Building robot material searching method based on RFID tag

Abstract

The invention discloses a building robot material searching method based on an RFID tag, and the method comprises the following steps: materials are packaged into stacks, enter a site, and are stacked in the site; after the building robot enters the site, a material stack is found and moved to the position beside the material stack, and the material stack serves as a starting point; a material stack required by the first construction area is searched by reading the RFID tag of each material stack; the found material stack is moved to the first construction area for construction; after construction of the first construction area is completed, the RFID tag read before is retrieved, and whether material stack information needed by the next construction area exists or not is confirmed; if so, directly moving to the side of the material stack and constructing; if not, the needed material stacks are searched for all the observed but unread material stacks; and materials required by the subsequent construction area are continuously searched, and construction is completed in the corresponding construction area until construction is completed. According to the invention, the material searching efficiency is greatly improved.

Description

Technical Field

[0001] The invention relates to the technical field of construction robots, and in particular to a material searching method of a construction robot based on RFID tags. Background Art

[0002] During construction, a lot of materials are usually stored in a relatively open area. Some materials have set use areas, so when constructing these areas, it is necessary to find the required materials from a large number of materials. Since materials are often moved during actual construction, even if all materials are counted and located in advance, the information obtained may be outdated and unreliable, so it is often necessary to temporarily find the required materials in the site. The traditional method is to manually inspect each material pile to determine whether it is the material being sought. Recently, the construction industry has promoted the digital traceability technology of building materials. RFID tags are affixed to the materials when they leave the factory. Manual handheld RFID readers can be used to read the information of each material pile and then determine whether it is the required material. However, due to cost reasons, the reading distance of the RFID tags used is usually within a few meters, and personnel need to be close to the material pile to stably read the information in the tag. Therefore, the existing technology requires a lot of manual operation for searching materials, which is time-consuming and laborious. When the site is large, the materials are scattered, and the line of sight is easily blocked, it is often easy to miss some materials, resulting in the need to search for materials again. Moreover, manual planning of the material search path is often not scientific enough, and there is room for compressing the material search time.

[0003] Therefore, it is urgent to solve the above problems. Summary of the invention

[0004] Purpose of the invention: The purpose of the present invention is to provide a material-finding method for a construction robot based on RFID tags. Based on RFID tags, the construction robot can quickly find the required material piles among the material piles scattered on the site, thereby improving the efficiency of the robot and reducing the intensity of manual labor.

[0005] Technical solution: To achieve the above purpose, the present invention discloses a material searching method for a construction robot based on RFID tags, comprising the following steps:

[0006] (1) Materials are packed into stacks and stacked on site;

[0007] (2) The construction robot enters the site, finds a material stack, and moves to the side of the material stack, taking the material stack as the starting point;

[0008] (3) The construction robot searches for the material stacks required for the first construction area by reading the RFID tags of each material stack. The RFID tag information read during the search process is stored inside the construction robot. If the construction robot performs an overscan map modeling of the global environment of the construction site, the material stacks in the global environment are grouped according to their locations, and the contour recognition algorithm is used to identify the number of material stacks in each material stack group. Then, a route is planned to visit the material stacks in each group one by one, and the RFID reader is used to search for the required material stacks in each group of material stacks. If the construction robot does not perform an overscan map modeling of the global environment of the construction site, the material stacks that can be observed at the starting point are grouped according to their locations, and the contour recognition algorithm is used to identify the number of material stacks in each material stack group. Then, a route is planned to visit the material stacks in each group one by one. After finding the required material stack, the construction robot interrupts the subsequent planned route and no longer visits the subsequent material stack groups.

[0009] (4) The construction robot moves the found material pile to the first construction area and carries out construction;

[0010] (5) After completing the construction of the first construction area, the construction robot retrieves the RFID tag read previously to confirm whether there is the material stack information required for the next construction area; if so, it can directly move to the material stack and repeat step (4); if not, it faces all the material stacks that have been observed but not read, repeats steps (3) and (4), and finds and uses the required material stacks;

[0011] (6) Repeat step (5) to find the materials required for the subsequent construction area and complete the construction in the corresponding construction area until all areas are completed.

[0012] Optionally, in step (1), the materials are packed into stacks, placed on pallets, and brought into the construction site together with the pallets; when the materials are packed, an ultra-high frequency (UHF) or active RFID tag is attached to ensure that the reading distance of the RFID tag reaches 4 m or more, and the RFID tag indicates the quantity and size of the materials in the material stack and the corresponding construction area information in the construction site.

[0013] Optionally, in step (2), if the construction robot stores a three-dimensional map of the site, after entering the site, the construction robot moves to the nearest material stack according to the three-dimensional map and uses the material stack as the starting point.

[0014] Optionally, in step (2), if the construction robot does not store a three-dimensional map of the site, the construction robot is at the entrance. If the construction robot can see all or part of the material stack in its field of view, it will move to the nearest material stack and use the material stack as the starting point; if the construction robot cannot see the material stack in its field of view, the construction robot will scan and model the entire construction site according to a preset inner spiral full traversal path planning algorithm, and read the RFID tags of the observed material stacks during the scanning and modeling process; then, after the scanning and modeling is completed, the construction robot will move to the nearest material stack and use the material stack as the starting point.

[0015] Optionally, in step (2), if the construction robot stores a three-dimensional map of the site, but when the material stacks in the site have moved, the expired three-dimensional map information should not be used, but the operation should be performed as if there is no three-dimensional map of the site. That is, when the construction robot is at the entrance, if the construction robot can see all or part of the material stacks in its field of view, it will move to the nearest material stack and use the material stack as the starting point; if the construction robot cannot see the material stack in its field of view, the construction robot will scan and model the entire construction site according to the preset inner spiral full traversal path planning algorithm, and read the RFID tags of the observed material stacks during the scanning and modeling process; then, after the scanning and modeling is completed, it will move to the nearest material stack and use the material stack as the starting point.

[0016] Optionally, the specific method for grouping in step (3) is: setting the reading distance of the RFID tag on the material stack to the radius of a circle, and then covering all the material stacks with as few circles of the radius as possible, and the position corresponding to each circle is a group of material stacks; according to this method, it is possible to read the information of each stack in the corresponding group at the center of each circle; when some material stacks belong to two or more groups of material stacks at the same time, such material stacks are preferentially assigned to the group with more stacks or closer to the current starting point.

[0017] Optionally, the specific method for planning the route in step (3) is as follows: when the construction robot faces all the material pile groups that have been observed and not visited, it needs to plan a path to reach these material pile groups, so that these material pile groups are reached only once and the required material piles can be found as quickly as possible; in the planned route, there is no need to plan to return to the current position; the planning goal is no longer the shortest total path, but the shortest expected time to find the required material pile in the currently observed and not visited material pile groups.

[0018] Optionally, the specific method for planning the route in step (3) is as follows: when the construction robot moves to the next group of material stacks, it may observe some material stacks that have not been observed before; then when the construction robot reaches the next group of material stacks, it re-plans the path for all material stacks that have been observed but not visited, and then moves according to the new optimal planned path to find the required material stacks.

[0019] Optionally, the specific method for planning the route in step (3) is as follows: when the construction robot has visited all the observed material stacks but cannot find the required material stack, it will scan and model the entire construction site according to the preset inner spiral full traversal path planning algorithm, and the trajectory gap of the scanning and modeling will be set within twice the reading distance of the RFID tag on the material stack, so that the construction robot can read the RFID tags of all material stacks in the construction site during the scanning and modeling process; if the required material stack still cannot be found, the construction robot will issue an alarm.

[0020] Optionally, the specific mathematical model for planning the route in step (3) is:

[0021] There are a total of M material stacks in the global scene of the construction site. The current construction robot has visited P material stacks and has not found the required material stacks. The P material stacks contain Q material stacks, Q ≥ P; the current construction robot is next to the P-th material stack, the P-th material stack is taken as the starting point, and the sequence number is set to 0; the J-th material stack that has not been visited has been observed, and the J-th material stack that has not been visited contains K material stacks, K ≥ J, K + Q ≤ M; the sequence numbers of the J-th material stack that has not been visited are 1, 2, ..., J, and the number of material stacks in each material stack is I i (i=1,2,…,J), The distance between the i-th group and the s-th group of materials is D is , i,s ​​is the serial number of the material chopping group, and the moving speed of the construction robot is a constant value V;

[0022] Plan the path for the material chops of group J that have not been visited. Assume that the path passes through groups 1, 2, ..., J of material chops in sequence, and the time spent on the corresponding path is W. 12…J , we can use enumeration or dynamic programming to plan the path of J groups of materials, with a maximum of For each planned path, the expected time to find the required material stack can be calculated according to the following formula:

[0023] W 12…J =P 1 *T 1 +P 2 *T 2 +…+P J *T J+P other *T other

[0024] Where P t (t=1,2,3,…,J) is the probability of finding the required material stack in the tth group of material stacks, T t is the time when the construction robot arrives at the tth group of material stacks, P other is the probability that the required material stack is not in the currently planned J-group material stack, nor in the previously visited P-group material stack, but in the remaining MQK stack material stack, T other After visiting the current J-group material stack, the time to continue to find the required material in the remaining MQK stacks of materials; because in the calculation of the expected time of all different paths currently planned for the J-group material stack, P other and T other is a constant value, so when comparing different path plans, we only need to compare P 1 *T 1 +P 2 *T 2 +…+P J *T J That's it;

[0025] Assume that, in addition to the previously visited Q stacks, the remaining unvisited MQ stacks have equal probabilities of being the required materials, both but

[0026]

[0027] T t =(D 01 +D 12 +…+D t-1,t ) / V

[0028] For other paths, W needs to be adjusted accordingly during calculation. 12…J and D 01 +D 12 +…+D t-1,t , use the parameters that match the actual order, and thus find the W value under different path planning. The path corresponding to the minimum T value is the optimal path under the current conditions.

[0029] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: the present invention utilizes RFID tags to mark materials used in different construction areas, so that the robot can identify the materials through an RFID reader; the present invention uses an improved path planning method to enable the construction robot to plan the moving path according to the observed and identified material stacks, and then find the required materials in the scattered material stacks as quickly as possible; the improved path planning method of the present invention is closer to the actual scenario of material searching, can shorten the material searching time, and the material searching process is more intelligent and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0032] It should be understood that the present invention can be implemented in different forms and should not be construed as being limited to the embodiments set forth herein. On the contrary, providing these embodiments will make the disclosure thorough and complete and fully convey the scope of the present invention to those skilled in the art. In the accompanying drawings, the sizes and relative sizes of the components may be exaggerated for clarity. The same reference numerals throughout represent the same components.

[0033] like Figure 1 As shown, a material searching method for a construction robot based on RFID tags in the present invention comprises the following steps:

[0034] (1) Materials are packed and brought in and stacked on site;

[0035] Materials need to be packed into stacks, placed on pallets, and brought into the site together with the pallets. When packing materials, an ultra-high frequency (UHF) or active RFID tag must be attached to ensure that the reading distance of the RFID tag reaches 4m or more. The RFID tag should indicate the quantity and size of the materials in the stack and the corresponding construction area information in the construction site.

[0036] (2) The construction robot enters the site. When entering the site, the construction robot uses the laser radar or visual camera installed on the body to find a material stack and moves to the side of the material stack, taking the material stack as the starting point;

[0037] If the construction robot stores a three-dimensional map of the site, after entering the site, the construction robot moves to the nearest material pile according to the three-dimensional map and uses it as the starting point;

[0038] If the construction robot does not store a three-dimensional map of the site, at the entrance, if the construction robot can see all or part of the material stack in its field of vision, it will move to the nearest material stack as the starting point; if the material stack is not visible in its field of vision, the construction robot will scan and model the entire construction site according to the preset inner spiral full traversal path planning algorithm, and read the RFID tags of the observed material stacks during the scanning and modeling process; then, after the scanning and modeling is completed, the construction robot will move to the nearest material stack as the starting point;

[0039] The laser sensor used for modeling has a large range, and the intervals between the construction robot's scanning tracks will also be relatively large. Therefore, during the scanning and modeling process, the construction robot may not pass by all the material stacks at close range, so it may not necessarily read the RFID information of all the material stacks during this process. During the entire material search process, if the construction robot obtains the information of some material stacks in advance, when it needs to find the material stack later, the construction robot can move directly to the material stack without searching;

[0040] If the construction robot stores a three-dimensional map of the site, but the materials in the site have been moved, it should not use the expired map information, but should operate in a mode without storing the three-dimensional map of the site;

[0041] (3) The construction robot searches for the material stacks required for the first construction area by reading the RFID information of each material stack. The RFID information read during the search process is stored inside the construction robot.

[0042] If the construction robot performs overscanning modeling on the global environment of the construction site, it will group the material stacks in the global environment according to their locations, identify how many material stacks are in each material stack group through the contour recognition algorithm, and then plan a route to visit each group of material stacks one by one, and use the RFID reader to find the required material stacks in each group of material stacks;

[0043] If the construction robot has not scanned and modeled the global environment of the construction site, it will group the material stacks that can be observed at the starting point according to their locations, identify how many material stacks are in each material stack group through the contour recognition algorithm, and then plan a route to visit each group of material stacks one by one;

[0044] After finding the required material stack, the construction robot interrupts the subsequent planned route and no longer visits the subsequent material stack groups;

[0045] The specific method of grouping is:

[0046] The reading distance of the RFID tag on the material stack is set to the radius of the circle, and then all the material stacks are covered with as few circles of the radius as possible. The position corresponding to each circle is a group of material stacks. In this way, the information of each stack in the corresponding group can be read at the center of each circle. According to this method, some material stacks may belong to two or more groups of material stacks at the same time. Such material stacks can be preferentially assigned to the group with more stacks or closer to the current starting point.

[0047] The specific method of planning a route is:

[0048] When faced with all the material stacks that have been observed and not visited, the construction robot needs to plan a path to reach these material stacks, so that these material stacks are reached only once and the required material stacks can be found as quickly as possible; when planning the route, there is no need to plan to return to the current position; the planning goal is no longer the shortest total path, but the shortest expected time to find the required material stack in the currently observed and unvisited material stacks;

[0049] There are a total of M material stacks in the global scene of the construction site. The current construction robot has visited P material stacks and has not found the required material stacks. The P material stacks contain Q material stacks, Q ≥ P; the current construction robot is next to the P-th material stack, the P-th material stack is taken as the starting point, and the sequence number is set to 0; the J-th material stack that has not been visited has been observed, and the J-th material stack that has not been visited contains K material stacks, K ≥ J, K + Q ≤ M; the sequence numbers of the J-th material stack that has not been visited are 1, 2, ..., J, and the number of material stacks in each material stack is I i (i=1,2,…,J), The distance between the i-th group and the s-th group of materials is D is , i,s ​​is the serial number of the material chopping group, and the moving speed of the construction robot is a constant value V;

[0050] Plan the path for the material chops of group J that have not been visited. Assume that the path passes through groups 1, 2, ..., J of material chops in sequence, and the time spent on the corresponding path is W. 12…J , we can use enumeration or dynamic programming to plan the path of J groups of materials, with a maximum of For each planned path, the expected time to find the required material stack can be calculated according to the following formula:

[0051] W 12…J =P 1 *T 1 +P 2 *T 2 +…+P J *T J +P other *T other

[0052] Where P t (t=1,2,3,…,J) is the probability of finding the required material stack in the tth group of material stacks, T t is the time when the construction robot arrives at the tth group of material stacks, P other is the probability that the required material stack is not in the currently planned J-group material stack, nor in the previously visited P-group material stack, but in the remaining MQK stack material stack, T other After visiting the current J-group material stack, the time to continue to find the required material in the remaining MQK stacks of materials; because in the calculation of the expected time of all different paths currently planned for the J-group material stack, P other and T other is a constant value, so when comparing different path plans, we only need to compare P 1 *T 1 +P 2 *T 2 +…+P J *T J That's it;

[0053] It can be assumed that, in addition to the previously visited Q stacks, the remaining unvisited MQ stacks have equal probabilities of being the required materials, both but

[0054]

[0055] T t =(D 01 +D 12 +…+D t-1,t ) / V

[0056] The above formula is calculated for the path that passes through groups 1, 2, ..., J of materials in sequence. For other paths, W needs to be adjusted accordingly during calculation. 12…J and D 01 +D 12 +…+D t-1,t The subscripts of the command are used, and the parameters that match the actual order are used. Thus, the W value under different path planning can be obtained, and the path corresponding to the minimum T value is the optimal path under the current conditions.

[0057] When the construction robot moves to the next group of material stacks, it may observe some material stacks that have not been observed before. When the construction robot reaches the next group of material stacks, it re-plans the path for all the material stacks that have been observed but not visited, and then moves according to the new optimal planned path to find the required material stacks.

[0058] When the construction robot has visited all the observed material stacks but cannot find the required material stack, it will scan and model the entire construction site according to the preset inner spiral full traversal path planning algorithm, and the trajectory gap of the scanning and modeling will be set within twice the reading distance of the RFID tag on the material stack. In this way, the construction robot can read the RFID tags of all material stacks in the construction site during the scanning and modeling process; if the required material stack is still not found, the construction robot will sound an alarm;

[0059] (4) The construction robot moves the found material pile to the first construction area and carries out construction;

[0060] (5) After completing the construction of the first construction area, the construction robot retrieves the RFID tag read previously to confirm whether there is the material stack information required for the next construction area; if so, it can directly move to the material stack and repeat step (4); if not, it faces all the material stacks that have been observed but not read, repeats steps (3) and (4), and finds and uses the required material stacks;

[0061] (6) Repeat step (5) to find the materials required for the subsequent construction area and complete the construction in the corresponding construction area until all planned areas are completed.

[0062] Example 1

[0063] The construction robot in the present invention can adopt the ceiling panel installation robot disclosed in CN118793252A. This embodiment is a factory where ceiling panels are being installed. The ceiling panel installation robot is used to search, fork, move, and lift materials to assist workers in construction. The factory area is 40m*80m, and the main obstacles in the space are 6 pillars. A total of 32 stacks of ceiling panels are required in the construction site of the factory. Currently, 16 stacks have been installed, and the corresponding half of the area of ​​ceiling panels has been installed.

[0064] The remaining 16 stacks of ceiling panels have all been delivered and are scattered on the factory floor. The ceiling panels are packed into stacks, placed on pallets, and delivered together with the pallets. Each stack of ceiling panels is equipped with an ultra-high frequency RFID tag with a stable reading distance of 4m. When reading the RFID tag, you can obtain the list of materials in the stack of ceiling panels, the dimensions of each ceiling panel, and the overall number of the stack of ceiling panels, which indicates which area of ​​the factory the stack of ceiling panels should be used for. RFID tags usually also record other factory information of the materials, such as the manufacturer, production date, material, weight, etc.; on the one hand, it is convenient for the acceptance, management, and subsequent maintenance of materials, and on the other hand, it is also convenient for on-site construction to allocate or request replenishment.

[0065] When the ceiling panels were brought in, most of them were directly transported to the areas where they were needed, but a small number of them may have been transported to places that did not correspond to the areas of use. The storage location of the ceiling panels in the factory was not reflected in the BIM model of the factory. Because these ceiling panels had not yet been brought in when the construction robot was scanning and modeling in the factory, these ceiling panels were usually not included in the robot's 3D map. Therefore, when the ceiling panel installation robot came in, it did not use the 3D map converted from the BIM file, nor did it use the map previously scanned and modeled, but used a 3D camera as a navigation and recognition tool.

[0066] When the ceiling panel installation robot enters the factory building, the 3D camera of the ceiling panel installation robot is started to identify the surrounding environment. By calling the YOLO V5 visual algorithm, the ceiling panel installation robot can perceive objects such as ceiling panel stacks, obstacles in the scene, and identify the outline of the ceiling panel stack and the pallet under the ceiling panel stack in the scene through open source data sets and algorithms. However, due to reasons such as light and obstacles, the ceiling panel installation robot can only recognize scenes within a radius of approximately 10 meters. There are 5 stacks of ceiling panels within this range, of which 3 stacks are stacked separately, each within 5-10 meters of other ceiling panels, and 2 stacks are stacked together, about 3 meters away from each other. According to the specific method for grouping in the present invention, these 2 stacks are classified as 1 group, and the other 3 stacks each become 1 group.

[0067] The set of ceiling panels closest to the ceiling panel installation robot contains only one stack of ceiling panels. The ceiling panel installation robot moves to the vicinity of this stack of ceiling panels and reads the serial number of the stack of ceiling panels through the RFID reader carried by the ceiling panel installation robot, which is 22. The robot currently plans to construct area 17, which does not match the serial number of the ceiling panels.

[0068] The ceiling panel installation robot needs to find the required ceiling panel with serial number 17 among the remaining 3 groups (4 stacks) of ceiling panels that have been observed but not visited. The ceiling panel installation robot needs to plan a path and visit these 3 groups of ceiling panels in turn.

[0069] The serial number settings of these 3 groups (4 stacks) of ceiling panels are shown in Table 1:

[0070] Table 1

[0071]

[0072] According to the measurement of the 3D camera of the ceiling panel installation robot, the distances between the four sets of ceiling panels are shown in Table 2:

[0073] Table 2

[0074] Group number\distance\group number 0 1 2 3 0 0 5 7.5 10 1 5 0 8 6 2 7.5 8 0 9 3 10 6 9 0

[0075] The ceiling board with serial number 17 is among the remaining 15 stacks of ceiling boards in the site, but it is not necessarily among the 3 groups (4 stacks) of ceiling boards that can be observed currently.

[0076] The ceiling panel installation robot plans a total of 3 groups (4 stacks) of ceiling panels that have been observed but not visited. There are three paths, namely 123 (i.e. first move to the group with sequence number 1, and then move to the groups with sequence numbers 2 and 3 in sequence), 132, 213, 231, 312, and 321. Then, the ceiling panel installation robot calculates the expected time to find the required ceiling panel for each path.

[0077] For the path with the order of 123, the mathematical model for planning the route in the present invention is used for calculation, without considering P other *T other ,

[0078] W 123 =P 1 *T 1 +P 2 *T 2 +P 3 *T 3

[0079]

[0080] Calculation can be obtained Where V is the moving speed of the robot.

[0081] The expected time to find the required material stack can be calculated in the same way for the other five paths. Then the path with the shortest expected time is selected as the current optimal planning path. The calculation time of the ceiling panel installation robot is within 1ms. When the number of groups and stacks is less than 50, the calculation time generally does not exceed 20ms.

[0082] The ceiling panel installation robot visits these ceiling panel stacks according to the optimal planning path to find ceiling panel stack No. 17. If other ceiling panel stacks that have not been visited are observed during the movement, they will be included in the path planning range and the path planning will be re-performed. If ceiling panel stack No. 17 is still not found after visiting all ceiling panel stacks observed during the movement, the entire factory will be fully moved to find ceiling panel stacks that have not been observed, and ceiling panel stack No. 17 will be found from them.

[0083] After the ceiling panel installation robot finds the No. 17 ceiling panel stack, it uses the YOLO V5 visual algorithm to identify the pallet under the material stack, determines the relative position between the robot and the pallet, and then adjusts its chassis to be close to and parallel to the edge of the pallet. It then extends the fork on the chassis into the pallet, lifts the pallet and the material, and then moves the fork into the ceiling panel installation robot to move the pallet and the material to the upper part of the ceiling panel installation robot chassis. The ceiling panel installation robot carries the No. 17 pallet and ceiling panels and moves to the planned construction area, namely Area 17.

[0084] After installing the No. 17 ceiling panel stack, the ceiling panel installation robot repeats the above steps to find the No. 18 material, fork, transport, lift and inspect it. Then it repeats the above steps for the No. 19, 20, ... 32 materials to complete the construction.

[0085] During the movement of the ceiling panel installation robot, including the aforementioned movement along the planned path to find the required ceiling panel stack, the ceiling panel installation robot uses a 3D camera to model the environment along the way. At the same time, the ceiling panel installation robot records the number of the material stack that has read the RFID tag, so as to facilitate the subsequent rapid positioning of the material stack with a specific number, without the need to re-plan the path and re-search for the material stack.

Claims

1. A construction robot material searching method based on RFID tags, characterized in that: The steps include: (1) Materials are packed into stacks and stacked on site; (2) The construction robot enters the site, finds a material stack, and moves to the side of the material stack, taking the material stack as the starting point; (3) The construction robot searches for the material stacks required for the first construction area by reading the RFID tags of each material stack. The RFID tag information read during the search process is stored inside the construction robot. If the construction robot performs an overscan map modeling of the global environment of the construction site, the material stacks in the global environment are grouped according to their locations, and the contour recognition algorithm is used to identify the number of material stacks in each material stack group. Then, a route is planned to visit the material stacks in each group one by one, and the RFID reader is used to search for the required material stacks in each group of material stacks. If the construction robot does not perform an overscan map modeling of the global environment of the construction site, the material stacks that can be observed at the starting point are grouped according to their locations, and the contour recognition algorithm is used to identify the number of material stacks in each material stack group. Then, a route is planned to visit the material stacks in each group one by one. After finding the required material stack, the construction robot interrupts the subsequent planned route and no longer visits the subsequent material stack groups. (4) The construction robot moves the found material pile to the first construction area and carries out construction; (5) After completing the construction of the first construction area, the construction robot retrieves the RFID tag read previously to confirm whether there is the material stack information required for the next construction area; if so, it can directly move to the material stack and repeat step (4); if not, it faces all the material stacks that have been observed but not read, repeats steps (3) and (4), and finds and uses the required material stacks; (6) Repeat step (5) to find the materials required for the subsequent construction area and complete the construction in the corresponding construction area until all areas are completed.

2. The material searching method of a construction robot based on RFID tags according to claim 1, characterized in that: In the step (1), the materials are packed into stacks, placed on pallets, and brought into the construction site together with the pallets. When the materials are packed, an ultra-high frequency (UHF) or active RFID tag is attached to ensure that the reading distance of the RFID tag reaches 4 m or more. The RFID tag indicates the quantity and size of the materials in the material stack and the corresponding construction area information in the construction site.

3. The material searching method of a construction robot based on RFID tags according to claim 1, characterized in that: In step (2), if the construction robot stores a three-dimensional map of the site, after entering the site, the construction robot moves to the nearest material stack according to the three-dimensional map and uses the material stack as the starting point.

4. The material searching method for a construction robot based on RFID tags according to claim 1, characterized in that: In step (2), if the construction robot does not store a three-dimensional map of the site, the construction robot is at the entrance. If the construction robot can see all or part of the material stack in its field of vision, it will move to the nearest material stack and use the material stack as the starting point. If the construction robot cannot see the material stack in its field of vision, it will scan and model the entire construction site according to the preset inner spiral full traversal path planning algorithm, and read the RFID tags of the observed material stacks during the scanning and modeling process; Then, after scanning and modeling, it moves to the material stack closest to itself and uses the material stack as the starting point.

5. The material searching method for a construction robot based on RFID tags according to claim 1, characterized in that: In step (2), if the construction robot has stored a three-dimensional map of the site, but when the material stack in the site has been moved, the expired three-dimensional map information should not be used, but the operation should be performed in accordance with the situation where the three-dimensional map of the site is not stored, that is, when the construction robot is at the entrance, if the construction robot can see all or part of the material stack in its field of view, it will move to the nearest material stack and use the material stack as the starting point; If the construction robot cannot see the material stack in its field of vision, it will scan and model the entire construction site according to the preset inner spiral full traversal path planning algorithm, and read the RFID tags of the observed material stacks during the scanning and modeling process; Then, after scanning and modeling, it moves to the material stack closest to itself and uses the material stack as the starting point.

6. The material searching method for a construction robot based on RFID tags according to claim 1, characterized in that: The specific method for grouping in step (3) is as follows: the reading distance of the RFID tag on the material stack is set to the radius of a circle, and then all the material stacks are covered with as few circles of the radius as possible, and the position corresponding to each circle is a group of material stacks; according to this method, the information of each stack in the corresponding group can be read at the center of each circle; when some material stacks belong to two or more groups of material stacks at the same time, such material stacks are preferentially assigned to the group with more stacks or closer to the current starting point.

7. The material searching method for a construction robot based on RFID tags according to claim 1, characterized in that: The specific method of planning the route in step (3) is as follows: the construction robot faces all the material stacks that have been observed and not visited, and needs to plan a path to reach these material stacks, so that these material stacks are reached only once and the required material stacks can be found as quickly as possible; When planning a route, there is no need to plan to return to the current location; The planning goal is no longer to minimize the total path, but to minimize the expected time to find the required material stack in the currently observed and unvisited material stack group.

8. The material searching method for a construction robot based on RFID tags according to claim 7, characterized in that: The specific method for planning the route in step (3) is as follows: when the construction robot moves to the next group of material stacks, it may observe some material stacks that have not been observed before; then when the construction robot reaches the next group of material stacks, it re-plans the path for all material stacks that have been observed but not visited, and then moves according to the new optimal planned path to find the required material stacks.

9. The method for searching materials by a construction robot based on RFID tags according to claim 8, characterized in that: The specific method of planning the route in step (3) is as follows: when the construction robot has visited all the observed material stacks but cannot find the required material stack, it will scan and model the entire construction site according to the preset inner spiral full traversal path planning algorithm, and the trajectory gap of the scanning and modeling will be set within twice the reading distance of the RFID tag on the material stack, so that the construction robot can read the RFID tags of all material stacks in the construction site during the scanning and modeling process; if the required material stack still cannot be found, the construction robot will issue an alarm.

10. The material searching method for a construction robot based on RFID tags according to claim 9, characterized in that: The specific mathematical model for planning the route in step (3) is: There are a total of M material stacks in the global scene of the construction site. The current construction robot has visited P material stacks and has not found the required material stacks. The P material stacks contain a total of Q material stacks, Q ≥ P; The current construction robot is next to the P-th material chopping group. The P-th material chopping group is taken as the starting point and the sequence number is set to 0; J groups of unvisited material chop groups have been observed, and the J groups of unvisited material chop groups contain K stacks of material stacks, K ≥ J, K + Q ≤ M; let the serial numbers of the J groups of unvisited material chop groups be 1, 2, ..., J, and the number of material stacks in each group of material chop groups is I i (i=1,2,…,J), The distance between the i-th group and the s-th group of materials is D is , i,s ​​is the serial number of the material chopping group, and the moving speed of the construction robot is a constant value V; Plan the path for the material chops of group J that have not been visited. Assume that the path passes through groups 1, 2, ..., J of material chops in sequence, and the time spent on the corresponding path is W 12…J , we can use enumeration or dynamic programming to plan the path of J groups of materials, with a maximum of For each planned path, the expected time to find the required material stack can be calculated according to the following formula: W 12…J =P1*T1+P2*T2+…+P J *T J +P other *T other Where P t (t=1,2,3,…,J) is the probability of finding the required material stack in the tth group of material stacks, T t is the time when the construction robot arrives at the tth group of material stacks, P other is the probability that the required material stack is not in the currently planned J-group material stack, nor in the previously visited P-group material stack, but in the remaining MQK stack material stack, T other After visiting the current J-group material stack, the time to continue to find the required material in the remaining MQK stacks of materials; because in the calculation of the expected time of all different paths currently planned for the J-group material stack, P other and T other is a constant value, so when comparing different path plans, we only need to compare P1*T1+P2*T2+…+P J *T J That's it; Assume that, in addition to the previously visited Q stacks, the remaining unvisited MQ stacks have equal probabilities of being the required materials, both but T t =(D 01 +D 12 +…+D t-1,t ) / V For other paths, W needs to be adjusted accordingly during calculation. 12…J and D 01 +D 12 +…+D t-1,t , use the parameters that match the actual order, and thus find the W value under different path planning. The path corresponding to the minimum T value is the optimal path under the current conditions.

Images