An adaptive sweeper with a pneumatic conveying system

Abstract

The present invention relates to an adaptive sweeper with a pneumatic conveying system, comprising: a vehicle body, wherein the vehicle body is provided with a camera group and an odor sensor group; a cleaning mechanism, which is arranged at the bottom of the vehicle body and is driven by a transverse driving member to move forward or backward along the driving direction of the vehicle body; a control system, wherein the control system obtains road pollutant image data at different angles through the camera group, obtains a first pollutant type and a pollution degree thereof through a first fused image recognition, fuses the first fused image with the road pollutant odor data according to the first pollution type to obtain a fused data feature, and obtains a second pollutant type and a pollution degree thereof through the fused data feature judgment, and if the first pollutant type and the second pollutant type are different and any of the pollution degrees is greater than a preset pollution threshold, then the contact time between the cleaning mechanism and the road pollutant is increased; and a pneumatic conveying system.

Description

Technical Field

[0001] The present invention relates to the field of self-adaptive washing and sweeping vehicles, and in particular to a self-adaptive washing and sweeping vehicle with a pneumatic conveying system. Background Art

[0002] A road sweeper is a road cleaning and maintenance vehicle within the road sweeper family. It performs multiple functions, including road cleaning, curb and curbstone scrubbing, low-pressure flushing, and spray dust removal. It typically includes a sweeping brush, a high-pressure nozzle, and a suction mechanism. The rotating sweeping brush cleans the road surface, while the high-pressure nozzle flushes away stubborn stains. The suction mechanism then collects road debris and waste liquid from the flushing process into a trash bin.

[0003] A road sweeper typically operates by an operator driving the vehicle across the road, scrubbing the surface. Traditional road sweepers typically utilize fixed brushes and high-pressure nozzles, positioned on either side of the vehicle. These brushes and nozzles are lowered or retracted by driving them longitudinally. Due to the wide variety of trash and stains on the road, small, lightweight items like fallen leaves and cigarette butts are easily swept and picked up by the sweeper. However, items like grease, candy, stickers, and mixtures of these items are difficult to remove due to their stickiness. Operators typically operate the sweeper at a constant speed, with the brushes and nozzles maintaining a fixed cleaning time with the road surface. Difficult-to-clean stains like grease, candy, and stickers are often tightly bonded to or mixed with other trash and the road surface. Furthermore, these items are often small, making them difficult for the operator or the sweeper's visual recognition system to detect. In the prior art, for example, the existing application with publication number CN115542735A, entitled Adaptive Control Method for the Operation Speed ​​of a Washing and Sweeping Vehicle Based on Machine Vision, controls the cleaning operation speed of the cleaning components through vision, but does not adaptively adjust the cleaning time between the sweeping brush and the high-pressure nozzle and the road surface according to the type of garbage, the degree of pollution, etc., resulting in problems such as incomplete cleaning.

[0004] The purpose of this invention is to design an adaptive sweeper with a pneumatic conveying system to solve the above problems in the prior art. Summary of the Invention

[0005] In response to the problems existing in the above-mentioned prior art, the present invention provides an adaptive sweeper with a pneumatic conveying system, which can effectively solve at least one problem existing in the above-mentioned prior art.

[0006] The technical solution of the present invention is:

[0007] An adaptive sweeper with a pneumatic conveying system, comprising:

[0008] A vehicle body, wherein the vehicle body is provided with a camera group and an odor sensor group;

[0009] A cleaning mechanism is provided at the bottom of the vehicle body, and is driven by a transverse driving member to move forward or backward along the travel direction of the vehicle body;

[0010] a control system connected to the camera group, the odor sensor group, and the transverse drive member, the control system acquiring road pollutant image data at different angles through the camera group, fusing the road pollutant image data at different angles to obtain a first fused image, identifying a first pollutant type and its pollution degree through the first fused image, acquiring road pollutant odor data, fusing the first fused image with the road pollutant odor data based on the first pollution type to obtain a fused data feature, determining a second pollutant type and its pollution degree through the fused data feature, and if the first pollutant type and the second pollutant type are different and either pollution degree is greater than a preset pollution threshold, controlling the transverse drive member to drive the cleaning mechanism to retreat along the travel direction of the vehicle body, thereby increasing the contact time between the cleaning mechanism and the road pollutants;

[0011] The pneumatic conveying system is arranged at the bottom of the rear end of the vehicle body and is used to suck up the pollutants on the road surface after being cleaned by the cleaning mechanism.

[0012] Furthermore, the camera group includes at least a front camera, a left camera, and a right camera; and the odor sensor group is arranged at the bottom of the vehicle body.

[0013] Furthermore, obtaining road surface pollutant image data at different angles, fusing the road surface pollutant image data at different angles to obtain a first fused image, and identifying the first pollutant type and its pollution degree through the first fused image include:

[0014] Acquire a front image, a left image, and a right image at the same time point, and fuse the front image, the left image, and the right image at the same time point to obtain a first fused image;

[0015] The first fused image is input into a pre-trained first pollutant detection model to obtain a first pollutant type and a pollution degree thereof. The pollutant types include: solid waste, particulate dust, and liquid pollutants. The pollution degrees include: high pollution, moderate pollution, and low pollution.

[0016] Furthermore, fusing the front image, the left image, and the right image at the same time point to obtain a first fused image includes:

[0017] The deep features of the front image, left image, and right image are extracted respectively through the pre-trained convolutional neural network;

[0018] The deep features of the front image, left image, and right image are weightedly fused using the attention mechanism, and the decoder is used to reconstruct the weighted fused features into the first fused image.

[0019] Furthermore, road pollutant odor data is obtained, and the first fused image and the road pollutant odor data are fused according to the first pollution type to obtain fused data features including:

[0020] Extract visual features of the first fused image through a pre-trained convolutional neural network ;

[0021] Input the road pollution odor data into the long short-term memory network model to extract the odor characteristics ;

[0022] Use the embedded features of the first pollutant type as the query vector , visual features are calculated separately through the attention mechanism and odor characteristics The corresponding weight is calculated as follows:

[0023] ,

[0024] ,

[0025] in, For visual features The corresponding weight, Odor characteristics The corresponding weight, is the attention mechanism function;

[0026] According to visual features and odor characteristics The corresponding weights will be visual features and odor characteristics Fusion is performed to obtain fused data features , the calculation formula is as follows:

[0027] .

[0028] Furthermore, determining the second pollutant type and its pollution degree by fusing the data features includes: inputting the fused data features into a pre-trained second pollutant detection model to obtain the second pollutant type and its pollution degree.

[0029] Furthermore, controlling the transverse driving member to drive the cleaning mechanism to move backward along the traveling direction of the vehicle body to increase the contact time between the cleaning mechanism and the road surface pollutants includes:

[0030] The type of pollutant, the degree of pollution, and the speed adjustment of the cleaning mechanism are used as state variables, and the time required for the cleaning mechanism to clean the road pollutants is used as the action variable to construct a cleaning time reinforcement learning model.

[0031] A reward function is constructed by cleaning the effect value. When the reward value of the reward function is higher than the reward threshold, it is marked as a positive reward signal, otherwise it is marked as a negative reward signal. The reinforcement learning model is optimized according to the reward signal.

[0032] Inputting the current first pollutant type and the second pollutant type and the corresponding pollution degree into the cleaning time reinforcement learning model to obtain the cleaning time required;

[0033] The driving speed of the transverse driving member to drive the cleaning mechanism backward along the driving direction of the vehicle body is controlled so that the contact time between the cleaning mechanism and the road surface pollutants meets the required cleaning time when the vehicle body maintains the driving speed.

[0034] Furthermore, when the vehicle body maintains the driving speed, if the maximum contact time between the cleaning mechanism and the road pollutants is less than the time required for cleaning, an alarm signal is issued.

[0035] Furthermore, the cleaning mechanism includes a sweeping brush and a high-pressure nozzle.

[0036] Therefore, the present invention provides the following effects and / or advantages:

[0037] This application fuses images of road pollutants from different perspectives (such as the front, left, and right) to form a complete description of scene characteristics. Image fusion retains key information from each perspective, more comprehensively reflects the shape, color, and distribution characteristics of pollutants, and reduces the risk of occlusion or information loss. Inputting the fused first fused image into a pre-trained pollutant detection model can effectively extract the deep characteristics of the pollutants. Accurately identify the type of the first pollutant (such as solid waste, liquid pollutants, particulate dust) and its degree of pollution. Therefore, by controlling the cleaning mechanism that can move forward or backward in the direction of travel of the vehicle, it can maintain or slowly pass through the road pollutants, thereby increasing the cleaning time and improving the cleaning effect.

[0038] This application overcomes the limitations of single-modal data by fusing image and odor data. Images provide information on the spatial distribution, appearance, shape, and occlusion effects of pollutants, while odors complement chemical properties (such as toxicity, harmfulness, or corrosiveness). Multimodal fusion enables a more comprehensive representation of pollutant characteristics, providing a more accurate reference for pollution identification and cleaning strategy generation. Based on the characteristics of the first pollutant type, the fusion weights of image and odor data are dynamically adjusted to ensure a matching of importance.

[0039] This application can provide richer feature judgments for second pollutant detection by fusing data features, helping the model to further explore hidden information from a single type of pollutant.

[0040] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description and the drawings.

[0041] It is to be understood that both the foregoing general description and the following detailed description of the present invention are exemplary and explanatory and are intended to provide further explanation of the invention as claimed. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 Schematic diagram of the structure of an embodiment of the present invention.

[0043] Description of reference numerals:

[0044] Vehicle body 1, cleaning mechanism 2, transverse driving component 3, pneumatic conveying system 4. DETAILED DESCRIPTION

[0045] In order to facilitate understanding by those skilled in the art, the structure of the present invention is further described in detail with reference to the embodiments and the accompanying drawings:

[0046] refer to Figure 1 , an adaptive sweeper with a pneumatic conveying system, comprising:

[0047] A vehicle body 1, wherein the vehicle body 1 is provided with a camera group and an odor sensor group;

[0048] A cleaning mechanism 2 is provided at the bottom of the vehicle body 1 and is driven by a transverse driving member 3 so as to move forward or backward along the travel direction of the vehicle body 1;

[0049] In this embodiment, the vehicle body 1 can be directly adopted from existing technology. The vehicle body 1 serves as a driving force for the cleaning mechanism 2 and other components to travel on the road. Furthermore, the vehicle body 1 is also the power source for the cleaning mechanism 2 and the transverse drive member 3. One of the core improvements of this embodiment is that the cleaning mechanism 2 is driven by the transverse drive member 3, enabling forward or backward movement. If the transverse drive member 3 drives the cleaning mechanism 2 backward while the vehicle is traveling forward, the speed of the cleaning mechanism 2 relative to the road surface is less than the speed of the vehicle relative to the road surface. This increases the contact time between the cleaning mechanism 2 and the road surface, thereby improving the cleaning effect of the cleaning mechanism 2.

[0050] a control system connected to the camera group, the odor sensor group, and the transverse drive member 3, wherein the control system acquires road pollutant image data from different angles through the camera group, fuses the road pollutant image data from different angles to obtain a first fused image, identifies a first pollutant type and its pollution degree through the first fused image, acquires road pollutant odor data, fuses the first fused image with the road pollutant odor data according to the first pollution type to obtain a fused data feature, determines a second pollutant type and its pollution degree through the fused data feature, and if the first pollutant type and the second pollutant type are different and either pollution degree is greater than a preset pollution threshold, controls the transverse drive member 3 to drive the cleaning mechanism 2 to retreat along the travel direction of the vehicle body 1, thereby increasing the contact time between the cleaning mechanism 2 and the road pollutants;

[0051] As mentioned in the background technology, grease, candy, stickers, and mixtures of these with other garbage are difficult to clean with a cleaning vehicle due to their stickiness. Furthermore, these grease, candy, stickers, and the like are small in size and easily stick to the road surface. It is difficult for the driver of vehicle 1 or the control system of vehicle 1 to recognize that these stubborn stains are among the road surface contaminants, and thus it is impossible to adaptively adjust the contact time between cleaning mechanism 2 and the road surface contaminants. This embodiment fuses the road surface contaminant images from a camera group to form a complete scene characteristic description, and then inputs the fused first fused image into a pre-trained contaminant detection model. This effectively extracts the deep features of the contaminants and accurately identifies the type of the first contaminant and its degree of contamination. Then, by fusing the odor data of road pollutants, such as grease, candy, stickers, etc. with special odors, the type of pollutants and their degree of pollution in the road pollutants after combining images and odors can be identified, so that it can be known whether the influence of stubborn stains on road pollutants is large. If it is large, it will affect the second pollutant type and its degree of pollution obtained by judging the fusion data characteristics, thereby obtaining a second pollutant type different from the first pollutant type. For example, if a pile of leaves contains a large amount of grease, the pollutant type calculated by the first fused image is leaves, and the pollutant type calculated by the second pollutant type may be grease; or if a pile of leaves contains a small amount of grease, the pollutant type calculated by the first fused image is leaves, and the pollutant type calculated by the second pollutant type may be leaves; therefore, by judging the pollutant types of the two, the proportion of stubborn stains can be identified.

[0052] Some pollutants, such as chemical liquids, oil stains, or decomposed garbage residues, may be more easily detected by smell alone. Odor sensors can provide important supplemental perception information by sensing the volatile chemical gases emitted by pollutants. Combining images and odors can more comprehensively identify the type and degree of pollution. Odor sensor groups are mainly used to detect volatile chemical gases or characteristic odors of road pollutants. Therefore, odor sensors can use a combination of multiple odor sensors, such as electrochemical gas sensors (which can detect ammonia and hydrogen sulfide gases produced by food spoilage in the market), metal oxide semiconductor gas sensors (which can detect volatile gases from chemical liquids and oil-based pollutants on the road), and PID photoionization gas sensors (which can detect volatile gases from paint, chemical cleaning agents, oil stains, etc.).

[0053] As mentioned above, the ability to calculate the first and second pollutant types allows for dynamic adjustment of the cleaning time based on whether these types match and the degree of contamination. Specifically, the cleaning mechanism 2 is driven forward or backward along the direction of travel of the vehicle 1 by a transverse drive 3. This allows the transverse drive 3 to extend rearward when the vehicle is traveling forward, extending the cleaning mechanism 2 rearward and prolonging its contact time with contaminants on the road surface. This allows the contaminants to be swept off the road surface and subsequently picked up by the vehicle's pneumatic conveying system 4.

[0054] The pneumatic conveying system 4 is arranged at the bottom of the rear end of the vehicle body 1 and is used to suck up road pollutants cleaned by the cleaning mechanism 2 .

[0055] Furthermore, the camera group includes at least a front camera, a left camera, and a right camera; and the odor sensor group is arranged at the bottom of the vehicle body 1.

[0056] Furthermore, obtaining road surface pollutant image data at different angles, fusing the road surface pollutant image data at different angles to obtain a first fused image, and identifying the first pollutant type and its pollution degree through the first fused image include:

[0057] Acquire a front image, a left image, and a right image at the same time point, and fuse the front image, the left image, and the right image at the same time point to obtain a first fused image;

[0058] The first fused image is input into a pre-trained first pollutant detection model to obtain a first pollutant type and a pollution degree thereof. The pollutant types include: solid waste, particulate dust, and liquid pollutants. The pollution degrees include: high pollution, moderate pollution, and low pollution.

[0059] In this step, image data from a single angle may have blind spots or incomplete information (e.g., due to occlusion or lighting), leading to inaccurate pollutant detection. By acquiring road surface image data from multiple angles, including the front, left, and right sides, and using deep feature extraction and an attention mechanism for weighted fusion, a first fused image with a global view is generated. Image data from different angles may have different viewpoints, resolutions, or inconsistent features, and direct fusion may result in information loss or redundancy. Deep features are extracted using a pretrained convolutional neural network and weighted fused using an attention mechanism to highlight key features and suppress irrelevant information.

[0060] Traditional methods struggle to simultaneously accurately identify multiple pollutant types (solid waste, particulate matter, and liquid pollutants) and their respective levels (high, medium, and low), especially in complex scenarios. This solution uses a pre-trained YOLO model as the primary pollutant detection model. Through training, the YOLO model can distinguish between solid waste, particulate matter, and liquid pollutants, and returns the results as specific pollutant types using a class confidence score.

[0061] When detecting pollutants, the YOLO model will output a bounding box of the target, which describes the rectangular range of the area where the pollutant is located. The total number of pixels in the bounding box The number of pixels covered by actual pollutants It can be used as an indicator to determine the degree of pollution. The calculation formula is: Specific: Highly polluted: Percentage of polluted pixels ≥80%, moderate pollution: 60%≤proportion of polluted pixels <80%, low pollution: percentage of polluted pixels <60%.

[0062] The ratio of polluted pixels to bounding box pixels is used as the basis for judgment, avoiding direct reliance on the absolute value of the polluted area. This is because the extent of road pollution can be affected by viewing angle and distance, while the ratio is relatively stable. Whether it is solid waste (evenly distributed lumps) or liquid pollutants (which may have irregular shapes or be scattered), the ratio as a criterion is universal.

[0063] Furthermore, fusing the front image, the left image, and the right image at the same time point to obtain a first fused image includes:

[0064] The deep features of the front image, left image, and right image are extracted respectively through the pre-trained convolutional neural network;

[0065] The deep features of the front image, left image, and right image are weightedly fused using the attention mechanism, and the decoder is used to reconstruct the weighted fused features into the first fused image.

[0066] Furthermore, road pollutant odor data is obtained, and the first fused image and the road pollutant odor data are fused according to the first pollution type to obtain fused data features including:

[0067] Extract visual features of the first fused image through a pre-trained convolutional neural network ;

[0068] Input the road pollution odor data into the long short-term memory network model to extract the odor characteristics ;

[0069] Use the embedded features of the first pollutant type as the query vector , visual features are calculated separately through the attention mechanism and odor characteristics The corresponding weight is calculated as follows:

[0070] ,

[0071] ,

[0072] in, For visual features The corresponding weight, Odor characteristics The corresponding weight, is the attention mechanism function;

[0073] According to visual features and odor characteristics The corresponding weights will be visual features and odor characteristics Fusion is performed to obtain fused data features , the calculation formula is as follows:

[0074] .

[0075] In this step, the data type collected by the odor sensor is usually an analog or digital quantity reflecting the characteristics of a specific gas. These odor data are usually time series data formed by sampling from multiple sensors at different time points. LSTM (long short-term memory network model) is good at processing this time series and can capture the dynamic characteristics of pollutant odors.

[0076] The importance of fusing visual features and odor features varies depending on the type of pollutant. In pollutant detection tasks involving multimodal features (visual and odor), different types of pollutants rely on visual and odor features to different degrees. For example, for solid garbage (such as paper scraps, plastic bottles), the visual features are very obvious (shape, outline, color), and the visual features contribute more to recognition, while the odor features may be irrelevant and have a lower weight. For liquid pollutants (such as oil stains, sewage), vision may only capture the area of ​​the pollutant, while the odor features supplement its composition information (such as whether there is a foul smell), and the odor features contribute more to judging the type of pollutant and the degree of pollution. In the attention mechanism, the query vector It is possible to dynamically focus on which parts of the visual and odor features are more important to this type of pollutant, thereby improving the accuracy of the fused data.

[0077] According to visual features and odor characteristics The corresponding weights will be visual features and odor characteristics Fusion is performed to obtain fused data features In the process, the corresponding weights are increased or decreased according to the detected images and odors. For example, for garbage mixed with leaves and grease, when a small amount of grease odor is detected in the leaves, the weight of the odor can be slightly increased or kept unchanged, so that the final fusion data feature is mainly based on leaves; or when a large amount of grease odor is detected in the leaves, the weight of the odor can be greatly increased, so that the final fusion data feature is mainly based on grease.

[0078] Furthermore, determining the second pollutant type and its pollution degree by fusing the data features includes: inputting the fused data features into a pre-trained second pollutant detection model to obtain the second pollutant type and its pollution degree.

[0079] Furthermore, controlling the transverse driving member 3 to drive the cleaning mechanism 2 to move backward along the traveling direction of the vehicle body 1 to increase the contact time between the cleaning mechanism 2 and the road surface pollutants includes:

[0080] The type of pollutant, the degree of pollution, and the speed adjustment of the cleaning mechanism 2 are used as state variables, and the time required for the cleaning mechanism 2 to clean the road pollutants is used as the action variable to build a cleaning time reinforcement learning model.

[0081] A reward function is constructed by cleaning the effect value. When the reward value of the reward function is higher than the reward threshold, it is marked as a positive reward signal, otherwise it is marked as a negative reward signal. The reinforcement learning model is optimized according to the reward signal.

[0082] Inputting the current first pollutant type and the second pollutant type and the corresponding pollution degree into the cleaning time reinforcement learning model to obtain the cleaning time required;

[0083] The driving speed of the transverse driving member 3 to drive the cleaning mechanism 2 backward along the driving direction of the vehicle body 1 is controlled so that the contact time between the cleaning mechanism 2 and the road surface pollutants meets the required cleaning time when the vehicle body 1 maintains the driving speed.

[0084] In this step, different pollutant types, pollution levels, etc. can be input as state variables, and the time required for the cleaning mechanism 2 to clean road pollutants can be used as action variables. For example, if the pollutant type is leaves mixed with grease, the pollution level is light, and the rotation speed of the sweeping brush during cleaning is 30r / min, then the cleaning mechanism 2 needs to clean for 10s, etc. A large amount of existing data can be used as training data for the reinforcement learning model, so as to obtain the ability to automatically output the required cleaning time according to the pollutant type, pollution level, and the operating speed adjustment amount of the cleaning mechanism 2.

[0085] Then, simply relying on the cleaning effect (such as the reduction of pollution level) may ignore the efficiency problem, while simply relying on efficiency (such as cleaning time) may sacrifice the cleaning effect. A reward function R based on cleaning effect and cleaning time is constructed, and the weights and Balancing the priorities of cleaning effectiveness and time. Cleaning effectiveness assessment relies on comparing the contamination level before and after cleaning, which requires extraction from multi-view images (rear, left, and right). Using image fusion technology, the rear, left, and right images are fused into a second fused image. This image is then input into a pre-trained first contaminant detection model (YOLO model) to determine the post-cleaning contamination level. This multi-view fusion reduces occlusions or omissions that may occur with a single viewpoint, improving the accuracy of contamination assessment.

[0086] Then, the required cleaning time can be automatically output based on the identified first pollutant type and second pollutant type and the corresponding pollution degree.

[0087] Finally, the output speed of the transverse drive member 3 is adjusted based on the required cleaning time. For example, if the vehicle body 1 is moving forward at 0.05 m / s and the cleaning time is 3 seconds, the transverse drive member 3 will continue to output backward at 0.05 m / s for 3 seconds, keeping the cleaning mechanism 2 above the road surface contaminants for 3 seconds, and then the transverse drive member 3 will retract.

[0088] Furthermore, when the vehicle body 1 maintains the driving speed, if the maximum contact time between the cleaning mechanism 2 and the road pollutants is less than the cleaning time required, an alarm signal is issued.

[0089] Since the output length of the transverse drive member 3 is limited, if the time required for cleaning cannot be met after the transverse drive member 3 is fully extended, the vehicle body 1 needs to stop driving. At this time, an alarm signal can be issued to remind the driver to stop for a certain time.

[0090] Furthermore, the cleaning mechanism 2 includes a sweeping brush and a high-pressure nozzle.

[0091] It should be noted that in the claims, any reference signs placed between parentheses shall not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claim. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. The invention can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the words first, second, third etc. does not indicate any order. These words may be interpreted as names.

[0092] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0093] In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "connect," "fixed," etc. should be understood broadly. For example, they may refer to fixed connection, detachable connection, or integration; mechanical connection or electrical connection; direct connection or indirect connection through an intermediate medium; internal communication between two components or interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0094] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials, or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms should not be understood as necessarily referring to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

Claims

1. An adaptive sweeper with a pneumatic conveying system, characterized by: include: A vehicle body (1), wherein the vehicle body (1) is provided with a camera group and an odor sensor group; A cleaning mechanism (2) is arranged at the bottom of the vehicle body (1), and the cleaning mechanism (2) is driven by a transverse driving member (3) to move forward or backward along the travel direction of the vehicle body (1); A control system is connected to the camera group, the odor sensor group, and the transverse driving member (3), wherein the control system obtains road pollutant image data at different angles through the camera group, fuses the road pollutant image data at different angles to obtain a first fused image, and obtains a first pollutant type and its pollution degree through the first fused image identification; the control system obtains road pollutant odor data, fuses the first fused image with the road pollutant odor data according to the first pollution type to obtain a fused data feature, and obtains a second pollutant type and its pollution degree through the fused data feature identification; if the first pollutant type and the second pollutant type are different and any of the pollution degrees is greater than a preset pollution threshold, the transverse driving member (3) is controlled to drive the cleaning mechanism (2) to retreat along the driving direction of the vehicle body (1), thereby increasing the contact time between the cleaning mechanism (2) and the road pollutants; Obtaining road pollutant odor data, fusing the first fused image and the road pollutant odor data according to the first pollution type, and obtaining fused data features including: Extract visual features of the first fused image through a pre-trained convolutional neural network ; Input the road pollution odor data into the long short-term memory network model to extract the odor characteristics ; Use the embedded features of the first pollutant type as the query vector , visual features are calculated separately through the attention mechanism and odor characteristics The corresponding weight is calculated as follows: ， ， in, For visual features The corresponding weight, Odor characteristics The corresponding weight, is the attention mechanism function; According to visual features and odor characteristics The corresponding weights will be visual features and odor characteristics Fusion is performed to obtain fused data features , the calculation formula is as follows: ； The pneumatic conveying system (4) is arranged at the bottom of the rear end of the vehicle body (1) and is used to suck up road pollutants cleaned by the cleaning mechanism (2).

2. The adaptive sweeper with a pneumatic conveying system according to claim 1, characterized in that: The camera group comprises at least a front camera, a left camera, and a right camera; the odor sensor group is arranged at the bottom of the vehicle body (1).

3. The adaptive sweeper with a pneumatic conveying system according to claim 1, characterized in that: Acquiring road pollutant image data at different angles, fusing the road pollutant image data at different angles to obtain a first fused image, and identifying a first pollutant type and its pollution degree through the first fused image includes: Acquire a front image, a left image, and a right image at the same time point, and fuse the front image, the left image, and the right image at the same time point to obtain a first fused image; The first fused image is input into a pre-trained first pollutant detection model to obtain a first pollutant type and a pollution degree thereof. The pollutant types include: solid waste, particulate dust, and liquid pollutants. The pollution degrees include: high pollution, moderate pollution, and low pollution.

4. The adaptive sweeper with a pneumatic conveying system according to claim 3, characterized in that: Fusing the front image, the left image, and the right image at the same time point to obtain a first fused image includes: The deep features of the front image, left image, and right image are extracted respectively through the pre-trained convolutional neural network; The deep features of the front image, left image, and right image are weightedly fused using the attention mechanism, and the decoder is used to reconstruct the weighted fused features into the first fused image.

5. The adaptive sweeper with a pneumatic conveying system according to claim 1, characterized in that: Obtaining the second pollutant type and its pollution degree by fusing data features includes: inputting the fused data features into a pre-trained second pollutant detection model to obtain the second pollutant type and its pollution degree.

6. The adaptive sweeper with a pneumatic conveying system according to claim 1, characterized in that: Controlling the transverse driving member (3) to drive the cleaning mechanism (2) to move backward along the travel direction of the vehicle body (1) to increase the contact time between the cleaning mechanism (2) and road surface pollutants includes: The pollutant type, pollution degree, and speed adjustment of the cleaning mechanism (2) are used as state variables, and the time required for the cleaning mechanism (2) to clean the road pollutants is used as the action variable to construct a cleaning time reinforcement learning model; A reward function is constructed by cleaning the effect value. When the reward value of the reward function is higher than the reward threshold, it is marked as a positive reward signal, otherwise it is marked as a negative reward signal. The reinforcement learning model is optimized according to the reward signal. Inputting the current first pollutant type and the second pollutant type and the corresponding pollution degree into the cleaning time reinforcement learning model to obtain the cleaning time required; The driving speed of the lateral driving member (3) to drive the cleaning mechanism (2) backward along the driving direction of the vehicle body (1) is controlled, so that the contact time between the cleaning mechanism (2) and the road surface contaminants meets the time required for cleaning while the vehicle body (1) maintains the driving speed.

7. The adaptive sweeper with a pneumatic conveying system according to claim 6, characterized in that: When the vehicle body (1) maintains the driving speed, if the maximum contact time between the cleaning mechanism (2) and the road pollutants is less than the time required for cleaning, an alarm signal is issued.

8. The adaptive sweeper with a pneumatic conveying system according to claim 1, characterized in that: The cleaning mechanism (2) comprises a sweeping brush and a high-pressure nozzle.

Images