Inertial navigation and visual fusion navigation method for multi-camera single-steering-wheel automatic guided vehicle

Abstract

The invention relates to the technical field of logistics, in particular to an inertial navigation and visual fusion navigation method for a multi-camera single-steering-wheel automatic guided vehicle, adopts a technical scheme of fusion navigation by adopting multiple means, and is a fusion navigation method based on vision SLAM (Simultaneous Localization and Mapping) and inertial navigation and two-dimensional code navigation as assistance. According to the method, correction in the state that visual navigation positioning is lost or navigation errors are too large is guaranteed, it is ensured that a vehicle body returns to a correct advancing route, meanwhile, the accumulative errors of inertial navigation are reduced, the problem of the accumulative errors of inertial navigation is solved through recalibration of the two-dimensional code, the positioning position in visual SLAM navigation is recalibrated through positioning data of inertial navigation, and the navigation accuracy is improved. The positioning accuracy of the SLAM is improved, the cargo inserting and taking efficiency is improved, and in this way, the technical problems that in the prior art, existing visual SLAM navigation excessively depends on feature points, and positioning is prone to being lost are solved.

Description

Technical Field

[0001] The present invention relates to the field of logistics technology, and in particular to an inertial navigation and vision fusion navigation method for a multi-camera single-steering wheel unmanned guided vehicle. Background Art

[0002] Logistics warehousing is an important part of the modern logistics system. Usually, unmanned guided vehicles equipped with visual SLAM navigation are used to insert and remove goods. In a standardized warehousing environment, due to the constant entry and exit of goods, if the goods themselves become feature point objects for visual SLAM mapping, the corresponding feature points will disappear after the goods are entered and exited, thus affecting visual SLAM navigation. However, if the shelves in the cleared state are simply used as feature point mapping objects, the shelves will be almost identical and unable to identify their own positions.

[0003] In summary, the existing visual SLAM navigation has the problems of over-reliance on feature points and easy loss of positioning. Summary of the invention

[0004] The purpose of the present invention is to provide an inertial navigation and vision fusion navigation method for a multi-camera single-steering wheel unmanned guided vehicle, aiming to solve the technical problems in the prior art that the existing visual SLAM navigation has excessive reliance on feature points and is prone to positioning loss.

[0005] To achieve the above purpose, the present invention adopts a multi-camera single-steering wheel unmanned guided vehicle inertial navigation and vision fusion navigation method.

[0006] The following steps are involved:

[0007] Step 1: Based on visual SLAM navigation, scan and build maps when the warehouse is in a clearing state, and post QR code labels in the warehouse;

[0008] Step 2: After the cargo pickup instruction is issued, the UGV plans the route according to the acquired target coordinates, and navigates towards the target point through visual SLAM. During the navigation process, the SLAM navigation will have a positioning coordinate A of the UGV. When the UGV is moving, the IMU inertial navigation will work at the same time, and there will be a positioning coordinate B of the inertial navigation. At this time, the mean fusion calculation based on SLAM will be performed based on the coordinate information of the two to reduce the driving error;

[0009] Step 3: When the AGV enters the shelf area, the recognition module on the side of the AGV recognizes the QR code label and parses the QR code to obtain the positioning coordinates C of the AGV relative to the QR code and determine whether the cargo is the target cargo.

[0010] Step 4: When the QR code of the target cargo is identified, the unmanned guided vehicle makes corresponding posture adjustments according to the location C given by the QR code of the target cargo. At this time, the navigation data of the unmanned guided vehicle is covered by the QR code coordinates, the unmanned guided vehicle makes corresponding posture adjustments, and then inserts and picks up the cargo.

[0011] Among them, in step one, the posted QR code label is posted on the shelf in the venue, and the QR code label includes coordinate positioning and cargo information.

[0012] Among them, in step 2, when the SLAM navigation is lost during the driving process of the unmanned guided vehicle, the unmanned guided vehicle continues to drive with the positioning coordinate B of the inertial navigation as the positioning data. When the visual SLAM finds the feature point again and continues to obtain the current positioning A, it returns to the state of coordinate navigation calculated by fusing the two navigations.

[0013] Among them, in step three, the recognition module on the side of the unmanned guided vehicle is a QR code camera.

[0014] Among them, in step 4, when the QR code of the target cargo is identified and it is determined that the unmanned guided vehicle is parallel to the cargo:

[0015] Based on the QR code scanning results, the camera calculates that the number of pixels from the center of the QR code to both sides is equal, so it is determined that the vehicle body is parallel to the shelf. At this time, there is no need to adjust the body posture. By calculating the center point of the QR code, the distance d1 of the target vertical line relative to the camera (X, Y) can be determined. Then, according to the vehicle body design parameters: the positioning of the vehicle body rotation point can obtain the vertical distance d2 between the vehicle body rotation center point (X2, Y2) and the camera (X, Y). From this, the distance between the point where the vehicle body and the target vertical line coincide after the rotation of the vehicle body can be calculated: d1+d2; then the vehicle body is controlled to move forward for a distance of d1+d2, and rotate 90° on the spot, so that the vehicle body is perpendicular to the shelf and coincides with the target vertical line, and finally moves forward to pick up the goods.

[0016] Among them, in step 4, when the QR code of the target cargo is identified and it is determined that the unmanned guided vehicle is not parallel to the cargo:

[0017] At this time, the camera performs decoding calculation based on the image of the QR code, and determines the deflection angle α between the camera and the vertical line of the QR code based on the pixel distortion of the QR code in the image, and then controls the body to adjust the posture, rotate α and move forward a short distance. At this time, the body is parallel to the shelf surface, and the camera continues to shoot and decode, and then inserts the goods according to the above-mentioned steps of inserting goods parallel to the shelf.

[0018] The invention discloses a method for inertial navigation and visual fusion navigation of a multi-camera single-steering wheel unmanned guided vehicle, a technical solution for fusion navigation by multiple means, a fusion navigation method based on visual SLAM as the main method, inertial navigation and two-dimensional code navigation as the auxiliary method, which ensures correction when visual navigation positioning is lost or navigation error is too large, ensures that the vehicle body returns to the correct route, and reduces the accumulated error of inertial navigation at the same time. The problem of accumulated error of inertial navigation is solved by recalibrating the two-dimensional code, and the positioning position in the visual SLAM navigation is recalibrated by the positioning data of inertial navigation, which improves the positioning accuracy of SLAM and the efficiency of inserting and taking goods. Under the premise of accurate navigation, the action of inserting and taking goods after turning can be completed in one step, and there is no need to readjust the posture of the vehicle body again or multiple times due to misalignment of the position after turning so as to insert the goods again, thereby improving efficiency. In this way, the technical problems of the existing visual SLAM navigation in the prior art that the existing visual SLAM navigation is too dependent on feature points and the positioning is easy to lose are solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 It is a navigation logic diagram of the inertial navigation and vision fusion navigation method of the multi-camera single-steering wheel unmanned guided vehicle of the present invention.

[0021] Figure 2 It is a specific insertion positioning logic diagram of the inertial navigation and vision fusion navigation method of the multi-camera single-steering wheel unmanned guided vehicle of the present invention.

[0022] Figure 3 It is a schematic diagram of the actual working process when the unmanned guided vehicle of the present invention is not parallel to the cargo.

[0023] Figure 4 It is a schematic diagram of the actual working process when the unmanned guided vehicle of the present invention is parallel to the cargo. DETAILED DESCRIPTION

[0024] Embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, but should not be construed as limiting the present invention.

[0025] See also Figure 1 to Figure 4 ,in Figure 1 It is a navigation logic diagram of the inertial navigation and vision fusion navigation method of the multi-camera single-steering wheel unmanned guided vehicle of the present invention, Figure 2 It is a specific insertion positioning logic diagram of the inertial navigation and vision fusion navigation method of the multi-camera single-steering wheel unmanned guided vehicle of the present invention, Figure 3 This is a schematic diagram of the actual working process when the unmanned guided vehicle of the present invention is not parallel to the cargo. Figure 4 It is a schematic diagram of the actual working process when the unmanned guided vehicle of the present invention is parallel to the cargo.

[0026] The present invention provides an inertial navigation and vision fusion navigation method for a multi-camera single-steering wheel unmanned guided vehicle, comprising the following steps:

[0027] The following steps are involved:

[0028] S101, based on visual SLAM navigation, scan and build maps when the warehouse is in a clearing state, and post QR code labels in the warehouse;

[0029] According to this specific implementation, the QR code labels are posted on shelves in the venue, and the QR code labels include coordinate positioning and cargo information.

[0030] S102, when the cargo insertion and removal instruction is issued, the unmanned guided vehicle plans the route according to the acquired target coordinates, and navigates toward the target point through visual SLAM. During the navigation process, the SLAM navigation will have a positioning coordinate A of the unmanned guided vehicle. When the unmanned guided vehicle is moving, the IMU inertial navigation will work at the same time, and there will be a positioning coordinate B of the inertial navigation. At this time, the mean fusion calculation based on SLAM will be performed based on the coordinate information of the two to reduce the driving error;

[0031] According to this specific implementation, when the SLAM navigation is lost during the driving of the unmanned guided vehicle, the unmanned guided vehicle continues to drive with the positioning coordinate B of the inertial navigation as the positioning data. When the visual SLAM finds the feature point again and continues to obtain the current positioning A, it returns to the state of coordinate navigation calculated by fusing the two navigation methods.

[0032] S103, when the unmanned guided vehicle enters the shelf area, the recognition module on the side of the unmanned guided vehicle recognizes the posted QR code label and parses the QR code, thereby obtaining the positioning coordinates C of the unmanned guided vehicle relative to the QR code and determining whether the cargo is the target cargo;

[0033] According to this specific implementation, the recognition module on the side of the unmanned guided vehicle is a QR code camera.

[0034] S104. When the QR code of the target cargo is identified, the unmanned guided vehicle performs a corresponding posture adjustment of the unmanned guided vehicle according to the location C given by the QR code of the target cargo. At this time, the navigation data of the unmanned guided vehicle is covered by the QR code coordinates, the unmanned guided vehicle performs a corresponding posture adjustment, and then inserts and takes the cargo.

[0035] According to this specific implementation, when the QR code of the target cargo is identified and it is determined that the unmanned guided vehicle is parallel to the cargo:

[0036] The camera calculates the number of pixels from the center of the QR code to both sides based on the QR code scanning result, and thus determines that the vehicle body is parallel to the shelf. At this time, there is no need to adjust the body posture. By calculating the center point of the QR code, the distance d1 of the target vertical line relative to the camera (X, Y) can be determined. Then, according to the vehicle body design parameters: the positioning of the vehicle body rotation point can obtain the vertical distance d2 between the vehicle body rotation center point (X2, Y2) and the camera (X, Y). From this, the distance of the point where the vehicle body and the target vertical line coincide after the rotation of the vehicle body can be calculated: d1+d2; then the vehicle body is controlled to move forward for a distance of d1+d2, and rotates 90° on the spot, so that the vehicle body is perpendicular to the shelf and coincides with the target vertical line, and finally moves forward to pick up the goods;

[0037] When the QR code of the target cargo is identified and it is determined that the unmanned guided vehicle is not parallel to the cargo:

[0038] At this time, the camera performs decoding calculation based on the image of the QR code, and determines the deflection angle α between the camera and the vertical line of the QR code based on the pixel distortion of the QR code in the image, and then controls the body to adjust the posture, rotate α and move forward a short distance. At this time, the body is parallel to the shelf surface, and the camera continues to shoot and decode, and then inserts the goods according to the above-mentioned steps of inserting goods parallel to the shelf.

[0039] The inertial navigation and visual fusion navigation method of a multi-camera single-steering wheel unmanned guided vehicle of the present invention is used, and a technical solution of fusion navigation is performed by multiple means. The fusion navigation method is mainly based on visual SLAM, and inertial navigation and two-dimensional code navigation are used as auxiliary methods. The correction of visual navigation positioning loss or excessive navigation error is ensured, and the vehicle body is ensured to return to the correct route of travel, while reducing the cumulative error of inertial navigation. The problem of cumulative error of inertial navigation is solved by recalibrating the two-dimensional code. The positioning position in the visual SLAM navigation is recalibrated by the positioning data of inertial navigation, which improves the positioning accuracy of SLAM and the efficiency of inserting and removing goods. Under the premise of accurate navigation, the action of inserting and removing goods after turning can be completed in one step, and there is no need to readjust the posture of the vehicle body again or multiple times due to misalignment of the position after turning, so as to insert and remove goods, thereby improving efficiency. In this way, the technical problems of the existing visual SLAM navigation in the prior art that the existing visual SLAM navigation is too dependent on feature points and the problem of easy loss of positioning are solved;

[0040] The method of pasting a QR code on the shelf is used for identification and navigation. The QR code pasted on the ground in the prior art can carry not only positioning information but also other information such as goods. On the other hand, it solves the problem that the ground-stick QR code is easily damaged when pasted on the ground, which is inconvenient for later maintenance.

[0041] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A multi-camera single-steering wheel unmanned guided vehicle inertial navigation and vision fusion navigation method, characterized in that: The following steps are involved: Step 1: Based on visual SLAM navigation, scan and build maps when the warehouse is in a clearing state, and post QR code labels in the warehouse; Step 2: After the cargo pickup instruction is issued, the UGV plans the route according to the acquired target coordinates, and navigates towards the target point through visual SLAM. During the navigation process, the SLAM navigation will have a positioning coordinate A of the UGV. When the UGV is moving, the IMU inertial navigation will work at the same time, and there will be a positioning coordinate B of the inertial navigation. At this time, the mean fusion calculation based on SLAM will be performed based on the coordinate information of the two to reduce the driving error; Step 3: When the AGV enters the shelf area, the recognition module on the side of the AGV recognizes the QR code label and parses the QR code to obtain the positioning coordinates C of the AGV relative to the QR code and determine whether the cargo is the target cargo. Step 4: When the QR code of the target cargo is identified, the unmanned guided vehicle makes corresponding posture adjustments according to the location C given by the QR code of the target cargo. At this time, the navigation data of the unmanned guided vehicle is covered by the QR code coordinates, the unmanned guided vehicle makes corresponding posture adjustments, and then inserts and picks up the cargo.

2. The inertial navigation and vision fusion navigation method for a multi-camera single-steering wheel unmanned guided vehicle as claimed in claim 1, characterized in that: In step one, the QR code label is posted on the shelf in the venue, and the QR code label includes coordinate positioning and cargo information.

3. The inertial navigation and vision fusion navigation method for a multi-camera single-steering wheel automated guided vehicle as claimed in claim 2, characterized in that: In step 2, when the SLAM navigation is lost during the driving process of the unmanned guided vehicle, the unmanned guided vehicle continues to drive with the positioning coordinate B of the inertial navigation as the positioning data. When the visual SLAM finds the feature point again and continues to obtain the current positioning A, it returns to the state of coordinate navigation calculated by fusing the two navigations.

4. The inertial navigation and vision fusion navigation method for a multi-camera single-steering wheel unmanned guided vehicle as claimed in claim 3, characterized in that: In step three, the recognition module on the side of the unmanned guided vehicle is a QR code camera.

5. The inertial navigation and vision fusion navigation method for a multi-camera single-steering wheel unmanned guided vehicle as claimed in claim 4, characterized in that: In step 4, when the QR code of the target cargo is identified and the unmanned guided vehicle is judged to be parallel to the cargo: Based on the QR code scanning results, the camera calculates that the number of pixels from the center of the QR code to both sides is equal, so it is determined that the vehicle body is parallel to the shelf. At this time, there is no need to adjust the body posture. By calculating the center point of the QR code, the distance d1 of the target vertical line relative to the camera (X, Y) can be determined. Then, according to the vehicle body design parameters: the positioning of the vehicle body rotation point can obtain the vertical distance d2 between the vehicle body rotation center point (X2, Y2) and the camera (X, Y). From this, the distance between the point where the vehicle body and the target vertical line coincide after the rotation of the vehicle body can be calculated: d1+d2; then the vehicle body is controlled to move forward for a distance of d1+d2, and rotate 90° on the spot, so that the vehicle body is perpendicular to the shelf and coincides with the target vertical line, and finally moves forward to pick up the goods.

6. The inertial navigation and vision fusion navigation method for a multi-camera single-steering wheel automated guided vehicle as claimed in claim 5, characterized in that: In step 4, when the QR code of the target cargo is identified and it is determined that the AGV is not parallel to the cargo: At this time, the camera performs decoding calculation based on the image of the QR code, and determines the deflection angle α between the camera and the vertical line of the QR code based on the pixel distortion of the QR code in the image, and then controls the body to adjust the posture, rotate α and move forward a short distance. At this time, the body is parallel to the shelf surface, and the camera continues to shoot and decode, and then inserts the goods according to the above-mentioned steps of inserting goods parallel to the shelf.

Images