CN110497419A - Construction waste sorting robot - Google Patents

Abstract

The present invention relates to the technical field of mobile robots and path planning, in particular to a construction waste sorting robot, which includes a sorting robot body, and the sorting robot body includes the following modules: manipulator: used to pick up waste on construction sites ;Movement module: used to drive the body of the sorting robot to move; image acquisition device: used to drive the body of the sorting robot to obtain image data of the construction site in real time; collection module: used to store waste picked up by the manipulator; control module: Automatically identify waste and waste category data in the image data, and control the mobile module to drive the sorting robot body to move to the waste, control the manipulator to pick up waste, and store the waste to the collection module according to the category data preset position in . The invention has the effect of being able to sort wastes at the construction site.

Description

Construction waste sorting robot

technical field

The invention relates to the technical field of mobile robots and path planning, in particular to a construction waste sorting robot.

Background technique

Currently, construction waste refers to waste or hazardous materials generated during the construction or demolition of buildings. In today's daily life, the construction industry has caused many negative impacts on the natural environment, including construction waste.

When the existing building waste is generated during the construction or demolition of buildings, if the construction waste cannot be removed in time, in addition to the impact of the construction waste on the surrounding environment, the construction workers on the construction site will also step on it by mistake. onto the construction waste, causing unnecessary damage. Therefore, it is necessary to clean up the construction waste in time. However, due to the large amount of construction waste on the construction site, it will take a lot of time and energy to clean up the construction waste manually, which is not conducive to the progress of the project, and there may be cleaning Not quite, so there is room for improvement.

Contents of the invention

The object of the present invention is to provide a construction waste sorting robot capable of sorting the waste at the construction site.

Above-mentioned invention purpose of the present invention one is achieved by the following technical solutions:

A construction waste sorting robot, the construction waste sorting robot includes a sorting robot body, and the sorting robot body includes the following modules:

Manipulator: used to pick up waste on construction sites;

Mobile module: used to drive the movement of the sorting robot body;

Image acquisition device: used for the mobile module to drive the sorting robot body to acquire image data of the construction site in real time;

A collection module: used to store the waste picked up by the manipulator;

Control module: automatically identify the waste and the category data of the waste in the image data, and control the moving module to drive the sorting robot body to move to the waste, and control the manipulator picking up the waste, and storing the waste to a preset location in the collection module according to the category data.

By adopting the above technical solution, the control module on the sorting robot body can analyze and process the image data of the construction site acquired by the image acquisition device, so as to obtain the position of the waste in the image data of the construction site, thereby controlling the mobile module and Manipulator, move the sorting robot body to the waste, and use the manipulator to pick up the waste, and according to the result of image recognition, identify the category data of the picked up waste, and classify the waste according to the category data Storage, on the one hand, it can automatically pick up the waste on the construction site, which has the effect of environmental protection; at the same time, when the waste is identified from the image data, it can be classified and stored according to the type of waste, so that the stored waste can be It is more reasonable, helps the subsequent recycling and utilization of waste, and further contributes to environmental protection.

The present invention is further set as: the control module adopts the following method steps to identify the waste and the category data:

S10: If the image data is acquired, identifying the waste and the category data from the image data;

S20: According to the waste in the image data, use a full-coverage path planning algorithm to calculate, and obtain corresponding moving path data;

S30: Generate and send a moving message to the moving module according to the moving path data, so that the moving module drives the sorting robot to move according to the moving message;

S40: When the sorting robot body moves to the waste, send a picking message to the manipulator, so that the manipulator picks up the waste, and stores the waste into the collection module according to at the corresponding position.

By adopting the above technical solution, after the control module acquires the image data, it uses the preset model to identify the waste and the corresponding category data on the construction site in the image data, so that the sorting robot can Data, pick up and classify waste; at the same time, use full-coverage path planning algorithm to plan movement path data based on the location of waste, which can make the moving route of the sorting robot body more reasonable and reduce repeated routes , which improves the efficiency.

The present invention is further set to: Step S10 includes:

S11: Using a preset waste recognition model, process the image data to obtain a corresponding processing result;

S12: Obtain the image of the waste and the corresponding category data from the processing result.

By adopting the above technical solution, the waste identification model is pre-trained, and after the control module acquires the image data, it can recognize the position of the waste and the category data corresponding to the waste from the image data, so that the control module can be based on The position and category data control the motion of the mobile module and the manipulator.

The present invention is further set to: before step S10, the waste recognition model is obtained through training through the following method steps:

S101: According to the category data, obtain several corresponding pictures of waste to be tested category by category;

S102: Input the waste to-be-tested pictures into the CNN network one by one to obtain corresponding waste feature values;

S103: According to the characteristic value of the waste, use the RPN network to process each picture of the waste to be tested, and obtain the corresponding suggestion window;

S104: Map the suggestion window to the CNN network, and use the RoI pooling layer to process each suggestion window to obtain a fixed-size feature map corresponding to each suggestion window;

S105: According to the category data, perform training on the feature map to obtain a waste recognition model capable of identifying waste corresponding to the category data.

By adopting the above-mentioned technical scheme, by adopting the Faster R-CNN algorithm, extracting the feature value of each waste according to the waste category data, generating a suggestion window and processing it through the pooling layer, and then obtaining the building that can be trained and obtained. A waste identification model for each waste on site and the corresponding category data.

The present invention is further set to: step S20 includes:

S21: Perform rasterization processing on the image data to obtain raster image data;

S22: Obtain the position data of each waste in the raster image data, and call the A* algorithm to calculate according to the position data, so as to obtain corresponding movement path data.

By adopting the above technical solution, using rasterization processing and A* algorithm, after obtaining the position of the waste, it is possible to calculate the optimal route when the sorting robot body picks up the waste on the construction site, thereby reducing the number of sorting The repetitive movement of the robot body improves the efficiency of picking.

The present invention is further set to: step S40 includes:

S41: Acquire category numbers in the category data in advance, divide corresponding waste storage areas in the collection module according to the category numbers, and mark each of the waste storage areas according to the category data;

S42: When the manipulator picks up the waste, store the waste in the waste storage area corresponding to the collection module according to the category data corresponding to the waste.

By adopting the above technical solution, the storage area corresponding to each type of data is set in the collection module in advance, and when the waste on the construction site is sorted, the picked waste can be placed in the corresponding area of the collection module. Contribute to subsequent recycling.

Preferably, the control module also includes a control system composed of the following modules:

An image recognition module, configured to recognize the waste and the category data from the image data if the image data is acquired;

A movement path calculation module, configured to use a full-coverage path planning algorithm to perform calculations based on the waste in the image data to obtain corresponding movement path data;

A movement control module, configured to generate and send a movement message to the movement module according to the movement path data, so that the movement module drives the sorting robot to move according to the movement message;

a sorting and collecting module, configured to send a picking message to the manipulator when the sorting robot body moves to the waste, so that the manipulator picks up the waste, and stores the waste according to the to the corresponding position at the collection module.

By adopting the above technical solution, after the control module acquires the image data, it uses the preset model to identify the waste and the corresponding category data on the construction site in the image data, so that the sorting robot can Data, pick up and classify waste; at the same time, use full-coverage path planning algorithm to plan movement path data based on the location of waste, which can make the moving route of the sorting robot body more reasonable and reduce repeated routes , which improves the efficiency.

In summary, the beneficial technical effects of the present invention are:

1. The control module on the sorting robot body can analyze and process the image data of the construction site obtained by the image acquisition device, so as to obtain the position of the waste in the image data of the construction site, thereby controlling the mobile module and manipulator, and sorting The body of the picking robot moves to the waste, and uses the manipulator to pick up the waste, and according to the result of image recognition, recognizes the category data of the picked waste, and classifies and stores the waste according to the category data. It can automatically pick up the waste on the construction site, which has the effect of environmental protection;

2. When identifying waste from image data, classify and store waste according to the type of data, which can make the stored waste more reasonable, help the subsequent recycling and utilization of waste, and further contribute to environmental protection .

Description of drawings

Fig. 1 is a flowchart of a method for a control module to identify waste and category data in an embodiment of the present invention;

Fig. 2 is an implementation flow chart of step S10 in the method for the control module to identify waste and category data in an embodiment of the present invention;

Fig. 3 is another flow chart of the method for the control module to identify waste and category data in an embodiment of the present invention;

Fig. 4 is an implementation flow chart of step S20 in the method for the control module to identify waste and category data in an embodiment of the present invention;

Fig. 5 is an implementation flow chart of step S40 in the method for the control module to identify waste and category data in an embodiment of the present invention;

Fig. 6 is a functional block diagram of the control system in an embodiment of the present invention.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings.

Embodiment one:

The invention discloses a construction waste sorting robot, comprising: a sorting robot body, the sorting robot body includes a control module, and a manipulator coupled to the control module, a moving module, an image acquisition module and a collection module.

The manipulator is controlled by the control module, and can pick up the corresponding items according to the instructions sent by the control module. In this embodiment, the specific items picked up are wastes from the construction site.

The image acquisition device sends image data to the control module. In this embodiment, the image acquisition device may specifically be a camera. The image acquisition device acquires the image data of the construction site in real time, and sends the image data to the control module for processing.

The mobile module is controlled by the control module, and can move the sorting robot body according to the instructions sent by the control module, so that the sorting robot body can move to the side of the waste.

The collection module is used to store the waste picked up by the manipulator. The collection module can be fixed on the sorting robot body, or it can be set in one of the areas of the construction site. The sorting robot body can know that the area is in the construction site. Location.

Embodiment two:

In one embodiment, as shown in Figure 1, the control module uses the following method steps to identify waste and category data, and pick up and collect waste:

S10: If the image data is acquired, identify waste and category data from the image data.

In this embodiment, image data refers to image data of actual conditions in a construction site. Waste refers to the waste generated during the construction or demolition of buildings on construction sites, where the waste includes but not limited to screws, stones, boards and other garbage. Here, the category data refers to data corresponding to each type of waste.

Specifically, a model capable of identifying waste and the corresponding category data of the waste from the image is set in the control module in advance. The data captures the location of waste on construction sites and the type of waste identified for each type of waste.

S20: According to the waste in the image data, use a full-coverage path planning algorithm to calculate, and obtain corresponding moving path data.

In this embodiment, the full-coverage path planning algorithm refers to an algorithm that obtains a shortest path that traverses all places except obstacles within a certain area or spatial range. The moving path data refers to the movement and sorting route of the sorting robot body obtained by calculating through the full coverage path algorithm according to the location of the waste on the construction site.

Specifically, after the control module recognizes the waste and the position of the waste on the construction site from the image data, the full-coverage path algorithm is used to calculate the shortest route for the sorting robot to move and pick up the waste on the construction site, and The shortest route is used as the movement route data.

S30: Generate and send a moving message to the moving module according to the moving path data, so that the moving module drives the sorting robot to move according to the moving message.

In this embodiment, the moving message refers to a message for controlling the movement of the mobile module.

Specifically, the path specifically included in the movement path data and the current position of the sorting robot body generate the movement message, thereby driving the movement module to move, and driving the sorting robot to move according to the movement path data calculated in step S20.

S40: When the sorting robot body moves to the waste, send a picking message to the manipulator, so that the manipulator picks up the waste, and stores the waste to the corresponding position of the collection module.

In this embodiment, the picking message refers to a message that controls the manipulator to perform operations such as rotation, clamping, and picking.

Specifically, when the moving module drives the sorting robot body to move to the corresponding waste according to the moving message, the control module sends the picking message to the manipulator to pick up the waste and store it in the corresponding location of the collecting module.

In this embodiment, the control module on the sorting robot body can analyze and process the image data of the construction site acquired by the image acquisition device, so as to obtain the position of the waste in the image data of the construction site, thereby controlling the mobile module and Manipulator, move the sorting robot body to the waste, and use the manipulator to pick up the waste, and according to the result of image recognition, identify the category data of the picked up waste, and classify the waste according to the category data Storage, on the one hand, it can automatically pick up the waste on the construction site, which has the effect of environmental protection; at the same time, when the waste is identified from the image data, it can be classified and stored according to the type of waste, so that the stored waste can be It is more reasonable, helps the subsequent recycling and utilization of waste, and further contributes to environmental protection.

In one embodiment, as shown in FIG. 2 , in step S10, if the image data is acquired, the waste and category data are identified from the image data, which specifically includes the following steps:

S11: Process the image data using a preset waste recognition model to obtain a corresponding processing result.

In this embodiment, the waste recognition model refers to a pre-trained model capable of recognizing waste from image data of a construction site.

Specifically, after the waste recognition model is trained, the waste recognition model is stored and set in the control module, and after the control module obtains the image data of the construction site, it invokes the waste recognition model to process the image data processing to obtain the corresponding processing results.

S12: Obtain images of waste and corresponding category data from the processing results.

Specifically, the processing result includes whether the target is waste or the target is not waste. Further, from the processing results, the results whose target is waste are screened out, and the category data corresponding to the waste is obtained.

In one embodiment, as shown in FIG. 3, before step S10, the waste recognition model is obtained through training through the following method steps:

S101: According to the category data, obtain a plurality of corresponding waste to-be-tested pictures category by category.

In this embodiment, the pictures of waste to be tested refer to pictures of waste that need to be trained.

Specifically, according to the category data of the waste, a picture of each category of waste is obtained as the picture of the waste to be tested. Among them, in order to ensure the effectiveness of the trained waste recognition model in recognizing waste, there is no limit to the number of images to be obtained for each type of waste.

S102: Input the images of the waste to be tested into the CNN network one by one to obtain the corresponding characteristic value of the waste.

In this embodiment, the CNN network refers to a type of feed-forward neural network (Feed-forward Neural Networks) that includes convolution calculation and has a deep structure, and is one of the representative algorithms of deep learning. The convolutional neural network has the ability of representation learning, which can classify the input information invariant to translation according to its hierarchical structure. The waste feature value refers to a picture of the characteristics of the waste corresponding to each category of data.

Specifically, according to the category data, each waste to-be-tested picture is input into the CNN network one by one, and each waste to-be-tested picture is down-sampled, convoluted and pooled through the CNN network, and the corresponding drinking water is obtained. The corresponding waste characteristic value of the waste.

S103: According to the characteristic value of the waste, use the RPN network to process each picture of the waste to be tested to obtain a corresponding suggestion window.

In this embodiment, the RPN network, namely Region Proposal Network, refers to a network used to find a predefined number of regions that may contain objects in a feature map.

Specifically, according to the waste feature value corresponding to each waste test picture, use the RPN network to generate at least 300 suggestion windows in the waste feature value corresponding to each waste test picture.

S104: Map the suggestion window to the CNN network, and use the ROI pooling layer to process each suggestion window to obtain a fixed-size feature map corresponding to each suggestion window.

In this embodiment, the ROI pooling layer refers to an algorithm that generates a fixed-size feature map from proposal windows of different sizes.

Specifically, input the waste feature value into the ROI pooling layer, obtain the corresponding location area according to the suggestion window, divide several sections in the location area, and finally perform max pooling on each section, and finally get the corresponding location area for each suggestion window Feature maps of fixed size.

S105: According to the category data, the feature map is trained to obtain a waste recognition model capable of identifying waste corresponding to the category data.

Specifically, after the feature map of each category of waste in the category data is obtained according to the category data, the feature map is trained using the existing deep learning method, and finally the corresponding image that can identify the category data is obtained. Waste identification model for waste.

Among them, the waste identification model includes the following algorithm to identify waste:

Among them, A, B and D represent non-negative constants, i and j respectively represent the labels of neurons in deep learning, I _i represents the external input ω _ij represents the weight coefficient of neurons i and j, through the formula It is used to identify obstacles, so that when the control module controls the mobile module to move, it can avoid the identified obstacles. Wherein, the obstacle refers to the waste that is heavy, cannot be picked up by the manipulator, and will block the movement of the sorting robot. It is used to ensure that the sorting robot tends to the target, that is, waste. is a positive coefficient, and δ _it represents the angle between the vector from the sorting robot body to neuron i and the vector from the sorting robot body to neuron t.

In one embodiment, as shown in FIG. 4, in step S20, according to the waste in the image data, a full-coverage path planning algorithm is used for calculation to obtain corresponding moving path data, which specifically includes the following steps:

S21: Perform rasterization processing on the image data to obtain raster image data.

In this embodiment, raster image data refers to a data form in which space is divided into regular grids, each grid is called a unit, and corresponding attribute values are assigned to each unit to represent entities.

Specifically, rasterization processing is performed on the image data to obtain raster image data of the image data.

S22: Obtain the position data of each waste in the raster image data, and call the A* algorithm to calculate according to the position data, and obtain the corresponding movement path data.

In this embodiment, the A* algorithm refers to the most effective direct search method for finding the shortest path in a static road network, and is also an effective algorithm for solving many search problems. The closer the distance estimate in the algorithm is to the actual value, the faster the final search.

Specifically, in the raster image data, each waste is mapped to the raster image data, and the A* algorithm is used for calculation, so as to obtain the moving path data.

After using the method from step S101 to step S105 to identify the waste and obstacles, the waste is used as the target point for the movement of the sorting robot body, the control module controls the moving module to avoid obstacles, and calculates the moving position of the sorting robot The shortest route of the target point can be calculated by using the following A* algorithm function:

f(x)=g(x)+h(x)

Among them, g(x) represents the cost from the initial node to any node x in the grid image, and h(x) represents the cost from node x to the target node. In this embodiment, it is specifically moving from the sorting robot body to The distance of the identified litter. When the sorting robot body moves from the current position, that is, the initial position to the identified waste, the size of g(x) and h(x) are adjusted respectively by the A* algorithm, so that the number of nodes of f(x) is the smallest, that is Find out the minimum distance from the body of the sorting robot to the waste.

Preferably, a corresponding model for identifying obstacles can be added to the control module, and the criterion for judging the obstacle can be based on the inability of the mobile module to cross, that is, for example, the obstacle is a raised wooden board or stone, etc., and When the mobile module is moving, because the height of the wooden plank or stone is too high, the mobile module cannot go over the protrusion, so the protrusion is judged as an obstacle. Further, if it is determined to be a waste, the mobile module can be controlled to move to the waste for a few seconds until the waste is picked up; if it is determined to be an obstacle, the route in the moving path data can be to bypass the obstacle .

In one embodiment, as shown in FIG. 5, in step S40, when the sorting robot body moves to the waste, it sends a picking message to the manipulator, so that the manipulator picks up the waste, and stores the waste according to the collection method. The corresponding position of the module includes the following steps:

S41: Obtain the number of categories in the category data in advance, divide the corresponding waste storage areas in the collection module according to the number of categories, and mark each waste storage area according to the category data.

In this embodiment, the waste storage area refers to an area for storing picked up waste.

Specifically, according to the number of categories in the category data, that is, how many kinds of wastes there are in total, according to the number of categories, divide the corresponding number of areas in the collection module as waste storage areas, and store each waste according to the category data The area is marked.

S42: When the manipulator picks up the waste, store the waste in the waste storage area corresponding to the collection module according to the category data corresponding to the waste.

Specifically, when the manipulator picks up the waste, the waste is stored in the waste storage area corresponding to the collection module according to the category data corresponding to the waste.

It should be understood that the sequence numbers of the steps in the above embodiments do not mean the order of execution, and the execution order of each process should be determined by its functions and internal logic, and should not constitute any limitation to the implementation process of the embodiment of the present invention.

Embodiment three:

In one embodiment, a control system of a construction waste sorting robot is provided, which is the same as the method of identifying waste and category data by the control module in the second embodiment above, and picking and collecting waste correspond. As shown in FIG. 6 , the control system includes an image recognition module 10 , a movement path calculation module 20 , a movement control module 30 and a sorting and collection module 40 . The detailed description of each functional module is as follows:

An image recognition module 10, configured to identify waste and category data from the image data if the image data is acquired;

The moving path calculation module 20 is used to calculate using the full coverage path planning algorithm according to the waste in the image data to obtain corresponding moving path data;

The mobile control module 30 is used to generate and send a mobile message to the mobile module according to the mobile path data, so that the mobile module can drive the sorting robot to move according to the mobile message;

The sorting and collecting module 40 is used to send a picking message to the manipulator when the body of the sorting robot moves to the waste, so that the manipulator picks up the waste and stores the waste to the corresponding position of the collection module.

Preferably, the image recognition module 10 includes:

The image processing sub-module 11 uses a preset waste recognition model to process the image data to obtain corresponding processing results;

The image recognition sub-module 12 acquires images of wastes and corresponding category data from the processing results.

Preferably, the control system also includes the following modules:

The picture-to-be-test acquisition module 101 is used to acquire a plurality of corresponding waste to-be-test pictures by category according to the category data;

The feature extraction module 102 is used to input the pictures of waste to be tested into the CNN network one by one to obtain the corresponding waste feature value; the RPN network module 103 is used to use the RPN network to make each piece of waste according to the waste feature value. Process the image of the object to be tested to obtain the corresponding suggestion window;

The pooling processing module 104 is used to map the suggestion window into the CNN network, and use the RoI pooling layer to process each suggestion window to obtain a fixed-size feature map corresponding to each suggestion window;

The training module 105 is configured to train the feature map according to the category data to obtain a waste recognition model capable of identifying waste corresponding to the category data.

Preferably, the moving path calculation module 20 includes:

The raster processing sub-module 21 is used for rasterizing the image data to obtain raster image data;

The calculation sub-module 22 is used to obtain the position data of each waste in the raster image data, and call the A* algorithm to calculate according to the position data to obtain the corresponding moving path data.

Preferably, the sorting and collecting module 40 also includes:

The classification sub-module 41 is used to obtain the number of categories in the category data in advance, divide the corresponding waste storage area in the collection module according to the number of categories, and mark each waste storage area according to the category data;

The classified storage sub-module 42 is configured to store the waste in the waste storage area corresponding to the collection module according to the type data corresponding to the waste when the manipulator picks up the waste.

For the specific limitation of the control system, please refer to the limitation of the method for identifying waste and category data, and picking and collecting waste in the second embodiment above, and details will not be repeated here. Each module in the above control system can be fully or partially realized by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, and can also be stored in the memory of the computer device in the form of software, so that the processor can invoke and execute the corresponding operations of the above-mentioned modules.

Claims (10)

1. A construction waste sorting robot, characterized in that, the construction waste sorting robot includes a sorting robot body, and the sorting robot body includes the following modules: Manipulator: used to pick up waste on construction sites; Mobile module: used to drive the movement of the sorting robot body; Image acquisition device: used for the mobile module to drive the sorting robot body to acquire image data of the construction site in real time; A collection module: used to store the waste picked up by the manipulator; Control module: automatically identify the waste and the category data of the waste in the image data, and control the moving module to drive the sorting robot body to move to the waste, and control the manipulator picking up the waste, and storing the waste to a preset location in the collection module according to the category data. 2. The construction waste sorting robot according to claim 1, wherein the control module uses the following method steps to identify the waste and the category data: S10: If the image data is acquired, identifying the waste and the category data from the image data; S20: According to the waste in the image data, use a full-coverage path planning algorithm to calculate, and obtain corresponding moving path data; S30: Generate and send a moving message to the moving module according to the moving path data, so that the moving module drives the sorting robot to move according to the moving message; S40: When the sorting robot body moves to the waste, send a picking message to the manipulator, so that the manipulator picks up the waste, and stores the waste into the collection module according to at the corresponding position. 3. The construction waste sorting robot according to claim 2, wherein step S10 comprises: S11: Using a preset waste recognition model, process the image data to obtain a corresponding processing result; S12: Obtain the image of the waste and the corresponding category data from the processing result. 4. The construction waste sorting robot according to claim 2, characterized in that, before step S10, the waste identification model is obtained through training in the following method steps: S101: According to the category data, obtain several corresponding pictures of waste to be tested category by category; S102: Input the waste to-be-tested pictures into the CNN network one by one to obtain corresponding waste feature values; S103: According to the characteristic value of the waste, use the RPN network to process each picture of the waste to be tested, and obtain the corresponding suggestion window; S104: Map the suggestion window to the CNN network, and use a RoIpooling layer to process each suggestion window to obtain a fixed-size feature map corresponding to each suggestion window; S105: According to the category data, perform training on the feature map to obtain a waste recognition model capable of identifying waste corresponding to the category data. 5. The construction waste sorting robot according to claim 2, wherein step S20 comprises: S21: Perform rasterization processing on the image data to obtain raster image data; S22: Obtain the position data of each waste in the raster image data, and call the A* algorithm to calculate according to the position data, so as to obtain corresponding movement path data. 6. The construction waste sorting robot according to claim 2, wherein step S40 comprises: S41: Acquire category numbers in the category data in advance, divide corresponding waste storage areas in the collection module according to the category numbers, and mark each of the waste storage areas according to the category data; S42: When the manipulator picks up the waste, store the waste in the waste storage area corresponding to the collection module according to the category data corresponding to the waste. 7. The construction waste sorting robot according to claim 1, wherein the control module further comprises a control system composed of the following modules: An image recognition module, configured to recognize the waste and the category data from the image data if the image data is acquired; A movement path calculation module, configured to use a full-coverage path planning algorithm to perform calculations based on the waste in the image data to obtain corresponding movement path data; A movement control module, configured to generate and send a movement message to the movement module according to the movement path data, so that the movement module drives the sorting robot to move according to the movement message; a sorting and collecting module, configured to send a picking message to the manipulator when the sorting robot body moves to the waste, so that the manipulator picks up the waste, and stores the waste according to the to the corresponding position at the collection module. 8. The construction waste sorting robot according to claim 1, wherein the image recognition module comprises: The image processing sub-module uses a preset waste recognition model to process the image data to obtain corresponding processing results; The image recognition submodule acquires the image of the waste and the corresponding category data from the processing result. 9. The construction waste sorting robot according to claim 1, wherein the control system further comprises the following modules: The picture-to-be-test acquisition module is used to acquire several corresponding waste to-be-test pictures by category according to the category data; The feature extraction module is used to input the waste to-be-tested pictures into the CNN network one by one to obtain the corresponding waste feature value; The RPN network module is configured to use the RPN network to process each picture of the waste to be tested according to the characteristic value of the waste to obtain a corresponding suggestion window; A pooling processing module, configured to map the suggestion window into the CNN network, and use the RoIpooling layer to process each of the suggestion windows to obtain a fixed-size feature map corresponding to each of the suggestion windows; The training module is configured to train the feature map according to the category data to obtain a waste recognition model capable of identifying waste corresponding to the category data. 10. The construction waste sorting robot according to claim 1, wherein the moving path calculation module comprises: The raster processing submodule is used to perform rasterization processing on the image data to obtain raster image data; the calculation submodule is used to obtain the position data of each waste in the raster image data, and According to the location data, the A* algorithm will be invoked for calculation, and the corresponding movement path data will be obtained.