Multi-path autonomous identification deployment intelligent internet of things wheelchair cruise system and control method

Abstract

The application provides a multi-path autonomous identification and deployment intelligent Internet of Things wheelchair cruise system and a control method, and focuses on solving the technical bottleneck problems of single obstacle detection sensor in the existing intelligent wheelchair cruise process, lack of multi-vehicle cruise path planning experience, and inability to accurately avoid obstacles for dynamic obstacles. When riding, only the disease needs to be stated or the social security card needs to be inserted, and the wheelchair can automatically lead the patient to the corresponding clinic for medical treatment. When laboratory test is needed, the wheelchair voice prompts the patient and automatically carries the patient; when the wheelchair is in front of the test order machine, the test order machine is automatically ordered through communication between the devices. The current situation that the old people cannot use intelligent devices is avoided, the hospital medical process is simplified, the cost of manpower and material resources spent by the hospital for the above special personnel is saved, and the crowd faced by the smart medical treatment is more extensive.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of control technology, and particularly relates to a multi-path autonomous identification and deployment intelligent Internet of Things wheelchair cruise system and a control method. BACKGROUND

[0002] With the progress of science, the combination of medical treatment and technology has achieved the effect of intelligent medical treatment. The conventional intelligent medical guide device currently in use has a development goal of realizing intelligent medical guide and diagnosis through a man-machine interaction system, and the main function is to replace the service work of an artificial medical guide station, and the target user group is mainly young and middle-aged patients with experience in using intelligent devices and who can move freely. However, for most elderly patients, the operation of the conventional intelligent medical guide device is slightly complex, which makes them have a great psychological resistance to the too intelligent device system, and often because of inconvenience in movement and lack of physical strength, the independent medical treatment process is difficult. Therefore, simply relying on an intelligent voice self-service system still cannot solve the problems of accompanying, finding a way, and preferential delivery of the medical treatment process of elderly patients.

[0003] At present, various intelligent distribution vehicles exist, but the utilization rate of path resources is not high, and the number of devices that can operate simultaneously is small. Therefore, how to improve the transportation efficiency of the intelligent distribution vehicle while ensuring stable operation and improve its transportation method is a problem that needs to be solved urgently at present. SUMMARY

[0004] Based on the above-mentioned deficiencies existing in the prior art, the present application provides an intelligent Internet of Things wheelchair cruise system and control method based on multi-path autonomous identification and deployment, which can plan the most suitable path according to the current road condition information and crowd dense point situation after the patient enters the hospital, and go to the hospital.

[0005] In one aspect, the present application provides an intelligent Internet of Things wheelchair cruise system based on multi-path autonomous identification and deployment, comprising:

[0006] A micro control unit for storing control codes of the entire system except the obstacle detection module.

[0007] A Bluetooth MESH communication module for realizing communication of the micro control unit, monitoring and cloud platform supervision system, while maintaining system stability and uninterrupted network.

[0008] A multi-sensor information fusion obstacle detection module for monitoring the obstacle condition around the wheelchair, and outputting the detected obstacle signal when a collision is about to occur.

[0009] A tracking module for participating in the right angle bend judgment in path planning and the relative position judgment with the ground mark.

[0010] A state monitoring module is configured to monitor the running speed of the wheelchair and the fault condition of the wheelchair.

[0011] A vehicle body recognition module is configured to recognize vehicles on the corresponding path segment and clear the path occupation information.

[0012] A voice recognition and broadcast module is configured to recognize the corresponding disease and broadcast module by monitoring the disease keyword, recognize the destination information in the user voice instruction, and broadcast the arrival position information in real time.

[0013] In another aspect, the application also provides a control method of the intelligent Internet of Things wheelchair cruise system based on the multi-path autonomous recognition and deployment as described above, comprising the following steps:

[0014] Step 1: The patient gets on the wheelchair, and the system detects that someone gets on and the voice recognition and broadcast module prompts the patient to insert the social security card into the card slot or orally describe the symptoms after detecting that someone gets on, determines the target clinic after recognizing the disease, and determines the target clinic.

[0015] Step 2: The system executes the path resource utilization maximization algorithm and the video recognition algorithm to determine the driving route and start running.

[0016] Step 3: The wheelchair automatically takes the patient to the destination.

[0017] Compared with the prior art, the application has the following beneficial effects:

[0018] The application focuses on solving the technical bottleneck problems of single obstacle sensor in the existing intelligent wheelchair cruise process, lack of multi-vehicle cruise path planning experience, and inability to accurately avoid obstacles for dynamic obstacles, and facilitates the medical treatment of the elderly, pregnant women and disabled people, mainly in the form of one-stop medical treatment by riding an intelligent wheelchair. When riding, only the disease needs to be orally described or the social security card needs to be inserted, and the wheelchair can automatically lead the patient to the corresponding clinic for medical treatment.

[0019] When laboratory test orders are needed, the wheelchair prompts the patient and automatically takes him / her to the test order machine. When the wheelchair is in front of the test order machine, the test order machine is automatically ordered through communication between devices. The current situation that the elderly cannot use smart devices is avoided, the hospital medical treatment process is simplified, and the cost of manpower and material resources spent by the hospital for the above special personnel is saved, so that the intelligent medical treatment faces a wider population.

[0020] The video recognition technology and the path resource utilization maximization algorithm improve the stability and safety during running. More importantly, the number of devices that can be stably operated in the space is greatly improved through the algorithm, and the transportation efficiency of the tracking vehicle is improved.

[0021] In addition, the application helps the elderly, the disabled, pregnant women and other patients with difficulty in moving to provide one-stop function coverage from entering the hospital gate to standing in front of the doctor, thereby effectively reducing the consumption of manpower and material resources in the hospital guide management. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 Flowchart of the visit process of the embodiment of the application;

[0023] Figure 2 General block diagram of the system module of the embodiment of the application;

[0024] Figure 3 Path resource utilization maximization algorithm before the operation of the embodiment of the application;

[0025] Figure 4 Device control flow during the operation of the embodiment of the application;

[0026] Figure 5 Path resource utilization maximization algorithm during the operation of the embodiment of the application;

[0027] Figure 6 Device control flow when entering and exiting the elevator of the embodiment of the application;

[0028] Figure 7 Circuit diagram of the six-way infrared tracking module of the embodiment of the application;

[0029] Figure 8 Relative position of the six-way infrared sensor and the ground mark of the embodiment of the application;

[0030] Figure 9 Infrared transmitting and receiving circuit diagram of the vehicle body recognition module of the embodiment of the application;

[0031] Figure 10 Video recognition algorithm of the embodiment of the application. DETAILED DESCRIPTION

[0032] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments. Obviously, the drawings described below are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0033] As Figure 2As shown, the present application provides a hospital intelligent Internet of Things wheelchair cruise system based on multi-path autonomous identification deployment. The multi-sensor information fusion technology is applied to the obstacle detection system. Through the cooperation of multiple sensors with their respective micro control units, the surrounding environmental information is monitored in parallel. The output signal is simplified through an OR gate circuit, and the accuracy of the surrounding obstacle information judgment is improved. The front-end tracking module uses multiple infrared photoelectric sensors, so that the wheelchair can still return to the original track when entering or exiting the elevator or receiving a collision. Specifically, it includes:

[0034] Micro control unit, using STM32F103ZET6, used to store the control code of the entire system except the obstacle detection module.

[0035] Bluetooth MESH communication module, using JDY-24M chip, used to realize the communication between micro control unit, monitoring and cloud platform supervision system, while maintaining system stability and uninterrupted network.

[0036] Multi-sensor information fusion obstacle detection module, composed of STM32F103C8T6 single-chip microcomputer, ultrasonic module and infrared obstacle avoidance module, used to monitor the obstacle situation around the wheelchair. When a collision is detected, the detected obstacle signal is output.

[0037] Tracking module, composed of six-way infrared photoelectric sensor, its circuit connection is as Figure 7 As shown, used for participating in the right angle bend judgment in path planning and the relative position judgment with the ground mark.

[0038] State monitoring module, used to monitor the running speed of the wheelchair and its own fault condition. When the wheelchair stops running due to abnormal reasons caused by severe unpredictable collision or other special circumstances, an alarm signal will be sent to the labview monitoring center to provide the supervisor, so as to handle it in time.

[0039] Vehicle recognition module, used to identify the vehicles on the corresponding path segment and clear the path occupation information. Specifically, a specific frequency infrared transmitter is installed on the top of the wheelchair. An infrared receiver that receives a specific frequency band and the corresponding micro control unit and Bluetooth module are installed on the top of each path (ceiling), so as to clear the path occupation information.

[0040] Voice recognition and broadcast module, mainly composed of voice recognition LD3320 and voice broadcast NV180C and its corresponding peripherals, used to identify the corresponding disease by monitoring the disease keyword, identify the destination information in the user voice instruction, and broadcast the arrival position information in real time. The information is sent to the micro control unit through the serial port.

[0041] DC speed reduction motor, used for various movements of the wheelchair.

[0042] A power module for powering the entire system.

[0043] Further, the cloud platform supervision system comprises a SQL database and a floor-by-floor labview monitoring center. The labview monitoring center is arranged on each floor and communicates with each module through a Bluetooth MESH communication module. The SQL database communicates with the labview monitoring center through a network to realize real-time updating and broadcasting of path occupation information and crowd dense points, and to realize real-time monitoring of the operation of each module.

[0044] Further, the ground mark is a hospital indoor traffic map in the form of white background black straight line segments. The corresponding path segments are written in the micro control unit, and the path segment information labels are formed to form a binary hospital indoor traffic map, which is stored in the micro control unit and the labview monitoring center for use by the entire system.

[0045] Further, the communication mode used in the system is a Bluetooth MESH communication mode. The communication mode is networked through a plurality of JDY-24M Bluetooth modules. Each floor of the wheelchair, the Bluetooth node and the labview monitoring center uses an independent network label. The wheelchair changes the network label on the wheelchair to the corresponding floor after passing through the floor.

[0046] The vehicle body recognition module circuit diagram of the system is shown in Figure 9 The transmitting circuit generates a sine signal of the corresponding frequency through a sine wave generating circuit, and then drives an infrared transmitting tube through a triode. The frequency of the infrared light emitted by the vehicle is 1KHZ~100KHZ, and the progressive frequency is 1KHZ. If there are one hundred vehicles, the frequencies of the infrared light emitted by the one hundred vehicles are 1~100KHZ. The frequency amount of the device can be increased by changing the progressive frequency. An infrared receiving head is arranged above each path segment. When the receiving head receives the infrared light, it enters the filter circuit to filter out the noise, and then passes through the comparison circuit to form a square wave signal for the micro control unit to read. After reading, if it is a frequency within the frequency band, the occupation information of the road is cleared and sent to the labview control center of the corresponding floor through Bluetooth MESH.

[0047] As shown in Figure 1As shown, the embodiments of the present application also provide a control method of the hospital intelligent Internet of Things wheelchair cruise system based on the multi-path autonomous identification and deployment as described above. When a patient gets on the wheelchair, the patient orally states the departure place and symptoms or inserts a social security card. The system will voice recognize the illness, determine the target clinic, execute the path resource utilization maximization algorithm before running, determine the driving route, and start running. During the running, the system executes the control process during running and the path resource utilization maximization algorithm during running, plans the driving route, continues running, and finally successfully arrives at the destination for treatment. The specific steps are as follows:

[0048] Step 1: The patient gets on the wheelchair, and the system detects that someone gets on the wheelchair. The voice recognition and broadcast module prompts the patient to insert the social security card into the card slot or orally state the symptoms after the system detects that someone gets on the wheelchair. After recognizing the illness, the target clinic is determined.

[0049] Step 2: The system executes the path resource utilization maximization algorithm (hereinafter referred to as the path resource maximization algorithm) and the video recognition algorithm, determines the driving route, and starts running.

[0050] Step 3: The wheelchair automatically takes the patient to the destination.

[0051] Further, the step 1 is specifically as follows:

[0052] Step 1.1: The patient enters the hospital and goes to the place where the intelligent wheelchair is placed to get on the wheelchair. After the wheelchair detects that someone gets on the wheelchair, the voice recognition and broadcast module prompts the patient to insert the social security card into the card slot or orally state the symptoms.

[0053] Step 1.2: When the patient orally states the symptoms: the voice recognition module actively recognizes the keywords, matches the keywords with the keyword library, and then sends the information to the micro control unit. After the micro control unit recognizes, the corresponding sentence in the voice broadcast module is driven to respond to the patient, and the target clinic is determined.

[0054] When the patient inserts the social security card: the illness that the patient needs to re-visit is recognized from the inserted social security card in the hospital treatment system, and the target clinic is determined.

[0055] Further, the step 2 is specifically as follows:

[0056] The system executes the path resource maximization algorithm before running and the video recognition algorithm to obtain the path communication information and determine the driving route.

[0057] The system executes the path resource maximization algorithm during running and the video recognition algorithm, simultaneously uses the tracking module and the speed measurement module in the state monitoring module to track the ground marks and drive at a uniform speed.

[0058] In particular, the system features a path resource utilization maximization algorithm, which is divided into two parts: a path resource maximization algorithm before operation and a path resource maximization algorithm during operation.

[0059] Furthermore, the pre-run path resource maximization algorithm performs the following operations:

[0060] Determine N path segments between the departure and destination, and combine path occupancy information, crowd density points, and wheelchair failure conditions into path information;

[0061] Receive path information broadcast by the control center;

[0062] Calculate the path occupancy information with the N path codes respectively;

[0063] There are M false positives among N results, indicating that there are M idle feasible paths. Calculate these numbers together with the number of densely populated points to obtain the number of M types of 1s. Select the route with the least number of 1s as the running route.

[0064] The running routes are combined into new path information and sent to the control center. When the next information broadcast by the control center is consistent with the sent one, the driving route is determined and the wheelchair starts running.

[0065] like Figure 3 The specific algorithm details are as follows:

[0066] This application assumes that there are 23 route segments between the departure and destination points. These 23 route segments (i.e., one straight line is a route segment) are numbered from route segment 0 to route segment 22. Each route segment corresponds to a binary code, where 0 represents an unoccupied route segment and 1 represents an occupied route segment.

[0067] Each surveillance camera monitors a road section and corresponds to a binary code, 0 represents that there is no crowd density point on the road section, and 1 represents that there is a crowd density point on the road section.

[0068] The communication information is a 48-bit binary code: 23 bits of binary data (representing path occupancy information), 23 bits of binary data (representing crowded points), and 2 bits of binary data (representing wheelchair malfunctions). Alternative routes between these points are stored in the STM32F103ZET6. When the monitoring system detects that a wheelchair is at the end of a road section, the occupancy information for that section is cleared.

[0069] Furthermore, Figure 5 As shown, the running path resource maximization algorithm performs the following operations:

[0070] Receive route information representing crowd density points from the broadcast. When a crowd density point appears on the route to be taken, set the current position to A and the current route to be No. 1;

[0071] Select n routes from the remaining routes that have the same path segment as route 1 before point A, and perform calculations on these n routes to obtain m routes that do not conflict with the currently occupied route segment.

[0072] Select the m routes and the one with the least number of conflicts with crowded points among the current routes as the new route;

[0073] The wheelchair stops moving and exchanges data with the control center, clearing the original path occupancy information and using the road section of the new route as the path occupancy information;

[0074] After the data exchange is successful, the driving route is determined and the operation continues.

[0075] like Figure 4 As shown, the control during system operation includes the following steps:

[0076] Scan the IO ports of the multi-sensor environmental monitoring module;

[0077] Scan the status of the tracking sensor;

[0078] Perform pid speed regulation to make the wheelchair move at a constant speed;

[0079] Read the path information in the Bluetooth MESH communication module;

[0080] The wheelchair continues to operate.

[0081] The program has already controlled the situation of passing the elevator on the way. The elevator point will be marked in the program. When the tracking module recognizes it, it will execute the elevator entry and exit operation program. The specific operation process is as follows Figure 6 At the same time, the system uses Bluetooth MESH communication to ensure that the system can also maintain network connectivity in the elevator, thus improving the stability of the system.

[0082] Furthermore, the video recognition algorithm used in this system uses the CSRNet network to perform accurate counting estimation and present high-quality density maps. It can identify the crowd density of the corresponding road section, represent it on the map in RGB format, and send the information to the labiview monitoring center at each level via Bluetooth MESH, such as Figure 10 As shown, the video recognition algorithm performs the following operations:

[0083] Input the current surveillance video screen;

[0084] Through the first part of the CSRNet network model: convolutional network for 2D feature extraction;

[0085] The second part of the CSRNet network model: the hole convolution network, is used to reduce the feature vector output by the convolution layer and improve the result;

[0086] The label file is subjected to Gaussian filtering processing as the GroundTruth (ground truth) of the model prediction;

[0087] The density map is output.

[0088] Further, the label file is derived from the labeling process of the data set and is mainly used to label the positions and speeds of various obstacle objects in the road scene, such as vehicles, pedestrians, corner intersections, etc. These label files are obtained by manual or automatic labeling of actual driving scene data. The specific content and format contain the following information:

[0089] 1) Object category: such as vehicles, pedestrians, etc.

[0090] 2) Whether truncated: to determine whether the object leaves the image boundary.

[0091] 3) Whether occluded: to determine the degree of occlusion of the object (completely visible, partially occluded, mostly occluded, completely occluded).

[0092] 4) Viewing angle: the angle of the object in the camera coordinate system.

[0093] 5) 2D bounding box: the two-dimensional bounding box coordinates of the object.

[0094] 6) Detection confidence: indicating the reliability of the detection.

[0095] Further, the GroundTruth (ground truth) refers to the true value or target value labeled in the data set. In the prediction phase, the algorithm needs to provide the true label (GroundTruth) of each sample. Assuming that the system is performing an image classification task, it needs to label each image as the corresponding category. Each label includes an image + file name, and these predicted values correspond to the GroundTruth. Each data in the data set corresponds to a label in the form of (x, t), where x is the input data and t is the corresponding label. If the label is correct, it is called GroundTruth, otherwise it is not. The data input into the network is (x, t), and the predicted data obtained by the model is (x, y). The prediction result y is compared with the label t, the loss is calculated by the minimum variance, and the true value and the predicted value are compared to judge the quality of the prediction model. That is, the GroundTruth in the system is the result of manual labeling, and in target detection, it needs to be compared with the predicted value.

[0096] The specific medical details of the examples of the present application are described below:

[0097] 1) The patient enters the hospital and goes to the wheelchair placement area at the entrance, and the wheelchair detects that someone has sat on it, and then prompts the patient to insert the social security card into the card slot or say the condition to determine the destination. At the same time, it will ask the departure place and plan the path. For example, the patient only needs to say "sore throat" and "I am at the entrance", and the system can automatically identify the destination and departure place and plan the path from the entrance to the respiratory department clinic room.

[0098] In this step, the LD3320 chip in the voice recognition module can actively identify the keywords and match them with the keywords in the dictionary, and then send the information to the STM32F103ZET6 single-chip microcomputer. At the same time, the single-chip microcomputer will drive the corresponding statement in the NV180C voice playback chip to respond to the patient. After determining the destination and departure place, the system will use the path resource utilization maximization algorithm to obtain the current path occupancy information and crowd concentration points through the Bluetooth MESH communication module before and during operation, so as to run the internal path resource utilization maximization algorithm for path planning, thereby improving the number of devices that can run simultaneously in the same area.

[0099] 2) During the operation of the wheelchair, the tracking module and the speed measurement module in the state monitoring module are used for ground mark tracking and uniform speed driving.

[0100] The circuit diagram of the tracking module is shown in Figure 7 The relative position of the six infrared sensors to the ground mark is shown in Figure 8 When the sensor is above the black substance, the receiving tube in the sensor cannot receive the reflected infrared light, thereby outputting a high level, and vice versa. The single-chip microcomputer can determine the relative position to the ground mark by reading the high and low levels of the six IO ports. The principle of uniform speed movement of the car is to determine the speed of each wheel by counting the number of level reversals of the speed measurement module within a certain period of time (20ms), and then perform PID adjustment. When the car is in the center, it will continue to move forward at a uniform speed. When the car position deviates to the left or right due to crossing obstacles or other reasons, the PID parameter setting is determined by the degree of deviation (i.e. the sensor number responding), so that the car can be repositioned in the center. When the car is at a crossroads or T-junction, the path reading in the program will be performed to execute the turning or straight driving program.

[0101] 3) If the wheelchair encounters an obstacle or a crowd that cannot be passed during operation, it will prompt the surrounding crowd to help or avoid, so as to continue running.

[0102] In this step, the path resource utilization maximization algorithm mentioned above will be used for path planning. If there are still crowded points on the route and the crowded points have not been evacuated upon arrival, the NV180C voice chip will be driven to play the corresponding speech segment to prompt passers-by.

[0103] When the IO port of the multi-sensor environmental monitoring system detects the presence of people and obstacles, the system will stop the movement of the wheelchair and play voice prompts to the crowd to avoid or help until there are no obstacles and then continue to run. During the operation process, the status of the tracking sensor is scanned. When a right-angle bend is detected, the wheelchair will move forward or turn according to the set route. After passing the right-angle bend, the speed will be adjusted to make it travel at a constant speed. During the constant speed driving process, the path information in the Bluetooth mesh will be read. If the path information changes, the system will select the path resource in operation and use the path provided by the maximization algorithm to change the driving direction. If the path information does not change, the wheelchair will continue to move.

[0104] 4) Upon reaching the destination, while the patient is receiving medical treatment, the system sends an arrival signal to the LabVIEW monitoring center on the corresponding floor. The LabVIEW monitoring center then checks to see if any path occupancy information has been cleared due to vehicle misidentification. If so, it will be cleared. After the treatment is complete, continue with step 1, only stating the destination.

[0105] 5) If the destination is not on the same floor as the departure point, the wheelchair will first take the elevator door on the departure floor as the destination. After arriving, the system will send an arrival signal to the Labview monitoring center on the corresponding floor. The Labview monitoring center will check whether there is any path occupancy information that has not been cleared in time due to the vehicle body not being recognized. If so, it will be cleared. Then follow Figure 6 The control process runs. When the vehicle arrives in front of the elevator, it prompts the patient to press the elevator button. After the front ultrasonic sensor detects the door opening, it turns around and enters the elevator. The ultrasonic ranging sensor is used to measure the distance. When the rear distance is about 50 cm, it stops running. The voice prompts people around to help press the elevator button of the corresponding floor or prompts the patient to stand up and operate. After arriving at the corresponding floor, the patient presses the button on the vehicle, and the vehicle will move forward. After driving out of the elevator door, the front-end six-way sensor tracking module adjusts the relative position with the ground mark so that the ground mark is in the center of the vehicle. Then, after reaching the corresponding floor, the single-chip machine sends a command to change the network label of the Bluetooth module on the vehicle and restarts to wait for network access. It requests path information from the LabVIEW monitoring center on the corresponding floor, and plans the path with the elevator as the new starting point and the corresponding clinic as the destination.

[0106] 6) After reaching the target diagnosis room, if the patient needs to take a test, the patient can repeat the voice control and say the destination of the test room. The system will automatically take the patient to the destination for testing. When taking the test, the patient only needs to insert the social security card medical card. The system will automatically connect to the hospital's medical system and take the patient to the test destination. When the wheelchair is in front of the test machine, through communication between the devices, the wheelchair will confirm the patient's identity information through the hospital's medical system and voice. After confirming the information, the test machine will automatically issue the test.

[0107] Obviously, the above embodiments of the present application are only examples for clearly illustrating the present application, and are not intended to limit the embodiments of the present application. For those skilled in the art, other different forms of changes or variations can be made on the basis of the above description, and all the embodiments cannot be exhausted here. Any obvious changes or variations derived from the technical solutions of the present application are still within the protection scope of the present application.

Claims

1. A control method for an intelligent IoT wheelchair cruise system based on multi-path autonomous identification and deployment, the intelligent IoT wheelchair cruise system based on multi-path autonomous identification and deployment comprising: Microcontrol unit, used to store the control code of the entire system except the fault detection module; Bluetooth MESH communication module, used to achieve communication between micro-control units, monitoring equipment and cloud platform supervision systems, while keeping the equipment stable and connected to the network; The multi-sensor information fusion obstacle detection module is used to monitor obstacles around the wheelchair and output a detected obstacle signal when an impending collision is detected; Tracking module, used to determine right-angle turns in path planning and relative position to ground markers; The ground markings are indoor traffic maps laid out in the form of black straight line markings on a white background; the corresponding path segments are written into the microcontroller unit and the path segment information is numbered, with 0 representing an unoccupied path segment and 1 representing an occupied path segment, thereby forming a binary indoor traffic map that is stored in the microcontroller unit and the LabVIEW monitoring center; Condition monitoring module, used to monitor the equipment's operating speed and its own fault conditions; The vehicle body recognition module is used to identify the wheelchair on the corresponding path segment and clear the path occupancy information. Specifically, an infrared transmitter with a set frequency is installed on the top of the wheelchair, and an infrared receiver with a set frequency band, as well as a corresponding micro control unit and Bluetooth MESH communication module, are installed on the top of each path segment to clear the path occupancy information. The device is added by incrementally changing the frequency of infrared light emitted by the wheelchair; The voice recognition and broadcast module is used to identify the corresponding condition by monitoring the condition keywords, recognize the destination information in the user's voice command, and broadcast the arrival location information in real time; The cloud platform monitoring system includes an SQL database and a floor-based host computer labview monitoring center; The host computer labview monitoring center is set up on each floor and communicates with each module via the Bluetooth MESH communication module; The SQL database and the host computer labview monitoring center use network communication to achieve real-time update and broadcast of path occupancy information and crowd density points, and at the same time monitor the operation of each module in real time; The method is characterized in that it comprises the following steps: Step 1: The patient is in a wheelchair. When the system detects that someone is sitting in the wheelchair, the voice recognition and announcement module prompts the patient to insert the social security card into the card slot or verbally state the symptoms. After identifying the condition, the target clinic is determined; Step 2: The system executes the path resource utilization maximization algorithm and video recognition algorithm to determine the driving route and start operation; Step 3: The wheelchair automatically takes the patient to the destination; The step 2 is specifically as follows: The system executes the pre-operational path resource maximization algorithm and video recognition algorithm to obtain path communication information and determine the driving route; The system executes the path resource maximization algorithm and video recognition algorithm in operation, and uses the tracking module and the speed measurement module in the status monitoring module to track the ground signs and drive at a constant speed; The pre-run path resource maximization algorithm performs the following operations: Determine N path segments between the departure and destination, and combine path occupancy information, crowd density points, and equipment failure conditions into path information; Receive path information broadcast by the control center; Calculate the path occupancy information with the N path codes respectively; There are M false positives among N results, indicating that there are M idle feasible paths. Calculate these numbers together with the number of densely populated points to obtain the number of M types of 1s. Select the route with the least number of 1s as the running route. The operation routes are combined into new path information and sent to the control center. When the information broadcasted by the control center next time is consistent with the sent information, the driving route is determined and the equipment starts to operate; The running path resource maximization algorithm performs the following operations: Receive route information representing crowd density points from the broadcast. When a crowd density point appears on the route to be taken, set the current position to A and the current route to be No. 1; Select n routes from the remaining routes that have the same path segment as route 1 before point A, and perform calculations on these n routes to obtain m routes that do not conflict with the currently occupied path segment; Select the m routes and the one with the least number of conflicts with crowded points among the current routes as the new route; The wheelchair stops moving and exchanges data with the control center, clearing the original path occupancy information and using the path segment of the new route as the path occupancy information; After the data exchange is successful, the driving route is determined and the operation continues.

2. The control method according to claim 1, characterized in that: The step 1 is specifically as follows: Step 1.1: The patient enters the hospital and walks to the smart wheelchair. Once the wheelchair detects someone sitting in it, the voice recognition and announcement module prompts the patient to insert their social security card into the card slot or verbally state their symptoms. Step 1.2: When the patient verbally describes their symptoms, the voice recognition module actively identifies keywords, matches them with keywords in the word library, and then sends the information to the microcontroller. After recognition, the microcontroller drives the corresponding sentence in the voice broadcast module to respond to the patient and determine the target clinic; When a patient inserts his / her social security card, the hospital's medical system will identify the patient's condition requiring follow-up and determine the target clinic.

3. The control method according to claim 1, wherein: The video recognition algorithm performs the following operations: Input the current surveillance video screen; Through the first part of the CSRNet network model: convolutional network for 2D feature extraction; The second part of the CSRNet network model: the dilated convolutional network, is used to reduce the feature vector output by the convolutional layer and improve the results; The label file is processed by Gaussian filtering as the GroundTruth predicted by the model; Output density map.

4. The control method according to claim 1, wherein: Also includes: Step 4: After arriving at the target clinic and completing the diagnosis, if the patient needs to undergo testing, the patient repeats the voice control and the wheelchair automatically takes the patient to the destination for testing; When picking up the test results, the patient inserts the social security card and medical card into the card slot. The system automatically connects to the hospital's medical system and takes the patient to the order collection destination. When the wheelchair is in front of the order collection machine, the wheelchair confirms the patient's identity information through the hospital's medical system and voice confirms the information. After confirming the information, the order collection machine automatically issues the order.

Images