiple medical robots, the system obtains multi-dimensional parameters, specifically including the inherent task priority of each medical robot, the waiting time, the distance to the elevator door, and the remaining capacity of the target elevator car. The system constructs a weighted formula by comprehensively considering the positive and negative correlations of the above parameters, and calculates the comprehensive access weight index by weighted summation of the multi-dimensional parameters based on preset weight coefficients.

[0026] In this embodiment, the calculation logic of the comprehensive traffic weight index adopts the following physical expression formula. Specifically, the system performs a weighted summation operation:

[0027]

[0028] Where W represents the comprehensive traffic weight index, which is the final numerical indicator used to evaluate the priority level of each medical robot in the queue. P task This represents the inherent priority weight of the task, reflecting the basic urgency of the task the medical robot is currently performing. In practical applications, such as emergency blood delivery tasks, P... task Set to 100; for routine medicine delivery tasks, P task Set to 50; for empty vehicle return tasks, P task Set it to 10. T represents the waiting time from when the medical robot initiated its elevator ride request to the current moment. current T is the current timestamp. request The timestamp for initiating the request. (D) distance V represents the Euclidean distance from the medical robot's current physical coordinates to the target elevator door. load This indicates the current remaining space capacity of the elevator car. This remaining space capacity can be provided by a cloud-based dispatch center or calculated using a visual assessment algorithm based on cameras installed on the ceiling inside the car, representing the available floor area. , , , These are all preset system adjustment coefficients used to balance the influence of parameters in various dimensions.

[0029] The edge computing gateway sorts all queuing medical robots in the current entry area in descending order according to the W value calculated by the above formula. For the medical robot with the highest W value, the edge computing gateway assigns it the first-tier grid closest to the elevator door's central axis and closest to the elevator door; the remaining medical robots are assigned to physical coordinate grids on the outer side according to the sorting result. After a medical robot moves to the designated waiting grid, it actively triggers a locking mechanism to enter a locked waiting state. Through the deep interactive coupling of the above weight model and grid allocation action, it is possible to use the fusion evaluation of multi-dimensional data to ensure that the most urgent tasks occupy the front physical waiting positions, improving the response speed of critical logistics. It should be noted that the specific values ​​of the coefficients in the above formula are only examples, and those skilled in the art can adjust them according to different business models of hospitals during morning and evening peak hours. , , , Dynamic parameter tuning can also achieve the scheduling logic of this application.

[0030] Building upon the above embodiments, to further guarantee absolute right-of-way for extremely urgent tasks, the edge computing gateway performs dynamic priority determination and adjustment after acquiring multi-dimensional parameters. In response to detecting that the medical robot's task tag is a life-saving task such as emergency rescue or blood sample transportation, the system forcibly sets the inherent priority of that task to the highest level and applies a preset weight coefficient corresponding to that inherent priority. Adjust to the first preset value, and at the same time adjust the weighting coefficient corresponding to the waiting time. Weighting coefficient corresponding to the distance from the elevator door The threshold is forcibly set to zero. This method of shielding against interference from conventional parameters ensures that the medical robot for emergency medical missions receives the highest queuing weight. Conversely, in response to the detection of conventional transportation missions, the system employs a dynamic tolerance mechanism. It determines a first dynamic coefficient based on the ratio of the distance to the elevator door to the standard waiting distance, and a second dynamic coefficient based on the ratio of the already waited time to a preset tolerance time. Subsequently, the first and second dynamic coefficients are multiplied by their corresponding preset weight coefficients and then summed to obtain the final comprehensive passage weight index. This mission-level adjustment method prevents conventional missions from accumulating excessively high weights due to long wait times, thus delaying emergency medical missions and achieving both safety and fairness in global scheduling.

[0031] Once the elevator reaches the current floor and the car door opens, the dispatching process enters the entry / exit conflict resolution phase and executes the delineation of virtual channels. If at this point the system determines that a medical robot inside the elevator car needs to exit (defined as set R),... out The medical robot outside the lobby needs to enter (defined as set R). in Concurrent operations, the edge computing gateway first sends data to set R. in and set Rout All medical robots within the system receive a prohibition command from the underlying layer. Subsequently, the edge computing gateway uses the elevator door's central axis as a reference and dynamically delineates asymmetrical one-way exit virtual channels and one-way entry virtual channels based on the effective width of the car door.

[0032] Specifically, the edge computing gateway obtains the current opening width data of the elevator car door through its interface with the elevator control board, that is, the effective width L when the car door is fully open. door The system identifies the physical coordinates of the elevator door's central axis. It then asymmetrically divides the current opening width data according to a first and a second ratio. If the judgment set R... out If a medical robot needs to exit the elevator, the edge computing gateway is located on the first side of the elevator door's central axis, allocating a width range corresponding to the left door frame to define a one-way exit virtual channel; on the second side of the elevator door's central axis, it allocates a corresponding width range to define a one-way entry virtual channel.

[0033] In this embodiment, to avoid forward physical collisions during entry and exit, the area of ​​the one-way exit virtual passage is defined as follows: This refers to extending half the door width towards the central axis of the door, with the left edge of the door frame as the origin. The corresponding area of ​​the one-way virtual elevator access passage is defined as follows: By employing this asymmetric spatial segmentation algorithm, it is possible to create mutually isolated bidirectional movement paths using the limited physical width of the car door, preventing the medical robot from spatially deadlocking at the intersection point.

[0034] After the channel is allocated, the edge computing gateway initiates the timing sequence by issuing timestamp tokens to control the token issuance. The edge computing gateway prioritizes sending tokens to set R. out The medical robot waiting to exit the elevator issues an elevator token, at which point the set R... in The medical robot waiting to enter the elevator remains locked and in a waiting state. To ensure the exit action is completed, a safety boundary line is set at the end of the one-way exit virtual channel. This safety boundary line is typically set at a physical location 1 meter outside the elevator threshold. In response to the medical robot's physical center of mass crossing this safety boundary line with its exit token, the end-side odometer or positioning module generates a confirmation feedback signal to the edge computing gateway. Based on this confirmation feedback signal, the edge computing gateway confirms that the medical robot has left the interference area, then cancels and reclaims the exit token. Only after the exit token is reclaimed does the edge computing gateway send a notification to the set R. in The medical robot, which is ranked first in the priority list, is issued an entry token. By adopting this closed-loop token mechanism, and utilizing the deterministic feedback of the physical center of mass crossing the boundary, the chain-reaction collisions and friction accidents at the elevator door caused by premature release of right-of-way are eliminated.

[0035] During the scheduling process of a medical robot moving towards the elevator with an entry token, dynamic disturbances such as pedestrians or medical carts often occur randomly in the elevator lobby. To enhance the robustness of the scheduling algorithm in the face of sudden disturbances in the physical space, the system is configured with a feedback-based dynamic adjustment interruption mechanism for the environmental potential field and performs dynamic interruption and replanning. Specifically, the system acquires the local occupancy grid generated by the end-side sensors of the medical robot and marks it as the center of the repulsive potential field distribution. When the unidirectional elevator entry virtual channel is covered by the repulsive potential field for more than a set threshold, the current entry token is withdrawn and the path is replanned.

[0036] The specific execution logic of the aforementioned interruption mechanism is as follows: In response to the medical robot moving within the one-way elevator virtual passage, the medical robot continuously scans for dynamic obstacles within a set height range ahead using a multi-line LiDAR. The system sets a dynamic obstacle recognition filtering height. When an obstacle with a height greater than 0.2 meters is detected and does not bear a system identification tag, a two-dimensional local occupancy grid map is generated. The physical basis of this 0.2-meter height filtering threshold is to filter out minor undulations in the ground, threshold gaps, or flat debris, focusing on identifying substantial obstacles that may cause physical collisions.

[0037] The system extracts the center coordinates of obstacle outlines from a 2D local occupancy grid map, using these coordinates as the center of the repulsive potential field distribution. It then generates the corresponding Gaussian repulsive potential field distribution range based on the moving speed of the dynamic obstacle. The edge computing gateway establishes a potential field mapping based on the received obstacle position and suppresses and corrects the reachability of the surrounding grid. Subsequently, the edge computing gateway calculates the overlap area between the Gaussian repulsive potential field distribution range and the currently allocated one-way elevator virtual channel in the physical coordinate grid. If the overlap area exceeds 50% of the total area of ​​the one-way elevator virtual channel, the edge computing gateway triggers a hardware-level interrupt command. The edge computing gateway forcibly reclaims the current elevator entry token, causing the medical robot to stop, and calls recalculation logic to generate a guidance path that allows the medical robot to retreat to the nearest safe avoidance grid until the potential field dissipates and it re-initiates an entry request. By constructing a specific closed-loop response process for local occupancy grid feedback and repulsive potential field interruption, the fusion of edge-side perception and rapid edge decision-making can address the shortcomings of traditional fixed trajectory algorithms that are prone to failure in complex dynamic scenarios involving medical staff and patients.

[0038] In another embodiment of this application, the specific connection structure and signal link relationship of the hardware base for implementing the above-described control method are described in detail. The aforementioned medical robot cross-floor elevator scheduling system includes an edge computing gateway comprising an embedded processing unit and a near-field communication antenna array. To ensure that the radio frequency signal can cover the core working area of ​​the elevator lobby without blind spots, the near-field communication antenna array is oriented directly above the elevator door frame, with the antenna beam tilted downwards at a specific angle to completely cover all physical coordinate grids within a specific distance range of the elevator lobby. The embedded processing unit establishes a low-latency dedicated local area network connection with each medical robot within the physical coordinate grid through this near-field communication antenna array. The reception of entry requests, the synchronization of the comprehensive access weight index, and the issuance of elevator entry tokens are all strictly completed through this low-latency dedicated local area network connection. This design, which combines the communication module with a specific physical installation location, avoids cross-floor signal interference and ensures the accuracy of time slot token distribution.

[0039] Furthermore, to achieve accurate capture of unmarked dynamic obstacles, the externally mounted end-side sensors of the medical robot specifically include a multi-line LiDAR and a depth camera. In terms of physical spatial arrangement, the multi-line LiDAR is fixedly mounted at the top center of the medical robot, configured to provide 360-degree horizontal two-dimensional point cloud data for macroscopic contour tracking. The depth camera is mounted tilted downwards on the front face of the medical robot, specifically configured to acquire three-dimensional depth information of low-lying dynamic obstacles within a preset height range above the ground, compensating for the insufficient detection of the blind spots below by the top LiDAR. In terms of data processing, the main control computing unit strictly timestamps the point cloud sequence output by the multi-line LiDAR and the depth map output by the depth camera. After alignment, spatial registration is performed using an external parameter matrix, fusing the heterogeneous sensor data to generate the aforementioned two-dimensional local occupancy grid map. The main control computing unit compresses and packages the local occupancy grid and continuously uploads it to the edge computing gateway via a low-latency dedicated local area network connection. By using this hardware architecture that couples specific installation posture with spatial registration of multi-source sensors, it is possible to eliminate perception blind spots from a multi-dimensional perspective, thereby providing a data support foundation for the accurate establishment of the Gaussian repulsive potential field.

[0040] In another embodiment of this application, an electronic device is provided. This electronic device typically manifests as a core control board within an edge computing gateway or a server related to a cloud scheduling center, and includes a memory and a processor communicatively connected to the memory. The memory stores computer instructions; when the processor executes the computer instructions, it is able to implement all the steps of the scheduling method for a medical robot to travel across floors in the various embodiments mentioned above.

[0041] In summary, this application divides physical coordinate grids within a specific distance range using an edge computing gateway and dynamically sorts queuing priorities based on a comprehensive access weight model that considers multi-dimensional parameters. Simultaneously, in the event of elevator door opening conflicts, an asymmetric virtual channel is defined based on the car door width, combined with a strict token control mechanism for elevator exit and entry, and an interruption feedback method based on edge radar potential field scanning. These core features, along with the auxiliary hardware and software collaborative architecture, solve the pain points of local congestion and physical deadlock in the lobby area for multiple medical robots, while ensuring the absolute priority of emergency transport tasks and maintaining system safety under dynamic and complex disturbances, thus improving the overall efficiency of hospital logistics scheduling.

[0042] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A scheduling method for medical robots to travel across floors in an elevator, characterized in that, include: The system receives applications for entry into the medical robot's domain and divides a specific distance range in the waiting hall into multiple physical coordinate grids based on a pre-stored high-precision two-dimensional grid map. The comprehensive passage weight index of the medical robot is calculated and sorted in descending order. The designated physical coordinate grid is assigned to it as the waiting grid, and its system state is changed to a locked waiting state. When the elevator car door opens, in response to the entry and exit conflict, an asymmetrical one-way exit virtual channel and one-way entry virtual channel are dynamically defined based on the effective width of the car door and the central axis of the elevator door. An elevator token is first issued to the medical robot waiting to exit and withdrawn after confirming that it has crossed the safety boundary. Then, an entry token is issued to the medical robot waiting to enter. The local occupancy grid generated by the end-side sensor of the medical robot is obtained and marked as the center of the repulsive potential field distribution. When the one-way entry virtual channel is covered by the repulsive potential field for more than a set threshold, the current entry token is withdrawn and the path is replanned.

2. The scheduling method for medical robots to travel across floors using elevators as described in claim 1, characterized in that, The steps of calculating the comprehensive access weight index of the medical robots, sorting them in descending order, assigning them designated physical coordinate grids as waiting grids, and changing their system state to a locked waiting state include obtaining multi-dimensional parameters, including the inherent task priority of the medical robots, the waiting time, the distance to the elevator door, and the remaining capacity of the target elevator car; calculating the comprehensive access weight index by weighted summation of the multi-dimensional parameters based on preset weight coefficients; sorting multiple medical robots according to the comprehensive access weight index from largest to smallest, and assigning the physical coordinate grid closest to the elevator door to the corresponding ranked medical robot according to the sorting result.

3. The scheduling method for medical robots to travel across floors using elevators as described in claim 2, characterized in that, After obtaining the multi-dimensional parameters, the system is configured to, in response to detecting that the medical robot's task is emergency rescue or blood sample transport, set the inherent priority of the task to the highest level, adjust the preset weight coefficient corresponding to the inherent priority of the task to a first preset value, and set the weight coefficient corresponding to the waiting time and the distance from the elevator door to zero; in response to detecting a regular transport task, determine a first dynamic coefficient based on the ratio of the distance from the elevator door to the standard distance, and determine a second dynamic coefficient based on the ratio of the waiting time to the preset tolerance time. The first dynamic coefficient and the second dynamic coefficient are multiplied by their respective preset weight coefficients and then superimposed to obtain the final comprehensive passage weight index.

4. The scheduling method for medical robots to travel across floors using elevators as described in claim 1, characterized in that, The steps of dynamically defining asymmetrical one-way exit virtual channels and one-way entry virtual channels based on the effective width of the elevator door centerline include: obtaining the current opening width data of the elevator car door and identifying the physical coordinates of the elevator door centerline; dividing the current opening width data according to a first ratio and a second ratio; and defining the one-way exit virtual channel on the first side of the elevator door centerline based on the width corresponding to the first ratio. The one-way virtual elevator entrance channel is defined on the second side of the central axis of the elevator door based on the width corresponding to the second ratio.

5. The scheduling method for medical robots to travel across floors using elevators as described in claim 4, characterized in that, The step of acquiring a locally occupied grid generated by the end-side sensors of the medical robot and marking it as the center of the repulsive potential field distribution, and withdrawing the current entry token and replanning the path when the one-way elevator virtual channel is covered by the repulsive potential field for more than a set threshold, is configured as follows: In response to the medical robot moving within the one-way elevator virtual channel, a two-dimensional locally occupied grid map is generated by continuously scanning dynamic obstacles within a set height range using an end-side LiDAR; the coordinates of the obstacle contour center in the two-dimensional locally occupied grid map are extracted as the center of the repulsive potential field distribution, and a corresponding Gaussian repulsive potential field distribution range is generated according to the moving speed of the dynamic obstacle; the overlap area between the Gaussian repulsive potential field distribution range and the one-way elevator virtual channel in the physical coordinate grid is calculated; in response to the overlap area accounting for more than 50% of the total area of ​​the one-way elevator virtual channel, an interrupt command is triggered, the entry token is revoked, a guidance path is generated to allow the medical robot to return to the nearest safe avoidance grid and re-initiate the domain entry request.

6. The scheduling method for medical robots to travel across floors using elevators as described in claim 1, characterized in that, The steps of prioritizing issuing an exit token to the medical robot waiting to exit and retrieving it after confirming that it has crossed the safety boundary include setting a safety boundary line at the end of the one-way exit virtual channel; generating a confirmation feedback signal in response to the physical centroid of the medical robot carrying the exit token crossing the safety boundary line; and canceling the exit token based on the confirmation feedback signal.

7. A scheduling system for medical robots to travel across floors in an elevator, characterized in that, include: The edge computing gateway is physically deployed inside the elevator lobby on each floor. It is used to receive the domain entry application of medical robots and divides the elevator lobby into multiple physical coordinate grids based on a pre-stored high-precision two-dimensional grid map. A medical robot is externally equipped with edge sensors, and its internal main control computing unit establishes a network connection with the edge computing gateway. The edge computing gateway calculates the comprehensive access weight index of the medical robot, sorts it in descending order, assigns a designated physical coordinate grid as the waiting grid for the medical robot, and instructs it to enter a locked waiting state. After the elevator door opens, the edge computing gateway dynamically delineates an asymmetrical one-way exit virtual channel and a one-way entry virtual channel based on the effective width of the elevator door and the central axis of the elevator door. The edge computing gateway first issues an elevator token to the medical robot waiting to exit and retracts it after confirming that it has crossed the safety boundary. Then, it issues an entry token to the medical robot waiting to enter. The edge computing gateway receives the locally occupied grid generated by the edge sensors and marks it as the center of the repulsive potential field distribution. When the one-way entry virtual channel is covered by the repulsive potential field for more than a set threshold, the current entry token is withdrawn and the path is replanned.

8. The scheduling system for medical robots to travel across floors using elevators as described in claim 7, characterized in that, The edge computing gateway includes an embedded processing unit and a near-field communication antenna array; the near-field communication antenna array is oriented above the elevator door frame to cover a specific distance range of the waiting hall. The embedded processing unit establishes a low-latency dedicated local area network (DLAN) connection with the medical robot located within the physical coordinate grid through the near-field communication antenna array, and receives the domain entry request and issues the elevator access token through the low-latency DLAN connection.

9. The scheduling system for medical robots to travel across floors in elevators as described in claim 8, characterized in that, The edge sensors include a multi-line LiDAR and a depth camera. The multi-line LiDAR is fixedly installed at the top center of the medical robot and configured to provide 360-degree horizontal two-dimensional point cloud data. The depth camera is installed at an angle downwards on the front end of the medical robot and configured to acquire three-dimensional depth information of low-lying dynamic obstacles within a preset height range above the ground. The data from the multi-line LiDAR and the depth camera are fused after time-stamp alignment and spatial registration to generate the local occupancy grid. The main control computing unit compresses and packages the local occupancy grid and continuously uploads it to the edge computing gateway via the low-latency dedicated local area network.

10. An electronic device, characterized in that, include: A memory for storing computer instructions; a processor communicatively connected to the memory; the processor, when executing the computer instructions, implements the scheduling method for a medical robot to travel across floors using an elevator as described in any one of claims 1 to 6.

Images