Single-steering-wheel AGV high-precision mapping and robust positioning method based on two-dimensional laser SLAM

Abstract

The invention discloses a high-precision mapping and robust positioning method of a single-steering-wheel AGV based on two-dimensional laser SLAM. The method comprises the following steps: step 1, acquiring data of a two-dimensional laser radar on the AGV; step 2, constructing a structured two-dimensional grid map of the target environment based on an SLAM method; and step 3, real-time accurate positioning data of the AGV body are obtained by judging the occupation condition of the grids in the two-dimensional grid map. Map information can be updated in real time and accurate positioning can be carried out through continuous laser scanning and two-dimensional laser SLAM, high precision can be kept even in a complex or changing environment, human intervention and adjustment needed due to environment change are reduced, and the cost and complexity of long-term maintenance are reduced.

Description

Technical Field

[0001] The present invention relates to the field of AGV control technology, and in particular to a high-precision mapping and robust positioning method for a single-steering wheel AGV based on two-dimensional laser SLAM. Background Art

[0002] Automated Guided Vehicle (AGV) is an automated transportation system. As an important part of modern logistics and production, its autonomous navigation and positioning capabilities are the key to improving its work efficiency and flexibility. With the rapid development of intelligent manufacturing and intelligent logistics, higher requirements are placed on the positioning accuracy and environmental adaptability of AGV. Two-dimensional laser simultaneous positioning and mapping (SLAM) technology has become a popular choice for AGV positioning and navigation due to its high precision and low cost. However, the positioning robustness and mapping accuracy of existing technologies in complex environments are still facing challenges.

[0003] At present, common problems faced by AGVs when using 2D laser SLAM for navigation include: inaccurate positioning in man-made structure environments, inaccurate mapping in diverse environments, and restrictions on the construction layout of the implementation scenario. In addition, the special mechanical structure of the single-steering wheel AGV also brings additional challenges to high-precision map construction and robust positioning.

[0004] The existing navigation technologies used in single-steering wheel AGVs can generally be divided into: magnetic navigation technology that lays magnetic strips / magnetic nails, visual navigation technology that requires posting QR code markers, reflective column navigation technology with auxiliary equipment, and laser SLAM autonomous navigation technology.

[0005] Magnetic navigation technology uses magnetic markers such as magnetic strips or magnetic nails installed on the ground for navigation in indoor environments, and the magnetic sensors on the AGV read these magnetic markers to achieve navigation. However, any modification to the ground layout requires the rearrangement of magnetic strips or magnetic nails, which is not flexible enough in a dynamically changing environment, and once the navigation path of the AGV needs to be changed, the magnetic markers must be physically rearranged, which is time-consuming and laborious in operation.

[0006] QR code navigation places QR code markers in the environment and uses visual systems such as cameras to identify these markers to locate AGVs. However, QR code markers must be within the line of sight to be recognized, which limits the flexibility and robustness of navigation. In addition, changes in lighting, obstruction or damage of the QR code will affect navigation accuracy.

[0007] In reflective column navigation, reflective columns are usually placed at specific locations in the environment as physical markers, and these reflective columns are identified and located by laser scanners to achieve AGV navigation. However, this method requires the pre-installation of reflective columns, and these physical markers need to be reconfigured every time the environment layout changes. In addition, when the environment changes (such as new obstacles, changes in the purpose of the area, etc.), the adjustment cost of the reflective column navigation system is high.

[0008] Therefore, it is necessary to provide an improved single-steering wheel AGV path tracking method to solve the above problems. Summary of the invention

[0009] The purpose of the present invention is to provide a high-precision mapping and robust positioning method for a single-steering wheel AGV based on two-dimensional laser SLAM. Through continuous laser scanning, the two-dimensional laser SLAM can update map information in real time and perform precise positioning. It can maintain high accuracy even in complex or changing environments, reduce the human intervention and adjustment required due to environmental changes, and reduce the cost and complexity of long-term maintenance.

[0010] The technical solution of the present invention is: a high-precision mapping and robust positioning method for a single-steering wheel AGV based on two-dimensional laser SLAM, comprising the following steps:

[0011] Step 1: Obtain the data of the two-dimensional laser radar on the AGV;

[0012] Step 2: construct a structured two-dimensional grid map of the target environment based on the SLAM method;

[0013] Step 3: Obtain real-time and accurate positioning data of the AGV body by determining the occupancy of the grids in the two-dimensional grid map.

[0014] In the above technical solution, in step 1, the two-dimensional laser radar is set on the top of the AGV body.

[0015] In the above technical solution, step 2 comprises:

[0016] Step 2.1, periodically scan the target environment through a two-dimensional laser radar to generate distance data, and convert the distance data into a point set in the robot coordinate system;

[0017] Step 2.2: Based on the point set, the point P detected in the robot coordinate system robot (x r ,y r ) is expressed as:

[0018]

[0019] Where d represents the measured distance and θ represents the measured angle;

[0020] Step 2.3: Point P in the robot coordinate system robot (x r ,y r ) is converted to P in the world coordinate system world (x w ,y w ) is expressed as:

[0021]

[0022] Step 2.4: P in the world coordinate system worl ( d x, w y) w Mapping onto the grid (i, j) in the two-dimensional grid map, we get:

[0023]

[0024] in, Indicates rounding down, w g Indicates the width of the grid, h g Indicates the height of the grid.

[0025] In the above technical solution, step 3 comprises:

[0026] Step 3.1: Determine the grid M i,j Occupancy:

[0027] P occ (M i,j )=f(P occ (M i,j ),Z t ,X t )

[0028] Among them, P occ (M i，j ) represents the grid M i,j The probability of being occupied or idle, f represents the update function, Z t represents the measurement value of the two-dimensional laser radar at time t, X t represents the unknown value of the AGV at time t;

[0029] Step 3.2: Update grid M using Bayesian rule i,j The occupation situation is expressed as:

[0030]

[0031] Among them, P(M i,j |Z t ,X t ) represents the grid Mi,j The posterior probability of being occupied, P(Z t |M i,j ,X t ) represents a given grid M i,j Occupied and AGV position X t When Z t The probability of i,j ) represents the grid M i,j Prior probability of being occupied, P(Z t |X t ) represents the marginal probability of the observed data.

[0032] In the above technical solution, P occ (M i，j )∈[0,1],P occ (M i，j ) is close to 1, indicating a high probability of being occupied, and close to 0, indicating idle or unknown.

[0033] The advantages of the present invention are:

[0034] Through continuous laser scanning, the two-dimensional laser SLAM of the present invention can update map information in real time and perform precise positioning, maintaining high accuracy even in complex or changing environments, reducing the need for human intervention and adjustment due to environmental changes, and reducing the cost and complexity of long-term maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0036] Figure 1 The figure is a flow chart of a method according to an embodiment of the present invention.

[0037] Figure 2 Schematic diagram of the motion model of a single-steering wheel AGV according to an embodiment of the present invention.

[0038] Figure 3 Schematic diagram of the static laser positioning result of the X-axis according to an embodiment of the present invention.

[0039] Figure 4 Schematic diagram of the static laser positioning result of the Y-axis of an embodiment of the present invention.

[0040] Figure 5 A schematic diagram of a two-dimensional grid map of a logistics plant area according to an embodiment of the present invention.

[0041] Figure 6 for Figure 5 A local enlarged schematic diagram of point a in the middle.

[0042] Figure 7 for Figure 5 A partial enlarged schematic diagram of point b in the middle.

[0043] Among them: 1. AGV; 2. Main steering wheel; 3. Auxiliary wheel. DETAILED DESCRIPTION

[0044] Example:

[0045] See also Figure 1 As shown, the present invention provides a high-precision mapping and robust positioning method for a single-steering wheel AGV based on two-dimensional laser SLAM, comprising the following steps:

[0046] Step 1: Obtain the data of the two-dimensional laser radar on the AGV;

[0047] Step 2: construct a structured two-dimensional grid map of the target environment based on the SLAM method;

[0048] Step 3: Obtain real-time and accurate positioning data of the AGV body by judging the occupancy of the grids in the two-dimensional grid map.

[0049] Kinematic analysis is performed in a Cartesian coordinate system, with AGV 1 located on the XOY plane, see Figure 2 As shown in the figure, we define counterclockwise rotation as positive, and the direction of the AGV is the positive direction of the X axis. c The center of the main steering wheel 2. o is the intersection of the AGV center axis and the connecting line of the auxiliary wheel 3, and is defined as the control reference point for AGV path tracking. c and P o The distance between the main steering wheel and the auxiliary wheel center. δ is the steering angle of the main steering wheel relative to the vehicle centerline. c It is the linear speed of the AGV's current steering wheel and points in the forward direction.

[0050] In the real-time control of AGV, O c The AGV instantaneous turning radius R is the center of the rotation circle. c for:

[0051]

[0052] The control principle of the pure tracking algorithm adopted in the present invention is lateral control, which has good robustness. Therefore, the instantaneous speed v in the X-axis direction is obtained. x for:

[0053] v x =v c cosδ

[0054] Therefore, the angular velocity of the AGV at the current position can be obtained for:

[0055]

[0056] The present invention constructs an environmental map through a two-dimensional laser radar, which provides an excellent prerequisite for subsequent AGV positioning, and also provides stable and reliable support for long-term work task operation.

[0057] The present invention sets up a two-dimensional laser radar on the top of the AGV. This installation method can ensure that the two-dimensional laser radar can obtain the surrounding environment information to the greatest extent, effectively avoid many potential obstructions, and obtain more accurate physical space structure detection.

[0058] The present invention uses a two-dimensional grid map with real space physical dimensions as a map reference. The target environment is represented as a two-dimensional grid map M that can be understood by a computer. In this embodiment, step 2 includes:

[0059] Step 2.1, periodically scan the target environment with a two-dimensional laser radar to generate distance data, and convert the distance data into a point set in the robot coordinate system, where each point represents the position of the object detected by the laser radar;

[0060] Step 2.2: Based on the point set, the point P detected in the robot coordinate system robot (x r ,y r ) is expressed as:

[0061]

[0062] Where d represents the measured distance and θ represents the measured angle;

[0063] Step 2.3: The position information (X, Y) and orientation Φ of the AGV in the world coordinate system are obtained from the laser odometer data. robot (x r ,y r ) is converted to P in the world coordinate system world (x w ,y w ) is expressed as:

[0064]

[0065] Step 2.4: P in the world coordinate system worl ( d x, w y) w Mapping onto the grid (i, j) in the two-dimensional grid map, we get:

[0066]

[0067] in, Indicates rounding down, w g Indicates the width of the grid, h g Indicates the height of the grid.

[0068] For each grid M i,j , can be divided into two states, namely, idle and occupied. In this embodiment, step 3 includes:

[0069] Step 3.1: Determine the grid M i,j Occupancy:

[0070] P occ (M i,j )=f(P occ (M i,j ),Z t ,X t )

[0071] Among them, P occ (M i，j ) represents the grid M i,j The probability of being occupied or idle, f represents the update function, Z t represents the measurement value of the two-dimensional laser radar at time t, X t represents the unknown value of the AGV at time t;

[0072] Step 3.2: Update the grid occupancy probability using the new measurement value and sensor position. t Then, through the inverse sensor model P(Z t |M i,j ,X t ) to update the probability P of each grid being occupied on the map occ (M i,j ), update the grid M by Bayesian rule i,j The occupation situation is expressed as:

[0073]

[0074] Among them, P(M i,j |Z t ,X t ) represents the grid M i,j The posterior probability of being occupied, P(Z t |M i,j ,X t ) represents a given grid M i,j Occupied and AGV position X t When Z t The probability of i,j ) represents the grid M i,j Prior probability of being occupied, P(Z t |Xt ) represents the marginal probability of the observed data and is usually used as a normalization constant to ensure that the sum of all probabilities is 1.

[0075] In the above technical solution, P occ (M i，j )∈[0,1],P occ (M i，j ) is close to 1, indicating a high probability of being occupied, and close to 0, indicating idle or unknown.

[0076] The present invention uses two-dimensional laser for high-precision mapping and real-time positioning. All points in each frame of the point cloud in the two-dimensional laser SLAM can participate in the calculation to ensure robust positioning performance. In the case that there is generally no large-scale change in indoor structures, the two-dimensional grid map has extremely high credibility and repeatability, providing an accurate map reference for subsequent trajectory tracking.

[0077] Based on the above technical solution, the present invention has the following beneficial effects:

[0078] 1. Improve positioning accuracy and environmental adaptability in structured environments

[0079] The present invention adopts real-time positioning technology based on two-dimensional laser sensors, which does not rely on preset physical markers, and significantly enhances the positioning accuracy of single-steering wheel AGV in complex structured environments. This technology ensures that the AGV can accurately identify its specific position in the environment in real time, thus having higher environmental adaptability and flexibility, and effectively responding to the challenges of various structured scenarios.

[0080] 2. Enhance environmental mapping and perception capabilities

[0081] The present invention implements a high-precision structured scene mapping method that can adapt to diverse environmental conditions and quickly and accurately construct detailed environmental maps. This not only improves the overall environmental perception capability, but also provides strong support for the efficient autonomous navigation of AGVs, enabling AGVs to position and navigate more flexibly in changing environments.

[0082] 3. Enhance the robustness of AGV control systems and reduce long-term maintenance costs

[0083] By installing a two-dimensional laser device on the top of the AGV to ensure the scanning field of view, without the need for additional auxiliary equipment, the implementation of the present invention improves the robustness of the AGV control system. At the same time, the positioning and mapping system is integrated into the single-steering wheel AGV control system, ensuring that the AGV can maintain stable positioning and navigation performance under changing environmental conditions, significantly reducing the maintenance cost of long-term operation, and providing enterprises with a cost-effective solution.

[0084] The solution of the present invention is verified by experiments below.

[0085] 1. AGV 2D laser static positioning accuracy

[0086] In order to verify the positioning algorithm of the 2D laser radar of the AGV forklift in a stationary state, the present invention selected 2000 sets of positioning results to verify the positioning accuracy of the static laser SLAM. Figure 3 As shown, the positioning result of the Y axis can be found in Figure 4 As shown. The X-axis is the positioning result in millimeters (mm). The Y-axis on the left side of the figure is the count of different positioning result positions, the Y-axis on the right side is the total positioning result sequence number, and the dots represent the positioning results.

[0087] Continue to see Figure 3 As shown, the positioning results of the algorithm of the present invention on the X-axis mostly fall within the interval of (35186, 35192), and only about 0.5% of the results are not within the interval. Through statistical analysis of the histogram and the normal distribution curve, it can be seen that based on the 3σ principle, the static positioning results of the algorithm of the present invention on the X-axis are within the error band of ±3mm.

[0088] Continue to see Figure 4 As shown, the Y-axis positioning results of the algorithm of the present invention mostly fall within the interval of (32388, 32396), and only about 0.95% of the results are not within the interval range. It can be concluded that the positioning results of the algorithm of the present invention in the Y-axis direction are within the error band of ±5mm. In summary, it can be shown that the positioning algorithm accuracy of the laser radar proposed in the present invention can reach ±5mm.

[0089] 2. 2D grid map and AGV running track

[0090] The present invention puts a single steering wheel AGV into the application scenario for mapping to verify the mapping effect. Figure 7 As shown, the two-dimensional grid map obtained by measuring length and width is locally enlarged to ensure the accuracy of the map.

[0091] See also Figure 5 As shown in the figure, a logistics map is 74.84 meters long at its longest point and 26.66 meters wide. Figure 6 As shown in the figure, the mapping positions of the shelves in the local enlargement at a are neat and orderly, and there is no mismatch or mismatch. Figure 7 As shown in the figure, in the reconstructed structured scene, the long side of the factory building is selected as the magnification reference, and the dotted line is used to represent the straight line. Compared with the mapping effect, the accuracy of the two-dimensional laser single-steering wheel AGV mapping can be verified.

[0092] According to the enlarged local image content, the two-dimensional laser mapping method of the present invention for the smart logistics industry can establish accurate grid maps in structured scenes, especially showing excellent performance in scenes with multiple man-made structures.

[0093] Of course, the above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable people familiar with the technology to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any modifications made according to the spirit of the main technical solution of the present invention should be included in the protection scope of the present invention.

Claims

1. A high-precision mapping and robust positioning method for a single-steering wheel AGV based on two-dimensional laser SLAM, characterized by: The following steps are involved: Step 1: Obtain the data of the two-dimensional laser radar on the AGV; Step 2: construct a structured two-dimensional grid map of the target environment based on the SLAM method; Step 3: Obtain real-time and accurate positioning data of the AGV body by determining the occupancy of the grids in the two-dimensional grid map.

2. The high-precision mapping and robust positioning method of a single-steering wheel AGV based on two-dimensional laser SLAM according to claim 1, characterized in that: In the step 1, the two-dimensional laser radar is arranged on the top of the AGV body.

3. The high-precision mapping and robust positioning method of a single-steering wheel AGV based on two-dimensional laser SLAM according to claim 1, characterized in that: The step 2 comprises: Step 2.1, periodically scan the target environment through a two-dimensional laser radar to generate distance data, and convert the distance data into a point set in the robot coordinate system; Step 2.2: Based on the point set, the point P detected in the robot coordinate system robot (x r ,y r ) is expressed as: Where d represents the measured distance and θ represents the measured angle; Step 2.3: Point P in the robot coordinate system robot (x r ,y r ) is converted to P in the world coordinate system world (x w ,y w ) is expressed as: Step 2.4: P in the world coordinate system worl ( d x, w y) w Mapping onto the grid (i, j) in the two-dimensional grid map, we get: in, Indicates rounding down, w g Indicates the width of the grid, h g Indicates the height of the grid.

4. The high-precision mapping and robust positioning method for a single-steering wheel AGV based on two-dimensional laser SLAM according to claim 1, characterized in that: The step 3 comprises: Step 3.1: Determine the grid M i,j Occupancy: P occ (M i,j )=f(P occ (M i,j ),Z t ,X t ) Among them, P occ (M i，j ) represents the grid M i,j The probability of being occupied or idle, f represents the update function, Z t represents the measurement value of the two-dimensional laser radar at time t, X t represents the unknown value of the AGV at time t; Step 3.2: Update grid M by Bayesian rule i,j The occupation situation is expressed as: Among them, P(M i,j |Z t ,X t ) represents the grid M i,j The posterior probability of being occupied, P(Z t |M i,j ,X t ) represents a given grid M i,j Occupied and AGV position X t When Z t The probability of i,j ) represents the grid M i,j Prior probability of being occupied, P(Z t |X t ) represents the marginal probability of the observed data.

5. The high-precision mapping and robust positioning method of a single-steering wheel AGV based on two-dimensional laser SLAM according to claim 4, characterized in that: P occ (M i，j )∈[0,1],P occ (M i，j ) is close to 1, indicating a high probability of being occupied, and close to 0, indicating idle or unknown.

Images