A semantic map construction method based on lidar and RFID

Abstract

The present invention discloses a method for constructing a semantic map based on laser radar and RFID. The semantic map consists of a grid map and a semantic information library; the semantic map construction method first uses SLAM to construct a grid map, and then the mobile robot obtains RFID tag information and laser radar scanning in the environment to obtain the coordinate information of the first object or the second object in the map, and finally updates the semantic information library. The present invention has the advantages of efficient and rapid acquisition of the coordinate information of the first object or the second object in the map; improves the robot's understanding of indoor semantic information, and obtains more accurate coordinate information of the first object label; does not perform precise identification on the acquisition of the coordinate information of the object label, so that it can meet the needs of navigation tasks; uses a graphic database to store semantic information, which enables other knowledge forms of the robot to be combined with map semantic information to better understand human language and maximize the use of semantic information.

Description

Technical Field

[0001] The present invention relates to a semantic map construction method, in particular to a semantic map construction method based on laser radar and RFID. Background Art

[0002] As robot map building technology becomes more mature, researchers have turned to semantic maps for extensive research. The key is semantic information, which is used to make the attributes and coordinates of objects on the map easier for robots to use. However, semantic map building technology has some flaws.

[0003] First, in terms of object semantic information extraction technology, researchers focus on segmenting objects using robot vision information, or using lidar technology combined with visual recognition technology to segment semantic objects. This segmentation requires a lot of algorithms and computing resources to achieve, and it is difficult to segment some objects using general visual recognition methods.

[0004] The second is the lack of recognition of some objects. Most semantic information extraction focuses on identifying larger objects such as sofas and pillows on sofas, while ignoring other objects such as TV remote controls, key chains, etc. These objects are very important for the robot's understanding of the environment and navigation tasks.

[0005] Third, the method of storing semantic information makes the semantic information base less reusable and less convenient to manage. Most map semantic information data is not stored in a suitable database after collection, and some semantic information is even directly stored in a file. Summary of the invention

[0006] The purpose of the present invention is to provide a method for constructing a semantic map based on laser radar and RFID. The present invention has the advantages of obtaining the coordinate information of the first object or the second object in the map efficiently and quickly; the first object is an object with a length and width greater than 0.45 meters, and the second object is an object with a length and width less than or equal to 0.45 meters, which improves the robot's understanding of indoor semantic information; the line segment information identified by laser radar scanning and RFID technology are used to jointly identify the first object label, and obtain more accurate coordinate information of the first object label; the acquisition of the coordinate information of the second object is not accurately identified, so that it can meet the needs of navigation tasks; the use of a graphic database to store semantic information can enable other knowledge forms of the robot to combine map semantic information to better understand human language and maximize the use of semantic information; the coordinate information of most objects in the indoor environment can be obtained, especially the coordinate information of objects with no length and width or small length and width that are not easy to extract, which often determines the execution of navigation tasks.

[0007] The technical solution of the present invention is a method for constructing a semantic map based on laser radar and RFID, characterized in that: the semantic map is composed of a grid map and a semantic information library; the semantic map construction method first uses SLAM to construct a grid map, and then a mobile robot obtains RFID tag information and line segment information scanned by the laser radar in the environment to obtain coordinate information of the first object or the second object in the map, and finally updates the semantic information library.

[0008] In the aforementioned method for constructing a semantic map based on lidar and RFID, the SLAM constructs a grid map using a gmapping algorithm; the semantic information base is composed of maps stored in a graphic database Neo4j.

[0009] In the aforementioned method for constructing a semantic map based on lidar and RFID, the method for updating the semantic information library is as follows: the semantic information library is initialized when defining the RFID tag information, the semantic information library obtains a graph-type node containing the RFID tag information, and the first object tag and the second object tag are obtained by classifying the RFID tags. The line segment information identified by the lidar scanning and the RFID tag information identified by the RFID device are fused and matched to obtain the coordinate information of the first object or the second object in the map, the coordinate information of the first object or the second object in the map is obtained and compared with the original coordinate information of the first object or the second object in the semantic information library. When the comparison results are different, the coordinate information of the first object or the second object in the map is obtained to overwrite the previously existing data in the semantic information library to obtain the latest map semantic information.

[0010] In the aforementioned semantic map construction method based on lidar and RFID, the RFID tag is classified as a first item tag or a second item tag; the first item is an item whose length and width are both greater than 0.45 meters, and the second item is an item whose length and width are both less than or equal to 0.45 meters; the RFID tags are respectively provided with RFID tag numbers, the RFID tag numbers are all different, the RFID tag numbers correspond to the first item or the second item information respectively, the first item or the second item information corresponds to the item name and the item length and width respectively, and the RFID tag number and the first item or the second item information are stored in a semantic information library, the semantic information library is initialized, and the RFID tags are correspondingly affixed to the first item or the second item.

[0011] In the aforementioned semantic map construction method based on lidar and RFID, the RFID tag information identification method is as follows: the first object tag or the second object tag is an RFID tag number obtained by identifying the RFID device to match the semantic information library, obtain the first object or the second object information, use the information of the first object or the second object and the line segment information identified by the lidar scan to match, and determine the coordinate information of the first object tag or the second object tag.

[0012] In the aforementioned semantic map construction method based on lidar and RFID, the information of the first object or the second object is matched with the line segment information scanned and identified by the lidar to determine the coordinate information of the first object tag or the second object tag; specifically, when the robot simultaneously identifies the first object or the second object and matches the line segment identified by the lidar, the robot will move to the target point 0.40 meters in front of the line segment, and use the enhanced signal strength RSSI returned by the tag to determine whether the first object or the second object tag matches. If it matches, the coordinate information is obtained. If it does not match, another lidar scan line segment is identified, and the robot is moved to 0.40 meters in front of the line segment for matching again.

[0013] In the aforementioned method for constructing a semantic map based on laser radar and RFID, obtaining the coordinate information of the first object tag includes the following steps:

[0014] The coordinate information of the first item tag is obtained based on the coordinates of the starting point A and the end point B of the line segment identified by the laser radar. Here, A and B identified by the radar are coordinates of the radar coordinate system. In the present invention, the radar and the robot belong to the same coordinate system on a two-dimensional scale. In the coordinate acquisition, A and B are first converted to the map coordinate system OXY according to the formula, and the first item is regarded as a rectangle ABCD on the radar scanning plane, where AB is the robot scanning line segment, and then the C and D coordinates are calculated, and the four coordinates are used as the coordinates of the first item tag;

[0015] map P= map R l l P+t (1.1)

[0016] in map p represents a point in the map coordinate system; map R l The rotation relationship from the radar coordinate system to the map coordinate system is obtained by the formula, where θ is the heading angle of the robot;

[0017]

[0018] l P represents the coordinates of a point in the radar coordinate system, which is represented by a column matrix; t represents the translation relationship from the radar coordinate system to the map coordinate system, that is, t = [x r ,y r ] T , where (x r ,y r ) represents the coordinates of the robot in the map coordinate system;

[0019] When the robot reaches a certain point (x r ,yr ) when its heading angle is θ, and the starting coordinate A of the radar scanning line segment is substituted lidar (x s ,y s )for l P = [x s ,y s ] T , that is, the coordinates of A in the robot coordinate system, we can get the coordinates of A in the map coordinate system (x A ,y A ); Similarly, we can get B(x B ,y B );

[0020] To obtain the C and D coordinates, the A and B coordinates can be calculated. According to the formula, the C and D coordinates (x C ,y C ), (x D ,y D );

[0021] (x C ,y C )=(x B +L4,y B +L3);(x D ,y D )=(x A +L4,y A +L3) (1.3)

[0022] Where L3 and L4 are calculated by the formula, and rat is the ratio of the length to the width of the rectangle ABCD. Specifically, when the radar scanning line segment is the length of the rectangle ABCD, rat is equal to the ratio of the length to the width of the rectangle ABCD, otherwise rat is equal to the ratio of the width to the length of the rectangle ABCD;

[0023]

[0024] In the aforementioned semantic map construction method based on laser radar and RFID, the coordinate information of the second object is divided into the second object tag inside the first object and the second object tag not inside the first object. Specifically, when the robot recognizes the second object tag, it will first recognize the first object tag multiple times: if the first object tag is recognized, the second object tag is classified as inside the first object, and the coordinates of the second object tag are set to the midpoint of the laser radar scanning segment during the first object tag matching process, that is, the midpoint of the radar scanning segment AB; if the first object tag cannot be recognized, the second object tag is classified as outside the first object, and the coordinates of the second object tag are the coordinates of the robot at this moment in the map (x r ,y r ).

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] The coordinate information of the first item tag of the present invention is obtained based on the laser radar scanning of the two-dimensional line segment contour of the first item tag, and the data is relatively accurate. The recognition of the coordinate information of the second item tag is based on the coordinate information of the robot body and the coordinate information of the first item tag. When the first item tag is recognized at the same time as the second item tag, the second item tag is classified into the first item, and the coordinates of the second item tag are set to the midpoint of the laser radar scanning line segment during the matching process of the first item tag, otherwise the coordinates of the second item tag are the position in the map when the robot recognizes the item tag.

[0027] Semantic map construction methods such as Figure 1 As shown in the figure, the semantic map consists of a grid map and a semantic information base. The grid map can obtain the geometric features of the environment space, which is beneficial to the robot's path planning, positioning and robot obstacle avoidance; the semantic information base is composed of node connections, which not only stores information such as items and coordinates, but also contains the spatial relationship between the first item and the second item, and can analyze the robot's service tasks to obtain the navigation target point.

[0028] Semantic information update method:

[0029] Semantic information updating methods such as Figure 2 As shown, the semantic information library is initialized when the RFID tag information is defined, and the graph database of the semantic information library obtains a graph node containing the RFID tag information. By classifying the RFID tags, the first object tag and the second object tag are obtained, and the line segment information identified by the laser radar scanning and the RFID tag information identified by the RFID device are fused and matched to obtain the coordinate information of the first object or the second object in the map. The coordinate information of the first object or the second object in the map is compared with the graph node coordinate information in the semantic information library. When the comparison results are different, the coordinate information of the first object or the second object in the map is obtained to overwrite the previously existing data in the semantic information library to obtain the latest map semantic information.

[0030] The present invention uses an RFID reader to identify RFID tags in the environment and obtain information about the first object and the second object. This identification technology is efficient and fast. The RFID tags are divided into first object tags and second object tags, which improves the robot's understanding of indoor semantic information. The line segment information (such as the line segment information) identified by laser radar scanning is Figure 3The robot uses a graphical database to store semantic information, so that the robot's map semantic information includes the first item information and the relationship between the second item, and enables the robot's other knowledge forms to be combined with map semantic information to better understand human language, thereby maximizing the use of semantic information.

[0031] The present invention can obtain the coordinate information of most objects in the indoor environment, especially the coordinate information of small objects with small length and width that are not easy to extract, which often determines the execution of navigation tasks. The robot semantic map is constructed by combining the grid map with the atlas semantic information library stored in the graphic database Neo4j. Objects in the indoor environment are classified as first objects or second objects; the first object is an object with a length and width greater than 0.45 meters, and the second object is an object with a length and width less than or equal to 0.45 meters. RFID tag data is designed based on this. The semantic information of the present invention is the coordinate information of the initially designed first object tag and the second object tag, as well as the ownership relationship between the first object and the second object. Figure 4 As shown, the first object is regarded as a rectangle ABCD, and OXY is a map coordinate system; in the present invention, the coordinates of the first object label are obtained as four coordinates, and the A and B coordinates (A and B in the first object label coordinate acquisition) of the radar scan are obtained by coordinate conversion, and the other two coordinates are obtained by formula. The present invention divides the second object label into whether it is located inside the first object label, and obtains the object label coordinate information through the line segment information identified by RFID technology and laser radar scanning, thereby extracting the object label coordinate information that is not easy to obtain. In the construction of the semantic map, a graph database is used as a semantic information library. The present invention visualizes the indoor semantic information as nodes and relationships in the graph database, and updates the acquired semantic information in real time. The semantic information acquired by the present invention includes the coordinate information of the first object or the second object in the map. In the present invention, the semantic information library is initialized to the RFID tag and the corresponding object information, and is respectively provided with RFID tag numbers, the RFID tag numbers are all different, the RFID tag numbers respectively correspond to the first object or the second object information, the first object or the second object information respectively corresponds to the object name and the object length and width, and the RFID tag number and the first object or the second object information are stored in the semantic information library. In the updating of the semantic information, the initial semantic information library data is matched with the tag information recognized by the RFID in real time to update the missing or outdated coordinate information in the semantic information library.

[0032] In summary, the present invention has the advantages of efficient and rapid acquisition of coordinate information of the first object or the second object in the map; the first object is an object with a length and width both greater than 0.45 meters, and the second object is an object with a length and width both less than or equal to 0.45 meters, which improves the robot's understanding of indoor semantic information; the line segment information identified by laser radar scanning and RFID technology are used to jointly identify the first object tag to obtain relatively accurate coordinate information of the first object tag; the acquisition of the coordinate information of the second object is not accurately identified, so that it can meet the needs of navigation tasks; the use of a graphic database to store semantic information can enable other knowledge forms of the robot to be combined with map semantic information to better understand human language and maximize the use of semantic information; the coordinate information of most objects in the indoor environment can be acquired, especially the coordinate information of objects with no length or width or small length and width that are difficult to extract, which often determines the beneficial effect of the execution of navigation tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flow chart of the semantic map construction method of the present invention; Figure 1 The left and right boxes in the middle represent the functions of the grid map and the semantic information base in the semantic map respectively.

[0034] Figure 2 It is a flow chart of the method for updating the semantic information base of the present invention; wherein the robot map coordinates are the coordinates of the robot in the map obtained in real time.

[0035] Figure 3 It is a line segment information diagram scanned and recognized by the laser radar of the present invention, wherein the robot is in the square box and the line segment scanned and recognized by the laser radar is in the elliptical box;

[0036] Figure 4 is the first object coordinate acquisition diagram of the present invention;

[0037] Figure 5 It is a grid map of an embodiment of the present invention, wherein the white area No. 1 is an idle state area, the gray area No. 2 is an unknown state area, and the black line segment pointed to by 3 represents an obstacle in the grid map;

[0038] Figure 6 It is the semantic information library diagram of the present invention;

[0039] Figure 7 is a comparison diagram of the coordinates of the first item label of the present invention;

[0040] Figure 8 is a second item tag recognition time diagram of the present invention;

[0041] Fig. 9 is the radar identification error of the present invention;

[0042] Fig.10The present invention obtains the A and B coordinate diagrams of the first object label coordinates multiple times. DETAILED DESCRIPTION

[0043] The present invention is further described below in conjunction with the accompanying drawings and embodiments, but they are not intended to limit the present invention.

[0044] Embodiment: A semantic map construction method based on lidar and RFID, the semantic map consists of a grid map and a semantic information library; the semantic map construction method first uses SLAM to construct a grid map, then a mobile robot obtains RFID tag information and line segment information scanned by the lidar in the environment to obtain coordinate information of a first object or a second object in the map, and finally updates the semantic information library.

[0045] The SLAM constructs a grid map using the gmapping algorithm; the semantic information base is composed of graphs stored in the graphic database Neo4j.

[0046] The method for updating the semantic information library is as follows: the semantic information library is initialized when the RFID tag information is defined, the semantic information library obtains a graph-type node containing the RFID tag information, and the first object tag and the second object tag are obtained by classifying the RFID tags, and the coordinate information of the first object or the second object in the map is obtained by fusion matching of the line segment information identified by the laser radar scanning and the RFID tag information identified by the RFID device, and the coordinate information of the first object or the second object in the map is obtained and compared with the original coordinate information of the first object or the second object in the semantic information library. When the comparison results are different, the coordinate information of the first object or the second object in the map is obtained to overwrite the previously existing data in the semantic information library to obtain the latest map semantic information.

[0047] The RFID tag is classified as a first item tag or a second item tag; the first item is an item whose length and width are both greater than 0.45 meters, and the second item is an item whose length and width are both less than or equal to 0.45 meters; the RFID tags are respectively provided with RFID tag numbers, the RFID tag numbers are all different, the RFID tag numbers correspond to the first item or the second item information respectively, the first item or the second item information corresponds to the item name and the item length and width respectively, and the RFID tag number and the first item or the second item information are stored in a semantic information library, the semantic information library is initialized, and the RFID tag is correspondingly attached to the first item or the second item.

[0048] The RFID tag information identification method: the first object tag or the second object tag is identified by an RFID device to obtain an RFID tag number to match the semantic information library, obtain the first object or the second object information, use the information of the first object or the second object and the line segment information identified by the laser radar scanning to match, and determine the coordinate information of the first object tag or the second object tag.

[0049] The information of the first object or the second object is matched with the line segment information scanned and identified by the laser radar to determine the coordinate information of the first object label or the second object label. Specifically, when the robot simultaneously identifies the first object or the second object and matches the line segment identified by the laser radar, the robot will move to the target point 0.40 meters in front of the line segment, and use the RSSI of the signal strength returned by the tag to determine whether the first object or the second object label matches. If it matches, the coordinate information is obtained. If it does not match, another laser radar scan line segment is identified, and the robot is moved to 0.40 meters in front of the line segment for matching again.

[0050] The first item tag coordinate information acquisition comprises the following steps:

[0051] The coordinate information of the first item tag is obtained based on the coordinates of the starting point A and the end point B of the line segment identified by the laser radar. Here, A and B identified by the radar are coordinates of the radar coordinate system. In the present invention, the radar and the robot belong to the same coordinate system on a two-dimensional scale. In the coordinate acquisition, A and B are first converted to the map coordinate system OXY according to the formula, and the first item is regarded as a rectangle ABCD on the radar scanning plane, where AB is the robot scanning line segment, and then the C and D coordinates are calculated, and the four coordinates are used as the coordinates of the first item tag;

[0052] map P= map R l l P+t (1.1)

[0053] in map p represents a point in the map coordinate system; map R l The rotation relationship from the radar coordinate system to the map coordinate system is obtained by the formula, where θ is the heading angle of the robot;

[0054]

[0055] l P represents the coordinates of a point in the radar coordinate system, which is represented by a column matrix; t represents the translation relationship from the radar coordinate system to the map coordinate system, that is, t = [x r ,y r ] T , where (x r ,yr ) represents the coordinates of the robot in the map coordinate system;

[0056] When the robot reaches a certain point (x r ,y r ) when its heading angle is θ, and the starting coordinate A of the radar scanning line segment is substituted lidar (x s ,y s )for l P = [x s ,y s ] T , that is, the coordinates of A in the robot coordinate system, we can get the coordinates of A in the map coordinate system (x A ,y A ); Similarly, we can get B(x B ,y B );

[0057] To obtain the C and D coordinates, the A and B coordinates can be calculated. According to the formula, the C and D coordinates (x C ,y C ), (x D ,y D );

[0058] (x C ,y C )=(x B +L4,y B +L3);(x D ,y D )=(x A +L4,y A +L3) (1.3)

[0059] Where L3 and L4 are calculated by the formula, and rat is the ratio of the length to the width of the rectangle ABCD. Specifically, when the radar scanning line segment is the length of the rectangle ABCD, rat is the ratio of the length to the width of the rectangle ABCD, otherwise rat is the ratio of the width to the length of the rectangle ABCD;

[0060]

[0061] The coordinate information of the second object is divided into the second object tag that is inside the first object and the second object tag that is not inside the first object. Specifically, when the robot recognizes the second object tag, it will first recognize the first object tag multiple times: if the first object tag is recognized, the second object tag is classified as inside the first object, and the coordinates of the second object tag are set to the midpoint of the laser radar scanning line segment during the first object tag matching process, that is, the midpoint of the radar scanning line segment AB; if the first object tag is not recognized, the second object tag is classified as outside the first object, and the coordinates of the second object tag are the coordinates of the robot at this moment in the map (x r ,yr ).

[0062] Grid ground of the embodiment of the present invention Figure 5 As shown, the black line segment pointed to by 3 represents an obstacle in the grid map, the gray area is the area marked by 2 in the unknown state diagram, and the white area is the area marked by 1 in the idle state diagram. The obstacles in the map are the first object, the second object, the wall or a few other indoor objects in the room that are not represented in this example. The rectangular boxes numbered ABCDE in the figure are the first objects of this solution, and the second object is not marked, and it is generally stored in the first object.

[0063] Semantic information base such as Figure 6 As shown in the figure, the laboratory is the experimental room with different labels attached to it. The semantic information base obtained from the experiment is the knowledge graph stored in neo4j. The semantic information is stored in the form of nodes in the graph stored in the graph database Neo4j. For example: Figure 6 The metal box in the middle node is a first item tag, and the pen is a second item tag, and the relationship between them is have and in; both the first item tag node and the second item tag node can be connected to the room lib node, and their relationship is also in and have. The scheme only completes the experiment in one room, so a room tag is set.

[0064] By comparing the coordinate information queried in the Neo4j database with the actual measured coordinate information, we can get Figure 7 The coordinate information of the first item label is shown in the figure. OXY is the robot map coordinate system, and rectangle ABCD is the actual position of the first item. The position of the first item measured by our solution is the rectangle A`B`C`D` with a 5-degree tilt in the figure. It can be seen that the distance from the actual measured straight line AB to the Y axis is FH=1.145 meters, while the distances from A` and B` to the Y axis measured by the solution are B`G=1.174 and A`I=1.137 meters respectively. The difference in the map is not big, and the position obtained by the robot is credible and relatively accurate.

[0065] like Figure 8 The figure shows the comparison of recognition time for different numbers of tags after a first object tag is identified. The horizontal axis is the number of first object tags, and the vertical axis is the time for identifying multiple first object tags. It can be seen that after the first object is identified, the measurement time for different numbers of second object tags on the first object is compared. The identification time for more than 10 second object tags is only 5 seconds, and many tags can also be identified and added to the database in a very short time.

[0066] Fig. 9is the radar recognition error, that is, the length of the line segment extracted by the radar when the robot is at different distances from the first object. The upper figure is the line segment extraction of the width of the table, and the lower figure is the line segment extraction of the width of the metal box. The actual values ​​of both are 0.52 meters. It is found that the extracted line segment of the metal box is 0.03-0.06 meters smaller than the actual value on average, and the extracted line segment of the table is 0.01-0.04 meters smaller than the actual value. The data obtained for the 0.52-meter line segment recognition are all less than 0.52 meters. For the metal box, the radar recognition line segment will be 0.05 meters smaller than the actual average, and for the table, the radar recognition line segment will be 0.03 meters smaller than the actual average. Finally, the experiment calibrated the error. When the metal object is recognized, the length of the metal box scanned by the radar is matched with the length and width of the first object after adding 0.05 meters, and the length of the line segment scanned by the radar except for the metal box is matched with the length and width of the first object after adding 0.03 meters.

[0067] Fig.10 The horizontal axis is X, the vertical axis is Y, and the pose1 and pose2 coordinates in the first object label coordinate information obtained in the first object recognition algorithm are, that is, Figure 7 The coordinates of the midpoint A and B are identified multiple times. The experimental results are as follows Fig.10 The coordinates of the points in the figure are the coordinates of point A or point B in the map coordinate system. It can be clearly known that for each recognition of the position coordinates A and B of a first item label, the position error of the coordinate point is within 0.03 meters. The experiment proves that the position obtained by coordinate transformation in the first item recognition is reliable and relatively stable.

[0068] The position of the second item tag, the first type: when the first item tag is identified at the same time as the second item tag, the second item tag is placed inside the first item, and the second item tag position is the midpoint of the A and B coordinates of the four coordinates of the first item, and its position indicates that the second item belongs to the first item. The second type is when the first item tag is not identified when the second item tag is identified, the coordinates of the second item tag are represented by the coordinates of the current robot in the map, and its position indicates that the second item does not belong to the first item.

[0069] Coordinate point information release

[0070] A service is defined in ROS. The robot service task queries the environment information by only creating a client and sending an ID request response data. The server will then send the location of the object corresponding to the tag number to facilitate robot navigation.

Claims

1. A semantic map construction method based on LiDAR and RFID, Features: The semantic map consists of a grid map and a semantic information library. The semantic map construction method first uses SLAM to construct a grid map, then the mobile robot obtains RFID tag information and line segment information scanned by the laser radar in the environment to obtain the coordinate information of the first object or the second object in the map, and finally updates the semantic information library. The SLAM constructs a grid map using the gmapping algorithm; the semantic information base is composed of graphs stored in the graphic database Neo4j; The method for updating the semantic information library is as follows: the semantic information library is initialized when the RFID tag information is defined, the semantic information library obtains a graph node containing the RFID tag information, the RFID tags are classified to obtain a first item tag and a second item tag, the line segment information identified by the laser radar scanning and the RFID tag information identified by the RFID device are fused and matched to obtain the coordinate information of the first item or the second item in the map, the coordinate information of the first item or the second item in the map is obtained and compared with the original coordinate information of the first item or the second item in the semantic information library, and when the comparison results are different, the coordinate information of the first item or the second item in the map is obtained to cover the previously existing data in the semantic information library to obtain the latest map semantic information; The RFID tags are classified as first item tags or second item tags; the first item is an item whose length and width are both greater than 0.45 meters, and the second item is an item whose length and width are both less than or equal to 0.45 meters; the RFID tags are respectively provided with RFID tag numbers, the RFID tag numbers are all different, the RFID tag numbers correspond to the first item or the second item information respectively, the first item or the second item information corresponds to the item name and the item length and width respectively, and the RFID tag number and the first item or the second item information are stored in a semantic information library, the semantic information library is initialized, and the RFID tags are correspondingly attached to the first item or the second item; The RFID tag information identification method: the first item tag or the second item tag is identified by an RFID device to obtain an RFID tag number to match the semantic information library, obtain the first item or the second item information, and use the information of the first item or the second item to match the line segment information identified by the laser radar scanning to determine the coordinate information of the first item tag or the second item tag; The information of the first object or the second object is matched with the line segment information scanned and identified by the laser radar to determine the coordinate information of the first object tag or the second object tag; specifically, when the robot simultaneously identifies the first object or the second object and matches the line segment identified by the laser radar, the robot moves to the target point 0.40 meters in front of the line segment, and uses the RSSI of the signal strength returned by the tag to determine whether the first object or the second object tag is matched. If it is matched, the coordinate information is obtained; if it is not matched, another laser radar scanning line segment is identified, and the robot is moved to 0.40 meters in front of the line segment for matching again; The first item tag coordinate information acquisition comprises the following steps: The coordinate information of the first item tag is obtained based on the coordinates of the starting point A and the end point B of the line segment identified by the laser radar. Here, A and B identified by the radar are coordinates of the radar coordinate system. In the present invention, the radar and the robot belong to the same coordinate system on a two-dimensional scale. In the coordinate acquisition, A and B are first converted to the map coordinate system OXY according to the formula, and the first item is regarded as a rectangle ABCD on the radar scanning plane, where AB is the robot scanning line segment, and then the C and D coordinates are calculated, and the four coordinates are used as the coordinates of the first item tag; map P= map R l l P+t(1.1) in map p represents a point in the map coordinate system; map R l The rotation relationship from the radar coordinate system to the map coordinate system is obtained by the formula, where θ is the heading angle of the robot; l P represents the coordinates of a point in the radar coordinate system, which is represented by a column matrix; t represents the translation relationship from the radar coordinate system to the map coordinate system, that is, t = [x r ,y r ] T , where (x r ,y r ) represents the coordinates of the robot in the map coordinate system; When the robot reaches a certain point (x r ,y r ) when its heading angle is θ, and the starting coordinate A of the radar scanning line segment is substituted lidar (x s ,y s )for l P = [x s ,y s ] T , that is, the coordinates of A in the robot coordinate system, we can get the coordinates of A in the map coordinate system (x A ,y A ); Similarly, we can get B(x B ,y B ); To obtain the C and D coordinates, the A and B coordinates can be calculated. According to the formula, the C and D coordinates (x C ,y C ), (x D ,y D ); (x C ,y C )=(x B +L4,y B +L3);(x D ,y D )=(x A +L4,y A +L3)(1.3) where L3 and L4 are calculated by the formula, rat is the ratio of the length to the width of the rectangle ABCD. Specifically, when the radar scanning line segment is the length of the rectangle ABCD, rat is equal to the ratio of the length to the width of the rectangle ABCD, otherwise rat is equal to the ratio of the width to the length of the rectangle ABCD; The coordinate information of the second object is divided into the second object tag that is inside the first object and the second object tag that is not inside the first object. Specifically, when the robot recognizes the second object tag, it will first recognize the first object tag multiple times: if the first object tag is recognized, the second object tag is classified as inside the first object, and the coordinates of the second object tag are set to the midpoint of the laser radar scanning line segment during the first object tag matching process, that is, the midpoint of the radar scanning line segment AB; if the first object tag is not recognized, the second object tag is classified as outside the first object, and the coordinates of the second object tag are the coordinates of the robot at this moment in the map (x r ,y r ).

Images