Unmanned bus overcrowding monitoring system and method

Abstract

The invention provides a driverless bus overcrowding monitoring system and method. The driverless bus overcrowding monitoring system comprises a sensing module, a central data processing module and an overcrowding prompting module, the sensing module is used for sensing the passenger state and the seat state to obtain state data; the central data processing module is used for fusing the state data based on the data contribution degree and processing the fused state data based on an overcrowding decision model to obtain an overcrowding decision; and the overcrowding prompting module is used for performing overcrowding prompting based on the overcrowding decision. The overcrowding monitoring system for the unmanned bus is an intelligent system integrating sensing, processing and prompting, the safety and the operation efficiency of the bus can be effectively improved by applying the overcrowding monitoring system, and more convenient and comfortable travel experience is provided for passengers.

Description

Technical Field

[0001] The invention belongs to the technical field of automobile overcrowding monitoring, and in particular relates to an overcrowding monitoring system and method for an unmanned bus. Background Art

[0002] In recent years, driverless cars have been put into use in different regions across the country. The use of unmanned vehicles to carry out passenger transport has become a representative mode of passenger transport at this stage and even in the future because it has the advantages of significantly reducing the employment costs of transport operators and providing passengers with a good riding experience. Considering that the traditional driver and in-vehicle safety officer are eliminated in driverless cars, and there are no other staff in the car, when the number of passengers exceeds the rated passenger capacity, it violates existing laws and regulations, affects driving safety, and affects the normal operation of the vehicle.

[0003] The existing mainstream methods for monitoring overcrowding are as follows: First, remote personnel use in-vehicle cameras to monitor whether there is overcrowding, but there is a problem that it is impossible to accurately determine whether there is overcrowding due to insufficient light at night; second, represented by utility model CN220962591U, the number of people on the vehicle is obtained and whether there is overcrowding is determined by installing information collection modules, communication modules, etc. on the highway, but the cost of modifying road infrastructure is high, the application scenario is limited to highways and is not suitable for urban roads, and the recognition accuracy is easily affected by environmental factors such as driving weather; third, represented by utility model CN209785252, infrared thermal imagers and cameras are installed above the lane to obtain the number of people on the vehicle and determine whether there is overcrowding, which is also costly and easily affected by environmental factors such as driving weather. The recognition accuracy is affected; fourth, represented by utility model CN220962591U, the overcrowding situation is detected by a visible light camera and a single infrared camera, but when there is an error in image recognition, it is impossible to determine whether there is overcrowding.

[0004] In summary, the above-mentioned existing overcrowding monitoring methods have their own technical shortcomings. Second, they do not take into account the current characteristics of driverless buses, such as the need for all passengers to be seated and standing is prohibited. Third, they do not take into account the reason why driverless buses are overcrowded is that passengers decide not to get off the bus after arriving at the station, or the number of passengers who get on the bus after the bus arrives at the station exceeds the number of remaining seats. Therefore, the existing overcrowding monitoring schemes are not suitable for driverless buses with no one in the bus. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a system and method for monitoring overcrowding in an unmanned bus, which can achieve accurate monitoring of the overcrowding situation in an unmanned bus and ensure the safety of passengers.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] An unmanned bus overcrowding monitoring system includes: a sensing module, a central data processing module and an overcrowding prompt module;

[0008] The sensing module is used to sense the passenger status and seat status and obtain status data;

[0009] The central data processing module is used to fuse the status data based on the data contribution, and process the fused status data based on the overcrowding decision model to obtain an overcrowding decision;

[0010] The overcrowding prompt module is used to provide an overcrowding prompt based on the overcrowding decision.

[0011] Preferably, the sensing module includes a door infrared sensor unit, a passenger seating area detection unit, a non-passenger seating area detection unit, and a passenger area seat pressure detection unit;

[0012] The door infrared sensor unit is used to record the number of times passengers get on and off the vehicle;

[0013] The passenger seating area detection unit is used to detect a first seat occupancy status and identify an occupant category;

[0014] The non-passenger seating area detection unit is used to detect the standing situation of passengers in the carriage;

[0015] The passenger area seat pressure detection unit is used to detect a second seat occupancy condition;

[0016] The first seat occupancy status and the second seat occupancy status are weighted to obtain a final seat occupancy status.

[0017] Preferably, the passenger seating area detection unit comprises:

[0018] An image acquisition subunit is used to acquire a seat image of a passenger seating area, and perform enhancement processing on the seat image of the passenger seating area to obtain an enhanced seat image;

[0019] A feature extraction subunit, used for constructing a feature extraction network based on a deformable convolution block and a C2f module, and performing feature extraction on the enhanced seat image based on the feature extraction network to obtain a multi-scale seat feature image;

[0020] A feature enhancement subunit is used to fuse multi-scale seat feature images through a bottom-up path and a top-down path based on the PANet feature enhancement network to obtain a fused feature image;

[0021] An anchor frame allocation subunit, used for adjusting the anchor frame size based on preset requirements, and allocating anchor frames for the fused feature image based on the PAA module;

[0022] A seat occupancy detection unit, configured to establish a detection head by combining a spatial attention mechanism and a convolution operation, and obtain a first seat occupancy status based on a bounding box regression value of an anchor box predicted and assigned by the detection head; wherein the first seat occupancy status includes an occupied seat and an empty seat;

[0023] The seat-occupying object classification unit is used to perform a global average operation on the fused feature image using a global average pooling layer to obtain a feature vector of a preset length; based on the feature vector, the seat-occupying object is classified to obtain an occupied object category; wherein the occupied object category includes seat-occupying passengers and seat-occupying items.

[0024] Preferably, the central data processing module includes a data receiving unit, a data fusion unit and an overcrowding decision unit;

[0025] The data receiving unit is used to receive the status data, wherein the status data includes the number of passengers getting on and off the bus, the final seat occupancy, the category of the occupied object, and the standing status of passengers in the carriage;

[0026] A data analysis unit is used to analyze the number of passengers getting on and off the bus, calculate the number of passengers getting on and off the bus, and the total number of passengers currently in the bus; analyze the final seat occupancy and the category of the occupant to obtain the number of seated passengers and the number of remaining seats; analyze the standing passengers in the bus to obtain the number of standing passengers;

[0027] The data fusion unit is used to fuse the data analysis results to obtain fused data;

[0028] The overcrowding decision unit is used to make an overcrowding decision on the driverless bus based on the fused data and the overcrowding decision model.

[0029] Preferably, the data fusion unit comprises:

[0030] Performing denoising and outlier detection on the data analysis results to obtain processed data analysis results;

[0031] The processed data analysis results are formatted uniformly and timestamped to obtain multi-dimensional time series data;

[0032] Extracting features from the multidimensional time series data to obtain data features;

[0033] Calculating and ranking the contribution of the data features to overcrowding decision-making to obtain data features that meet a preset contribution threshold;

[0034] Based on a deep neural network, a new feature representation is performed on the data features that meet a preset contribution threshold to obtain the fused data.

[0035] Preferably, the overcrowding decision unit comprises:

[0036] A model building unit, used to build the overcrowding decision model using a deep learning algorithm based on the driverless bus type, passenger capacity standard and existing status data set;

[0037] A model calculation unit, used for processing the fusion data to be decided based on the overcrowding decision model to obtain the overcrowding probability and overcrowding degree;

[0038] The decision-making judgment unit is used to judge the overcrowding probability and the overcrowding degree based on the corresponding driverless bus type and passenger carrying standard, and obtain an overcrowding decision.

[0039] Preferably, the overcrowding prompt module includes an overcrowding prompt unit for passengers outside the vehicle, a backstage overcrowding prompt unit, and a bus overcrowding prompt unit;

[0040] The overcrowding prompt unit for passengers outside the vehicle is used to provide an overcrowding prompt to passengers waiting to board the vehicle outside the vehicle based on the overcrowding decision;

[0041] The backstage overcrowding prompt unit is used to provide an overcrowding prompt to the backstage staff of the driverless bus based on the overcrowding decision;

[0042] The bus overcrowding reminder unit is used to provide an overcrowding reminder to the vehicle central control platform of the driverless bus based on the overcrowding decision.

[0043] The present invention also provides a method for monitoring overcrowding in an unmanned bus, which is used to implement the method, comprising the following steps:

[0044] Sense passenger status and seat status and obtain status data;

[0045] Based on the data contribution, the state data is integrated, and the integrated state data is processed based on the overcrowding decision model to obtain an overcrowding decision;

[0046] Based on the overcrowding decision, an overcrowding prompt is given.

[0047] Compared with the prior art, the beneficial effects of the present invention are as follows: through the collaborative work of the perception module and the central data processing module, the system can monitor the passenger capacity of the bus in real time, and immediately issue an early warning when overcrowding is found, effectively avoiding safety hazards caused by overcrowding. The system can accurately determine the passenger capacity of the bus, thereby improving the operating efficiency and service quality of the bus. Through the overcrowding reminder module, the system can promptly inform passengers of the passenger capacity of the bus, help passengers make reasonable travel choices, and avoid the inconvenience and trouble caused by overcrowding. The data and analysis results collected by the system can provide strong support for the intelligent management of bus companies, and help to optimize bus routes, adjust departure intervals and other decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0049] Figure 1 The figure is a schematic diagram of the structure of an overcrowding monitoring system for an unmanned bus according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0052] Embodiment 1

[0053] like Figure 1 As shown, an unmanned bus overcrowding monitoring system includes: a perception module, a central data processing module and an overcrowding prompt module;

[0054] A sensing module is used to sense the passenger status and seat status and obtain status data; a further embodiment is that the sensing module includes a door infrared sensor unit, a passenger seating area detection unit, a non-passenger seating area detection unit and a passenger area seat pressure detection unit;

[0055] The door infrared sensor unit is used to record the number of times passengers get on and off the vehicle; the door infrared sensor is installed at the vehicle door.

[0056] The passenger seating area detection unit is used to detect the first seat occupancy status and identify the category of the occupied object; in this embodiment, the passenger seating area detection unit uses a passenger seating area camera group, which is installed in the car to detect the position of each seat. A further embodiment is that the passenger seating area detection unit includes:

[0057] The image acquisition subunit is used to acquire the seat image of the passenger seating area, and enhance the seat image of the passenger seating area to obtain an enhanced seat image; in particular, the driverless bus may drive during the day, at night, or in a low-light environment during driving, so in a complex lighting environment, the present invention establishes a low-light image enhancement model to process the acquired low-light image and improve the image detection accuracy. The specific implementation steps are as follows:

[0058] A low-light image enhancement model is constructed based on an improved VGG neural network model for capturing feature information of seat images in the passenger seating area and a RetinexNet neural network model for enhancing the background brightness of the image. The VGG neural network model is improved by weight transfer, and the improved VGG neural network model includes convolution kernels and channels as well as a Haar wavelet transform layer. When the VGG neural network model is improved by weight transfer, different weights are set for content loss and style loss, and the image is transferred to the corresponding part of the model according to the weight values ​​of different image losses for image optimization and enhancement. At the same time, multiple iterative trainings are performed to ensure the fusion balance of the image at the content level and the style level.

[0059] The RetinexNet neural network model includes a decomposition network and an enhancement network. The decomposition network uses a five-layer convolutional neural network, which receives a pair of image inputs (low-light and normal images) and outputs illumination and reflection images. Through this step, the model can effectively separate the illumination and reflection components in the image, laying the foundation for subsequent enhancement processing. The enhancement network is a deeper network that uses a nine-layer convolutional structure to further enhance the brightness and quality of the decomposed reflection image. In addition, the network also includes a denoising step to ensure that no additional noise is introduced into the image during the enhancement process, maintaining the natural properties of the image.

[0060] A feature extraction subunit is used to construct a feature extraction network based on a deformable convolution block and a C2f module, and to extract features from the enhanced seat image based on the feature extraction network to obtain a multi-scale seat feature image;

[0061] A feature enhancement subunit is used to fuse multi-scale seat feature images through a bottom-up path and a top-down path based on the PANet feature enhancement network to obtain a fused feature image;

[0062] An anchor frame allocation subunit, used for adjusting the anchor frame size based on preset requirements, and allocating anchor frames for the fused feature image based on the PAA module;

[0063] A seat occupancy detection unit is used to establish a detection head by combining a spatial attention mechanism and a convolution operation, and obtain a first seat occupancy status based on a bounding box regression value of an anchor box predicted and assigned by the detection head; wherein the first seat occupancy status includes an occupied seat and an empty seat;

[0064] The seat-occupying object classification unit is used to perform a global average operation on the fused feature image using a global average pooling layer to obtain a feature vector of a preset length; based on the feature vector, the seat-occupying object is classified to obtain an occupied object category; wherein the occupied object category includes seat-occupying passengers and seat-occupying items.

[0065] The non-passenger seating area detection unit is used to detect the standing situation of passengers in the vehicle compartment; in this embodiment, the non-passenger seating area detection unit adopts a non-passenger seating area camera group installed in the vehicle.

[0066] The passenger area seat pressure detection unit is used to detect the second occupancy status of the seat; in this embodiment, the passenger area seat pressure detection unit adopts a passenger area seat pressure sensor installed under each seat.

[0067] The first seat occupancy status and the second seat occupancy status are weighted by preset weights to obtain the final seat occupancy status. In the present invention, in order to overcome the detection defects that may be caused by simple image acquisition in the prior art, a seat pressure sensor is installed, and the real-time data collected by the pressure sensor is weighted with the image target detection result to obtain the final seat occupancy status, thereby improving the accuracy of seat occupancy detection.

[0068] The central data processing module is used to fuse the status data based on the data contribution, and process the fused status data based on the overcrowding decision model to obtain the overcrowding decision;

[0069] A further embodiment is that the central data processing module includes a data receiving unit, a data fusion unit and an overcrowding decision unit;

[0070] A data receiving unit, used to receive status data, wherein the status data includes the number of passengers getting on and off the bus, the final seat occupancy, the category of the occupied object, and the standing status of passengers in the carriage;

[0071] The data analysis unit is used to analyze the number of passengers getting on and off the bus, calculate the number of passengers getting on and off the bus, and the total number of passengers in the bus; analyze the final seat occupancy and the type of occupants, obtain the number of seated passengers and the number of remaining seats; analyze the standing passengers in the car, and obtain the number of standing passengers;

[0072] A data fusion unit, used for fusing data analysis results to obtain fused data;

[0073] The overcrowding decision unit is used to make overcrowding decisions for driverless buses based on fused data and an overcrowding decision model.

[0074] A further implementation is that the data fusion unit comprises:

[0075] The data analysis results are de-noised and outliers are detected to obtain the processed data analysis results; specifically, smoothing is used to remove noise in the data due to various reasons (such as measurement errors, transmission errors, etc.) to improve the accuracy and reliability of the data. The isolation forest algorithm is used to identify and process outliers in the data (i.e., values ​​that are significantly different from most data points). These outliers may be caused by erroneous records, data corruption, or special events.

[0076] The processed data analysis results are formatted in a unified manner and timestamped to obtain multidimensional time series data; specifically, a unique timestamp is assigned to each data point to ensure that the data can be arranged and analyzed in chronological order.

[0077] Fourier transform is used to extract features from multidimensional time series data to obtain data features;

[0078] The gradient boosting tree is used to calculate and sort the contribution of data features to overcrowding decisions, and obtain data features that meet the preset contribution threshold. Specifically, according to the contribution calculation results, the features are sorted from high to low, and the features that meet the preset contribution threshold are screened out.

[0079] Based on the deep neural network, a new feature representation is performed on the data features that meet the preset contribution threshold to obtain fused data. Specifically, the convolutional neural network is used to encode and decode the screened features to learn new feature representations. The learned new feature representations are spliced ​​and fused with the original data features to obtain the spliced ​​data of the analysis results of each category of data, and the spliced ​​data of each category are spliced ​​into a feature vector to obtain the final fused data.

[0080] A further implementation method is that the overcrowding decision unit includes:

[0081] The model building unit is used to build an overcrowding decision model based on the type of driverless buses, passenger standards, and existing status data sets using a deep learning algorithm; specifically, the convolutional layer of the convolutional neural network is used to extract local features from the fused data, the ReLU nonlinear activation function is introduced in the activation layer, and the pooling layer downsamples the output of the convolutional layer to reduce the number of parameters and improve computational efficiency. The outputs of the convolutional layer and the pooling layer are converted into feature vectors of a preset length. A recurrent neural network is used to receive the feature vectors output by the convolutional neural network, as well as time series data (such as bus running time, changes in the number of passengers over time, etc.), and LSTM is used to process the time series data to capture the dynamic characteristics and dependencies in the data. According to the preset task requirements, the predicted value of the overcrowding probability or overcrowding degree is output.

[0082] The outputs of the convolutional neural network and the recurrent neural network are weighted and fused based on preset weights, and an additional fully connected layer is introduced on the fused feature vector to further extract and integrate features. The fully connected layer performs nonlinear transformation on the fused features to extract higher-level feature representations. The weights and biases of the fully connected layer are optimized through the back-propagation algorithm during the training process.

[0083] A model calculation unit, used for processing the fusion data to be decided based on the overcrowding decision model to obtain the overcrowding probability and overcrowding degree;

[0084] The decision-making judgment unit is used to judge the probability and degree of overcrowding based on the corresponding driverless bus type and passenger capacity standard, and obtain an overcrowding decision. Specifically, the probability and degree of overcrowding output by the model calculation unit are compared with the set threshold. If the probability of overcrowding exceeds the threshold or the degree of overcrowding reaches or exceeds the preset standard, it is judged as overcrowding. Specifically, for example, outside the seating area, as long as there are people standing, it is considered overcrowded; if all unmanned standing seats are occupied, it is judged to be about to be overcrowded.

[0085] The overcrowding prompt module is used to provide overcrowding prompts based on overcrowding decisions.

[0086] A further implementation method is that the overcrowding prompt module includes an overcrowding prompt unit for passengers outside the vehicle, a backstage overcrowding prompt unit, and a bus overcrowding prompt unit;

[0087] The overcrowding reminder unit for passengers outside the vehicle is used to remind passengers waiting to get on the vehicle based on the overcrowding decision; the overcrowding reminder for passengers outside the vehicle includes using sound, light, electrical signals, etc. at the door to remind and warn passengers outside the vehicle who still want to get on the vehicle when the vehicle is full.

[0088] The background overcrowding reminder unit is used to remind the background staff of driverless buses of overcrowding based on overcrowding decisions; 5G transmission and other methods are used to remind relevant personnel in the company's background that the vehicle is overcrowded.

[0089] The bus overcrowding reminder unit is used to give an overcrowding reminder to the vehicle central control platform of the driverless bus based on the overcrowding decision. The overcrowding reminder for driverless cars includes transmitting the overcrowding information to the vehicle central control platform through CAN signals, wireless LAN transmission, etc., and passing it to the driverless system to execute control instructions such as inhibiting the vehicle from starting and running, and keeping the door open, until the passengers in the car return to normal.

[0090] The data and analysis results collected by the system can provide strong support for the intelligent management of bus companies, and help make decisions such as optimizing bus routes and adjusting departure intervals.

[0091] On the other hand, the present invention also provides a method for monitoring overcrowding in an unmanned bus, which is used to implement the method, comprising the following steps:

[0092] Sense passenger status and seat status and obtain status data;

[0093] Based on the data contribution, the status data is integrated, and the integrated status data is processed based on the overcrowding decision model to obtain the overcrowding decision;

[0094] Based on the overcrowding decision, overcrowding prompts are given.

[0095] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. An unmanned bus overcrowding monitoring system, characterized in that: include: Perception module, central data processing module and overcrowding warning module; The sensing module is used to sense the passenger status and seat status and obtain status data; The central data processing module is used to fuse the status data based on the data contribution, and process the fused status data based on the overcrowding decision model to obtain an overcrowding decision; The overcrowding prompt module is used to provide an overcrowding prompt based on the overcrowding decision.

2. The driverless bus overcrowding monitoring system according to claim 1 is characterized in that: The sensing module includes a door infrared sensor unit, a passenger seating area detection unit, a non-passenger seating area detection unit, and a passenger area seat pressure detection unit; The door infrared sensor unit is used to record the number of times passengers get on and off the vehicle; The passenger seating area detection unit is used to detect a first seat occupancy status and identify an occupant category; The non-passenger seating area detection unit is used to detect the standing situation of passengers in the carriage; The passenger area seat pressure detection unit is used to detect a second seat occupancy condition; The first seat occupancy status and the second seat occupancy status are weighted to obtain a final seat occupancy status.

3. The driverless bus overcrowding monitoring system according to claim 2 is characterized in that: The passenger seating area detection unit comprises: An image acquisition subunit is used to acquire a seat image of a passenger seating area, and perform enhancement processing on the seat image of the passenger seating area to obtain an enhanced seat image; A feature extraction subunit, used for constructing a feature extraction network based on a deformable convolution block and a C2f module, and performing feature extraction on the enhanced seat image based on the feature extraction network to obtain a multi-scale seat feature image; The feature enhancement subunit is used to fuse the multi-scale seat feature images through a bottom-up path and a top-down path based on the PANet feature enhancement network to obtain a fused feature image; An anchor frame allocation subunit, used for adjusting the anchor frame size based on preset requirements, and allocating anchor frames for the fused feature image based on the PAA module; A seat occupancy detection unit, configured to establish a detection head by combining a spatial attention mechanism and a convolution operation, and obtain a first seat occupancy status based on a bounding box regression value of an anchor box predicted and assigned by the detection head; wherein the first seat occupancy status includes an occupied seat and an empty seat; The seat-occupying object classification unit is used to perform a global average operation on the fused feature image using a global average pooling layer to obtain a feature vector of a preset length; based on the feature vector, the seat-occupying object is classified to obtain an occupied object category; wherein the occupied object category includes seat-occupying passengers and seat-occupying items.

4. The driverless bus overcrowding monitoring system according to claim 2 is characterized in that: The central data processing module includes a data receiving unit, a data fusion unit and an overcrowding decision unit; The data receiving unit is used to receive the status data, wherein the status data includes the number of passengers getting on and off the bus, the final seat occupancy, the category of the occupied object, and the standing status of passengers in the carriage; A data analysis unit is used to analyze the number of passengers getting on and off the bus, calculate the number of passengers getting on and off the bus, and the total number of passengers currently in the bus; analyze the final seat occupancy and the category of the occupant to obtain the number of seated passengers and the number of remaining seats; analyze the standing passengers in the bus to obtain the number of standing passengers; The data fusion unit is used to fuse the data analysis results to obtain fused data; The overcrowding decision unit is used to make an overcrowding decision on the driverless bus based on the fused data and the overcrowding decision model.

5. The driverless bus overcrowding monitoring system according to claim 4 is characterized in that: The data fusion unit comprises: Performing denoising and outlier detection on the data analysis results to obtain processed data analysis results; The processed data analysis results are formatted uniformly and timestamped to obtain multi-dimensional time series data; Extracting features from the multidimensional time series data to obtain data features; Calculating and ranking the contribution of the data features to overcrowding decision-making to obtain data features that meet a preset contribution threshold; Based on a deep neural network, a new feature representation is performed on the data features that meet a preset contribution threshold to obtain the fused data.

6. The driverless bus overcrowding monitoring system according to claim 5, characterized in that: The overcrowding decision unit comprises: A model building unit, used to build the overcrowding decision model using a deep learning algorithm based on the driverless bus type, passenger capacity standard and existing status data set; A model calculation unit, used for processing the fusion data to be decided based on the overcrowding decision model to obtain the overcrowding probability and overcrowding degree; The decision-making judgment unit is used to judge the overcrowding probability and the overcrowding degree based on the corresponding driverless bus type and passenger carrying standard, and obtain an overcrowding decision.

7. The driverless bus overcrowding monitoring system according to claim 2, characterized in that: The overcrowding prompt module includes an overcrowding prompt unit for passengers outside the vehicle, a backstage overcrowding prompt unit, and a bus overcrowding prompt unit; The overcrowding prompt unit for passengers outside the vehicle is used to provide an overcrowding prompt to passengers waiting to board the vehicle outside the vehicle based on the overcrowding decision; The backstage overcrowding prompt unit is used to provide an overcrowding prompt to the backstage staff of the driverless bus based on the overcrowding decision; The bus overcrowding reminder unit is used to provide an overcrowding reminder to the vehicle central control platform of the driverless bus based on the overcrowding decision.

8. A method for monitoring overcrowding in an unmanned bus, used to implement the method described in any one of claims 1 to 7, characterized in that: The following steps are involved: Sense passenger status and seat status and obtain status data; Based on the data contribution, the state data is integrated, and the integrated state data is processed based on the overcrowding decision model to obtain an overcrowding decision; Based on the overcrowding decision, an overcrowding prompt is given.

Images