Intelligent robot narrow road passing path planning method

Abstract

The invention provides an intelligent robot narrow channel passing path planning method, which comprises the following steps of: acquiring a minimum rectangular frame of a chassis of an intelligent robot, and generating a square physical model for simulating the intelligent robot in a hierarchical cost map by taking a short side of the minimum rectangular frame as a side length; establishing a virtual layer, and generating a virtual wall on boundary straight lines on two sides of a narrow channel in the virtual layer; integrating the virtual layer into a hierarchical cost map, and increasing the barrier expansion radius of the narrow lane based on the square physical model to update the hierarchical cost map until the cost values of the narrow lane and the internal grids of the virtual wall are both greater than zero; and planning a passing path of the intelligent robot on the hierarchical cost map on which the virtual layer is overlaid. According to the method, the square physical model of the intelligent robot is simulated through reverse thinking, the virtual wall is established, the barrier expansion radius of the narrow road is increased on the basis of the square physical model, the layered cost map is updated, and the narrow road traffic capacity is improved.

Description

Technical Field

[0001] The present application belongs to the field of intelligent robot navigation technology, and more specifically, to a method for planning a narrow lane passage path for an intelligent robot. Background Art

[0002] Intelligent robots have the ability to perceive, make decisions and execute. They mainly involve key technologies such as multi-sensor information fusion, navigation and positioning, path planning, robot vision, intelligent control, and human-machine interface technology. They can replace humans in dangerous and complex labor in unstructured environments. For example, cleaning robots, firefighting robots, patrol robots, and transport robots.

[0003] The current intelligent robot navigation generally uses path planning algorithms and layered cost maps for global and local path planning. However, since its positioning accuracy is around 5 cm and its navigation accuracy is also around 5 cm, collisions are prone to occur when passing through narrow roads, especially irregular narrow roads, and it is even impossible to plan a path. Summary of the invention

[0004] The purpose of the embodiments of the present application is to provide a method for planning a narrow road passage path for an intelligent robot, so as to solve the technical problem of insufficient ability of the intelligent robot in the narrow road passage process in the prior art.

[0005] To achieve the above purpose, the technical solution adopted in this application is: to provide a method for planning a narrow passage path for an intelligent robot, comprising the steps of: Obtain the minimum rectangular frame of the chassis of the intelligent robot, and use the short side of the minimum rectangular frame as the side length to generate a square physical model of the simulated intelligent robot in the layered cost map; A virtual layer is created, and virtual walls are generated on the boundary straight lines on both sides of the narrow road in the virtual layer; Integrate the virtual layer into the layered cost map, and based on the square physical model, increase the obstacle expansion radius of the narrow road to update the layered cost map until the cost values ​​of the narrow road and the internal grid of the virtual wall are greater than zero; The layered cost map after superimposing the virtual layer plans the path of the intelligent robot.

[0006] In one embodiment of the present application, before establishing the virtual layer, a narrow road detection step is also included: In the layered cost map, the distance W between any two points on both sides of the obstacle is measured local ; If there are any two points with a width W local Satisfy R w <W local <R l , then these two points are identified as narrow points, where R lis the long side of the minimum rectangular box, R w is the short side of the minimum rectangular box; Perform curve fitting on the narrow points connected on both sides to obtain the boundary of the narrow road; The narrow lane area and the non-narrow lane area are defined according to the boundary of the narrow lane.

[0007] In one embodiment of the present application, a method for defining a narrow lane area and a non-narrow lane area according to a boundary of a narrow lane comprises the steps of: Get the starting point coordinates and the ending point coordinates of each narrow road boundary; Connecting the starting point coordinates of a pair of corresponding narrow lane boundaries to obtain a narrow lane entrance line, and connecting the ending point coordinates of a pair of corresponding narrow lane boundaries to obtain a narrow lane exit line; Connecting the boundaries of a pair of corresponding narrow lanes and their corresponding narrow lane entrance lines and narrow lane exit lines to obtain a narrow lane area; The area other than the narrow channel area in the irregular channel is set as the non-narrow channel area.

[0008] In one embodiment of the present application, a method for generating a virtual wall from boundary lines on both sides of a narrow road in a virtual layer includes the steps of: Based on the detected exit and entrance of each narrow passage, each passage is divided into two straight virtual walls; The virtual wall is parallel to or tangent to the boundary of the corresponding narrow road.

[0009] In one embodiment of the present application, the thickness Q of the virtual wall is w Satisfaction: Q w ≤W local -R w ; The length Q of the virtual wall width l Satisfaction: Q l ≥R l .

[0010] In one embodiment of the present application, a method for planning a path for an intelligent robot on a layered cost map after superimposing a virtual layer includes the steps of: Set up temporary target points in each non-narrow area of ​​the irregular channel; Generate a path from the starting point to the nearest temporary target point using a global path planning algorithm; After reaching the temporary target point, the global path planning algorithm is used to generate a path to the next temporary target point until the destination is reached.

[0011] In one embodiment of the present application, the temporary target point is located within a virtual wall of a narrow passage entrance or exit.

[0012] In one embodiment of the present application, the global path planning algorithm is the Theta star algorithm.

[0013] In one embodiment of the present application, a method for determining whether an intelligent robot has reached a temporary target point within a virtual wall comprises the steps of: Create virtual wall polygons; If each vertex of the minimum rectangular box is within the virtual wall polygon, it is determined that the intelligent robot has reached the temporary target point within the virtual wall.

[0014] In one embodiment of the present application, a method for creating a virtual wall polygon comprises the steps of: Assume that the starting point of the virtual wall of the first narrow road is A (x1, y1) and the end point is B (x2, y2), and the starting point of the virtual wall of the second narrow road is C (x3, y3) and the end point is D (x4, y4); If d1 is greater than d2, connect ABCD in sequence, otherwise connect ABDC in sequence to obtain a virtual wall polygon; .

[0015] The beneficial effect of the intelligent robot narrow road passage path planning method provided by the present application is that: compared with the existing technology, the square physical model of the intelligent robot is simulated by reverse thinking, a virtual wall is established and the obstacle expansion radius of the narrow road is increased based on the square physical model to update the layered cost map, and each grid inside the narrow road is fully assigned a cost value. It comprehensively overcomes technical bias from four aspects and improves the narrow road passage capacity. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, 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 application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0017] Figure 1 A schematic diagram of a hierarchical structure of a hierarchical cost map in the prior art; Figure 2 A schematic diagram of an intelligent robot in the prior art when planning a path; Figure 3 A flow chart of a method for planning a narrow passage path for an intelligent robot provided in an embodiment of the present application; Figure 4 A schematic diagram showing a comparison of methods for generating a minimum rectangular frame and a square physical model based on an intelligent robot; Figure 5 It is a schematic diagram of the inscribed circle and circumscribed circle of the minimum rectangular frame and square physical model; Figure 6A schematic diagram of an intelligent robot provided in an embodiment of the present application when planning a path in a regular channel; Figure 7 A schematic diagram of an intelligent robot provided in an embodiment of the present application when planning a path in an irregular channel. DETAILED DESCRIPTION

[0018] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0019] It should be noted that when an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or indirectly connected to the other element.

[0020] It should be understood that the orientation or position relationship indicated by terms such as "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside" and "outside" are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0021] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0022] First of all, it must be explained that layered cost maps are an existing mature technology for representing environmental information. Layered cost maps refer to dividing the data in the cost map into several isolated map layers according to different semantics. Each layer of the map only tracks one type of obstacle or constraint, and then modifies the main cost map used for path planning. At the same time, this layered cost map can represent complex situations so that navigation can be performed according to some different situations.

[0023] See also Figure 1 The layered cost map consists of three layers. The first layer is the Static Layer. Figure 1It is usually generated in advance using the SLAM (Simultaneous Localization and Mapping) algorithm to represent the location of walls and other static obstacles. The second layer is the Obstacle Layer, which collects data from high-precision sensors (such as lasers and RGB-D cameras), and the locations of sensor readings are marked as occupied. The third layer is the Inflation Layer, which inserts a buffer around each fatal obstacle to ensure that the intelligent robot does not collide with the obstacle. These three layers are combined into the main map (the final cost map) for route planning.

[0024] See also Figure 2 The layered cost map is in the form of a grid map. The cost value of each grid is 0-255. Each grid can be divided into the following five types according to the cost value: (1) Fatal conflict: The grid value is 254. There is an obstacle in the grid, and the geometric center of the intelligent robot is not allowed to exist in the grid. (2) Inscribed conflict: The grid cost value is 253. When the geometric center of the intelligent robot is located in this grid, there will be obstacles within the inscribed circle contour of the intelligent robot, and a conflict will inevitably occur. (3) Risk: The grid cost value is greater than 0 and less than 253. When the geometric center of the intelligent robot is located in the grid, there will be obstacles between the circumscribed circle and the inscribed circle contour of the intelligent robot, which poses a risk of conflict. (4) Free area: The grid value is 0, there is no obstacle. When the geometric center of the intelligent robot is located in this grid, all obstacles are outside the circumscribed circle of the intelligent robot and will not affect the intelligent robot. (5) Unknown area: The grid value is 255, which is outside the map boundary and cannot be reached by general intelligent robots.

[0025] In addition, based on the above-mentioned layered cost map, a variety of planning path algorithms can be run for navigation, such as Dijkstra (Dijkstra algorithm), A* (A star algorithm), D* (A star algorithm), Theta* (Theta star algorithm). Therefore, the layered cost map and any of the above-mentioned planning path algorithms constitute the core technology of intelligent robot navigation (hereinafter referred to as the existing navigation core technology).

[0026] Since the chassis contours are of different shapes, in order to adapt to the above core technologies, people in this field will simulate the chassis of the intelligent robot into a circle, square or rectangle to simplify the algorithm. However, no matter what shape is simulated, the area of ​​the chassis will be larger than the actual area (see Figure 4To the left algorithm route), in order to further reduce the probability of collision accidents, the actual chassis shape will be enlarged by a certain multiple. Technical personnel generally believe that the simulated chassis shape area must at least cover the actual chassis area, otherwise the risk of collision will inevitably increase, which will further limit the narrow road passing ability of the intelligent robot.

[0027] However, the present application is an improvement based on the layered cost map and any of the above-mentioned path planning algorithms, and uses reverse thinking to create a method for improving the narrow road passing ability of an intelligent robot.

[0028] Please also read Figures 3 to 6 , the narrow lane path planning method for an intelligent robot provided in an embodiment of the present application is now described. The narrow lane path planning method for an intelligent robot is applicable to an intelligent robot with a chassis of a rectangular, trapezoidal, triangular, rhombus, parallelogram, and irregular shapes (excluding squares and circles).

[0029] The method for planning a narrow passage path for an intelligent robot comprises the following steps: Step S1, obtain the minimum rectangular frame of the chassis of the intelligent robot, and use the short side of the minimum rectangular frame as the side length to generate a square physical model of the simulated intelligent robot in the layered cost map.

[0030] It is understandable that the chassis mentioned in the field of automatic navigation of intelligent robots generally refers to the area occupied by the intelligent robot from a bird's-eye view, which includes the fuselage chassis and surrounding protruding structures, such as outer casings, cameras, lidars, etc.

[0031] In step S1, the smallest rectangular frame of the chassis of the intelligent robot is first obtained. In the prior art, the smallest rectangular frame is directly or proportionally enlarged as the physical model of the intelligent robot. However, in this embodiment, the short side of the smallest rectangular frame is used as the side length to generate a square physical model of the simulated intelligent robot (see Figure 4 To the right algorithm route), therefore, its essence is to reduce the chassis area of ​​the intelligent robot. According to the classification introduction of the cost values ​​of the above five grids, the inscribed circle contour and the circumscribed circle contour of the square physical model after the processing of step S1 will become smaller than the minimum rectangular frame. Therefore, the inscribed conflicts and risky grids in the narrow road will be reduced, and the number of grids in the free area will increase significantly.

[0032] In this way, the direct effect of step S1 is to increase the passable area in the narrow road. However, since the square physical model does not actually completely cover the chassis of the intelligent robot, the intelligent robot will freely choose a path in the free area when planning the path. The direct negative impact of this result is that the risk of collision with obstacles when passing through a narrow road based on the existing planning path algorithm increases.

[0033] Step S2, creating a virtual layer, and generating virtual walls on the boundary straight lines on both sides of the narrow road in the virtual layer.

[0034] In step S2, the virtual layer is established based on the basic framework of the layered cost map, which is conducive to integrating the virtual layer into the layered cost map. After the virtual wall is generated on the boundary straight lines on both sides of the narrow road in the virtual layer, it is equivalent to extending the length of the narrow road, that is, there will be a large number of free area grids between the virtual wall and the narrow road.

[0035] Step S3, integrating the virtual layer into the layered cost map, and increasing the obstacle expansion radius of the narrow road based on the square physical model to update the layered cost map until the cost values ​​of the narrow road and the internal grid of the virtual wall are both greater than zero.

[0036] In step S3, this embodiment also increases the obstacle expansion radius of the narrow passage until the cost values ​​of the grids inside the narrow passage are all greater than zero, that is, the grids in the free area are cleared, so that each grid inside the narrow passage has a certain cost value.

[0037] Specifically, on the first hand, before entering the narrow passage, the intelligent robot can use the virtual wall as a buffer to effectively avoid obstacles at the entrance of the narrow passage when adjusting its direction; on the first hand, because the virtual wall is generated by the boundary straight lines on both sides of the narrow passage, and there is no grid of free area in the virtual wall, the use of the virtual wall allows the intelligent robot to adjust to the optimal posture in advance, and the optimal posture is that the direction of the intelligent robot is facing the entrance of the narrow passage, and its position is located on the central axis of the entrance of the narrow passage; on the third hand, when the intelligent robot enters the narrow passage and uses the existing path planning algorithm to navigate, it needs to consider the cost value at each step, thereby limiting its freedom of path selection and forcing the path to be planned in the center; on the fourth hand, because the obstacle radius of the narrow passage is expanded, the actual obstacle is smaller than the obstacle when planning the path, so the robustness just makes up for the aforementioned direct negative impact.

[0038] Step S4, planning the path of the intelligent robot on the layered cost map after superimposing the virtual layer.

[0039] Thus, in step S4, based on the four effects of step S3, conventional narrow roads (such as gates) and irregular narrow roads (such as S-shaped narrow roads) can be effectively passed. Most notably, the method of this embodiment can pass ultra-narrow terrains where the width of the narrow road is less than the long side of the smallest rectangular frame of the chassis and greater than the short side, while in the existing path navigation algorithm, such ultra-narrow terrain will be directly regarded as a closed channel.

[0040] Compared with the prior art, the intelligent robot narrow road path planning method provided in the present application simulates the square physical model of the intelligent robot with reverse thinking, establishes a virtual wall and increases the obstacle expansion radius of the narrow road based on the square physical model to update the layered cost map, fully assigns a cost value to each grid inside the narrow road, and comprehensively overcomes technical bias from four aspects to improve the narrow road passability.

[0041] In one embodiment of the present application, see Figure 7 In step S2, before establishing the virtual layer, a narrow road detection step is also included: In the layered cost map, the distance W between any two points on both sides of the obstacle is measured local ; If there are any two points with a width W local Satisfy R w <W local <R l , then these two points are identified as narrow points, where R l is the long side of the minimum rectangular box, R w is the short side of the minimum rectangular box; Perform curve fitting on the narrow points connected on both sides to obtain the boundary of the narrow road; The narrow lane area and the non-narrow lane area are defined according to the boundary of the narrow lane.

[0042] It is understandable that when there is W local <R w When W is lower than the limit that the intelligent robot can pass through, it is usually considered a closed channel; local Satisfy R w <W local <R l When W is triggered, the narrow lane traffic mode is triggered, that is, steps S2-S4 are executed; local >R l When using the system, it is based on the existing core navigation technology.

[0043] Different from the conventional narrow road detection definition method, in this embodiment, the distance W between any two points is detected. local The narrow point is identified as the narrow point, and then the narrow point is fitted with a curve to obtain the boundary of the narrow road. In this way, the irregular channel in the real scene is divided into several narrow roads in the form of segments according to its curvature and width. Then, the virtual wall is generated in the continuous irregular channel according to the shape of the narrow road, and the non-narrow road area (W local >R l ) as a buffer for adjusting to the optimal posture in advance.

[0044] That is, this embodiment can segment the irregular narrow passage in the conventional sense into the narrow passage defined in this embodiment, and independently execute the above-mentioned steps S2-S5 for each segmented narrow passage. In this way, the path of the intelligent robot is finely guided and corrected through multiple virtual walls, and each virtual wall of the segment is parallel to the boundary of the segmented narrow passage, ensuring that the intelligent robot can adapt to the curvature and inclination of the narrow passage, so as to improve the passability of the irregular narrow passage.

[0045] For further information, see Figure 7 In step S2, the method of defining the narrow road area and the non-narrow road area according to the boundary of the narrow road includes the steps of: Get the starting point coordinates and the ending point coordinates of each narrow road boundary; Connecting the starting point coordinates of a pair of corresponding narrow lane boundaries to obtain a narrow lane entrance line, and connecting the ending point coordinates of a pair of corresponding narrow lane boundaries to obtain a narrow lane exit line; Connecting the boundaries of a pair of corresponding narrow lanes and their corresponding narrow lane entrance lines and narrow lane exit lines to obtain a narrow lane area; The area other than the narrow channel area in the irregular channel is set as the non-narrow channel area.

[0046] It can be understood that, according to the boundary definition of the narrow road in this embodiment, it can be found that the non-narrow road area is located on both sides of the narrow road area, so that the virtual wall is located in the non-narrow road area. According to the above-mentioned narrow road detection steps, it can be seen that the space of the non-narrow road area is larger than the narrow road area, that is, the virtual wall is limited to be generated only in the area with sufficient space, thereby reducing the probability of collision of the intelligent robot when adjusting the posture, which is conducive to the adjustment of the posture of the intelligent robot.

[0047] For further information, see Figure 7 In step S2, the method for generating virtual walls from the boundary straight lines on both sides of the narrow road in the virtual layer includes the following steps: Based on the detected exit and entrance of each narrow passage, each passage is divided into two straight virtual walls; The virtual wall is parallel to or tangent to the boundary of the corresponding narrow road.

[0048] It can be understood that when the exit or entrance of the narrow passage is a straight line, the virtual wall is parallel to the corresponding narrow passage boundary, and when the exit or entrance of the narrow passage boundary is an arc, the virtual wall is tangent to the corresponding narrow passage boundary. In this way, the virtual wall and the narrow passage achieve a natural transition, optimizing the posture of the intelligent robot before entering the narrow passage.

[0049] Furthermore, in step S2, the thickness Q of the virtual wall w Satisfaction: Q w ≤W local -R w; The length Q of the virtual wall width l Satisfaction: Q l ≥R l .

[0050] It can be understood that when the thickness Q of the virtual wall w Satisfy Q w =W local -R w When the length of the virtual wall is Q l Satisfy Q l =R l When the robot is in a narrow and continuous irregular lane, the virtual wall can dynamically adjust the shape of the virtual wall according to the actual space size of each narrow lane, and then adjust the virtual wall as close to the optimal posture as possible in the limited space, and further refine the path of the robot to improve the navigation capacity of the narrow lane.

[0051] For further information, see Figure 7 In step S4, the method for planning the path of the intelligent robot on the layered cost map after superimposing the virtual layer includes the steps of: Set up temporary target points in each non-narrow area of ​​the irregular channel; Generate a path from the starting point to the nearest temporary target point using a global path planning algorithm; After reaching the temporary target point, the global path planning algorithm is used to generate a path to the next temporary target point until the destination is reached.

[0052] It is understandable that, on the one hand, the environment in the irregular channel is complex, and may also contain obstacles that change dynamically in real time and bifurcated sections. Therefore, by setting temporary target points in each non-narrow road area in the irregular channel, the path can be generated step by step to fully adapt to the above complex environment.

[0053] In addition, since this application is based on the existing navigation core technology, based on the narrow road width defined in this application, if the destination position is directly navigated, the navigation core technology may display an inaccessible or detour path. Therefore, the solution of this embodiment can also solve the above-mentioned incompatibility problem.

[0054] For further information, see Figure 7 , in step S4, the temporary target point is located within the virtual wall of the narrow passage entrance and exit.

[0055] It can be understood that, since the virtual wall is located in the non-narrow road area, when the temporary target point is located within the virtual wall of the narrow road entrance and exit, in the process of segmented path planning, the intelligent robot can be guided to adjust its posture before entering the narrow road to ensure that it is facing the narrow road entrance and walks out of the narrow road exit along the curvature of the narrow road to reduce collisions; at the same time, by landing in advance, it is helpful to reduce the complexity of re-planning the path.

[0056] Furthermore, in step S4, the global path planning algorithm is at least one of Dijkstra algorithm, A* algorithm, and D* algorithm. Optimally, the global path planning algorithm is Theta* algorithm.

[0057] It can be understood that since the present embodiment is based on a square physical model, the square physical model does not actually completely cover the chassis of the intelligent robot. There is a risk of the aforementioned direct negative impact during narrow road turning, such as tail-swinging collision or head-swinging collision. The Theta star algorithm is characterized in that it is suitable for generating smooth paths in complex environments. Therefore, the amplitude of tail-swinging and head-swinging can be reduced during movement along a relatively smooth path, thereby reducing collisions.

[0058] Since the Theta star algorithm needs to perform line-of-sight detection (LOS) every time a node is expanded to determine whether there is an obstacle-free straight path between the current node and the parent node, this process will increase the computational overhead, especially in complex environments (such as narrow roads), the frequency of line-of-sight detection is high, resulting in a longer running time than other algorithms; in addition, the Theta star algorithm generates a smooth path by skipping intermediate nodes through line-of-sight detection, so in complex environments (such as narrow roads), it is also necessary to use a local path control algorithm for local path planning and dynamic obstacle avoidance, such as the TEB algorithm (time elastic band algorithm) or the DWA algorithm (dynamic window method), which further increases the computational overhead and the running time. However, in this embodiment, since it is divided into several narrow roads in the form of segments, and then a virtual wall is generated in segments according to the shape of the narrow road inside the continuous irregular channel, the non-narrow road area is used as a buffer, and the virtual wall performs local detail posture adjustment, which can replace the local detail control of the local path control algorithm. In turn, it can reduce the computational overhead and improve the running speed.

[0059] It can be seen that the Theta star algorithm and the segmented narrow road and virtual wall solution in the embodiment of the present application complement each other and can reduce collisions and improve running speed compared to the prior art.

[0060] Further, after step S4, the method for determining whether the intelligent robot has reached a temporary target point in the virtual wall comprises the steps of: Create virtual wall polygons; If each vertex of the minimum rectangular box is within the virtual wall polygon, it is determined that the intelligent robot has reached the temporary target point within the virtual wall.

[0061] It can be understood that the virtual wall polygon is an area surrounded by a pair of virtual walls.

[0062] Further, after step S4, the method for creating a virtual wall polygon comprises the steps of: Assume that the starting point of the virtual wall of the first narrow road is A (x1, y1) and the end point is B (x2, y2), and the starting point of the virtual wall of the second narrow road is C (x3, y3) and the end point is D (x4, y4); If d1 is greater than d2, connect ABCD in sequence, otherwise connect ABDC in sequence to obtain a virtual wall polygon; .

[0063] It is understandable that the boundaries of the irregular narrow road are not parallel, and their lengths may be inconsistent due to terrain restrictions. Based on the above formula, d1 is equivalent to the distance from A to C plus the distance from B to D, and d2 is equivalent to the distance from A to D plus the distance from B to C. If d1 is greater than d2, connect ABCD in sequence; otherwise, connect ABDC in sequence. The virtual wall polygon obtained in this way can maximize the area and improve robustness.

[0064] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for planning a narrow passage path for an intelligent robot, characterized in that: Includes steps: Obtain the minimum rectangular frame of the chassis of the intelligent robot, and use the short side of the minimum rectangular frame as the side length to generate a square physical model of the simulated intelligent robot in the layered cost map; A virtual layer is created, and virtual walls are generated on the boundary straight lines on both sides of the narrow road in the virtual layer; Integrate the virtual layer into the layered cost map, and based on the square physical model, increase the obstacle expansion radius of the narrow road to update the layered cost map until the cost values ​​of the narrow road and the internal grid of the virtual wall are greater than zero; The layered cost map after superimposing the virtual layer plans the path of the intelligent robot.

2. The method for planning a narrow passage path for an intelligent robot according to claim 1, characterized in that: Before creating the virtual layer, the narrow road detection step is also included: In the layered cost map, the distance W between any two points on both sides of the obstacle is measured local ; If there are any two points with a width W local Satisfy R w <W local <R l , then these two points are identified as narrow points, where R l is the long side of the minimum rectangular box, R w is the short side of the minimum rectangular box; Perform curve fitting on the narrow points connected on both sides to obtain the boundary of the narrow road; The narrow lane area and the non-narrow lane area are defined according to the boundary of the narrow lane.

3. The method for planning a narrow passage path for an intelligent robot according to claim 2, characterized in that: The method for defining a narrow lane area and a non-narrow lane area according to the boundary of a narrow lane comprises the steps of: Get the starting point coordinates and the ending point coordinates of each narrow road boundary; Connecting the starting point coordinates of a pair of corresponding narrow lane boundaries to obtain a narrow lane entrance line, and connecting the ending point coordinates of a pair of corresponding narrow lane boundaries to obtain a narrow lane exit line; Connecting the boundaries of a pair of corresponding narrow lanes and their corresponding narrow lane entrance lines and narrow lane exit lines to obtain a narrow lane area; The area other than the narrow channel area in the irregular channel is set as the non-narrow channel area.

4. The method for planning a narrow passage path for an intelligent robot according to claim 3, characterized in that: The method for generating virtual walls from boundary lines on both sides of a narrow road in a virtual layer comprises the following steps: Based on the detected exit and entrance of each narrow passage, each passage is divided into two straight virtual walls; The virtual wall is parallel to or tangent to the boundary of the corresponding narrow road.

5. The method for planning a narrow passage path for an intelligent robot according to claim 4, characterized in that: The thickness Q of the virtual wall w Satisfaction: Q w ≤W local -R w ; The length Q of the virtual wall width l Satisfaction: Q l ≥R l .

6. The method for planning a narrow passage path for an intelligent robot according to claim 5, characterized in that: A method for planning a path for an intelligent robot based on a layered cost map after superimposing a virtual layer, comprising the steps of: Set up temporary target points in each non-narrow area of ​​the irregular channel; Generate a path from the starting point to the nearest temporary target point using a global path planning algorithm; After reaching the temporary target point, the global path planning algorithm is used to generate a path to the next temporary target point until the destination is reached.

7. The method for planning a narrow passage path for an intelligent robot according to claim 6, characterized in that: The temporary target point is located within the virtual wall of the narrow passage entrance and exit.

8. The method for planning a narrow passage path for an intelligent robot according to claim 7, characterized in that: The global path planning algorithm is the Theta star algorithm.

9. The method for planning a narrow passage path for an intelligent robot according to claim 7, characterized in that: The method for determining that the intelligent robot has reached a temporary target point within a virtual wall comprises the following steps: Create virtual wall polygons; If each vertex of the minimum rectangular box is within the virtual wall polygon, it is determined that the intelligent robot has reached the temporary target point within the virtual wall.

10. The method for planning a narrow passage path for an intelligent robot according to claim 9, characterized in that: The method for creating a virtual wall polygon comprises the steps of: Assume that the starting point of the virtual wall of the first narrow road is A (x1, y1) and the end point is B (x2, y2), and the starting point of the virtual wall of the second narrow road is C (x3, y3) and the end point is D (x4, y4); If d1 is greater than d2, connect ABCD in sequence, otherwise connect ABDC in sequence to obtain a virtual wall polygon; where d1 and d2 satisfy: 。

Images