Attribute value interaction LPA* dynamic planning method for disinfection robot

Abstract

The invention discloses an attribute value interaction LPA * dynamic planning method for a disinfection robot, and particularly relates to an A * path planning method which modifies attribute values of grid points through an interactive method and combines dynamic search. The method comprises the following steps: firstly, processing environment data to construct a multi-attribute map; modifying attribute values of the grid points through an interactive method; and finally, considering the influence of multiple attribute values on the cost values of the nodes when searching the adjacent nodes, guiding the algorithm to search the nodes with low attribute values, and dynamically generating a path according to the change of the attribute values of the obstacles and the grid points. According to the method, an LPA * path planning method is improved, a multi-attribute map is combined on the basis of LPA *, compared with a multi-attribute A * path planning method, the attribute values of grid points can be dynamically adjusted, the limitation that the multi-attribute A * path planning method only depends on environment data to generate the attribute values is overcome, and the application range of the A * method is further expanded.

Description

Technical Field

[0001] The present invention relates to the field of robot global path planning, and in particular to an attribute value interaction LPA* dynamic planning method for a disinfection robot. Background Art

[0002] Disinfection robots are a type of mobile robot dedicated to disinfection and sterilization. They are often used in hospitals, public places, industrial environments and other scenarios that require efficient and accurate disinfection. Disinfection robots can move autonomously and achieve full coverage disinfection of large areas through path planning and obstacle avoidance algorithms. Common disinfection methods include spraying disinfectants, ultraviolet disinfection and ozone disinfection. Path planning is one of the important contents of disinfection robot research. The robot needs to calculate the movement route according to the target position and adjust the movement route in real time according to the dynamic obstacles that appear during the movement. These are two core issues in the path planning process. Traditional disinfection robot path planning mainly relies on the A* algorithm or its variants, such as the Dijkstra algorithm. These algorithms perform well in static environments, but in dynamic environments, especially when faced with changing obstacles and path requirements, they often seem to be unable to cope with the situation. In order to meet this challenge, the LPA* (Lifelong Planning A*) method came into being. The LPA* method is an incremental heuristic search algorithm for path planning in dynamic environments. LPA* combines the advantages of the classic A* algorithm and the D*Lite algorithm, and can efficiently update the path when the environment changes. However, the LPA* method does not fully consider the impact of environmental information on the path in path planning, and lacks the ability to define and dynamically adjust the attributes of grid points. This patent is proposed to address the application limitations of the LPA* path planning method in multi-attribute environments. By combining multi-attribute maps and the LPA* dynamic planning method, a flexible and efficient path planning solution is provided, which is suitable for complex and changeable robot navigation environments. Using the attribute value interaction method, the attribute values ​​of the grid points can be adjusted in real time to ensure that the planned path better meets the actual environmental requirements. Summary of the invention

[0003] In view of the deficiencies in the prior art, the present invention proposes an attribute value interaction LPA* dynamic planning method for a disinfection robot, which solves the application limitations of the robot in a multi-attribute environment during the dynamic path planning process.

[0004] According to the present invention, a property value interaction LPA* dynamic programming method for a disinfection robot comprises the following steps:

[0005] Step 1: Obtain attribute values ​​of grid points from a variety of environmental data and construct a multi-attribute map;

[0006] Step 2: After constructing the multi-attribute grid map, define the attributes of the grid points interactively;

[0007] Step 3: Combine the LPA* dynamic programming method and apply the grid point attribute values ​​to the dynamic path planning process.

[0008] Preferably, in step 1: each grid point of the multi-attribute map can store multiple attribute information. For multiple attributes, the weight of each attribute is first determined. The main execution steps are: the first step is to normalize all attributes, the second step is to calculate the weight of each individual attribute map, the third step is to calculate the attribute value of each grid point, and the fourth step is to complete the construction of the multi-attribute map.

[0009] Preferably, in step 2: define grid point attributes interactively, including but not limited to changing, adding, and deleting grid point attribute values ​​by mouse clicks, keyboard input, and real-time updating of obstacle maps and attribute maps. When interacting with grid points in the grid map, perform the following judgment: if the grid point is an obstacle, delete the obstacle; if the grid point is not an obstacle, a prompt box pops up to prompt the user to enter the grid point attribute value that needs to be modified.

[0010] Preferably, in step three: in the process of searching for multiple neighboring nodes, the search is performed in combination with the attribute value cost, including but not limited to 4 neighboring nodes, 8 neighboring nodes, 16 neighboring nodes and other search methods. Mainly, when searching for neighboring nodes, the distance cost from the current node to the next neighboring node and the attribute value cost are used as the total cost, guiding the path planning process to search for the node with the lowest total cost. The present invention takes into account multiple attribute value factors in the path planning process, and uses the distance cost in the traditional search and the attribute value cost proposed by the present invention as the total cost, guiding the algorithm to search for nodes with low attribute values.

[0011] The beneficial effects of the present invention are as follows: the present invention proposes a multi-attribute grid map, and the multi-attribute factors proposed in the present invention can refer to factors such as environmental information, obstacles, and terrain. The attribute map supports the definition of multiple environmental attributes, expanding the limitation that the classic grid map can only store obstacle information. The multi-attribute grid map is applied to the LPA* dynamic programming method. Compared with the classic grid map where the grid points only store 0 and 1 information (such as Figure 1As shown in the figure: (a) is a grid digital map, and (b) is a grid color map. It is divided into a passable area and an obstacle area). The grid points in the multi-attribute map can store floating point numbers between 0 and 1, which are called attribute values ​​(the algorithm plans the optimal path according to the size of the attribute value, avoids generating path nodes at grid points with larger attribute values, and ensures that the total cost of path planning is minimized). Compared with the unimproved LPA* method, the total cost of the path nodes generated by the present invention is lower, the path planning process is more flexible, and can effectively respond to dynamic environmental changes, thereby improving the path planning ability of the disinfection robot in a complex environment. Through the above steps, the present invention can significantly improve the application effect of LPA* path planning in a multi-attribute environment, overcome the limitation that the multi-attribute A* path planning method can only rely on static environmental data to generate attribute values, and expand the scope of application of the A* algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a classic raster map;

[0013] Figure 2 It is the normalization process in the present invention;

[0014] Figure 3 It is the attribute value solving process in the present invention;

[0015] Figure 4 It is the high attribute value screening process in the present invention (t=0.3);

[0016] Figure 5 This is a schematic diagram of a newly added obstacle in the present invention;

[0017] Figure 6 This is a schematic diagram of modifying attribute values ​​in the present invention;

[0018] Figure 7 The search results of the LPA* algorithm before and after the introduction of attribute value cost. DETAILED DESCRIPTION

[0019] In order to further understand the content of the present invention, the present invention is described in detail in conjunction with the accompanying drawings and embodiments. It should be understood that the embodiments are only for explaining the present invention and are not intended to limit it.

[0020] The LPA* dynamic programming method based on attribute value interaction is implemented using the following steps:

[0021] Step 1: Obtain attribute values ​​of grid points from a variety of environmental data and construct a multi-attribute map;

[0022] Step 2: After constructing the multi-attribute grid map, define the attributes of the grid points interactively;

[0023] Step 3: Combine the LPA* dynamic programming method and apply the grid point attribute values ​​to the dynamic path planning process.

[0024] In the step 1, the specific steps include:

[0025] (1) Figure 2 The normalization process in the present invention is shown as follows: for n pieces of environmental information with different attributes, each piece of environmental information is first normalized to unify the dimensions. This paper uses the average normalization method, as shown in the following formula:

[0026]

[0027] Among them, in the i-th environmental data, norm i，j is the normalized value of the jth grid point, x i，j is the environmental data of the jth grid point, μ is the average value of the environmental data, x i , min is the minimum value in the environmental data, x i , max It is the maximum value in the environmental data. The maximum data in the map is 1. The data value is a relative indicator of the current environmental information. The larger the value, the greater the impact of the grid point on the path planning, and the less recommended the path is to pass through the current grid point.

[0028] (2) Calculation of environmental weights: After normalization in step (1), the total values ​​of each normalized environmental information are counted and the environmental weights are calculated using the following formula:

[0029]

[0030] Among them, w i is the weight of the ith environment, w i ∈[0,1] and ∑w i =1; δ is a weighting parameter, which is the proportion of values ​​greater than 0.5 in the environmental information; S i is the total value of the normalized environment information of the ith environment, and n is the number of environments.

[0031] (3) Figure 3 The following is the process of calculating the attribute value of the grid point in the present invention: The weight w of each environmental information is obtained by step (2): i Then, the attribute value of each grid point is obtained by the following formula:

[0032]

[0033] Among them, e i，j is the attribute value of the jth grid point in the i-th environmental information, which can be changed interactively in step 2, and the changed attribute value replaces the original attribute value. i，j ∈[0, 1].

[0034] The larger the attribute value, the less likely the algorithm recommends that the disinfection robot pass through that grid point. Figure 3 Taking the multi-attribute map as an example, the algorithm does not recommend that the robot generate a path at a grid point with an attribute value of 1.00. Instead, it will select grid points with smaller attribute values ​​around the grid point to generate a path. After solving the attribute values ​​of each grid point through the above steps, the multi-attribute map is constructed.

[0035] (4) Figure 4 The figure shows the process of selecting grid points with high attribute values ​​in the present invention: In actual application scenarios, different disinfection robots have different adaptability to the environment. In order to study the path planning ability of disinfection robots for grid points with high attribute values ​​in the environment, the present invention introduces the environmental threshold parameter t to select grid points with high attribute values. The weight of each environmental information is determined by the screening result, so that the construction of the multi-attribute map is more targeted. The processing process is as follows: if norm i If j is greater than t, the original value of the attribute is retained; if norm i If j is less than t, the attribute value is set to 0. The environmental threshold t is calculated by the following formula:

[0036]

[0037] Among them, n represents the number of environments, and the coefficient α i represents the influence coefficient of each environment on path planning, and t i Represents the recommended threshold of the robot in the i-th environment.

[0038] In the step 2, the specific steps include:

[0039] When a grid point is selected interactively, such as when a user selects a grid point in a grid map, the attribute editing function is triggered. At this time, the grid point is judged in the attribute editing function: if the current grid point is an obstacle, the obstacle is deleted; if the current grid point is not an obstacle, a prompt box pops up to prompt the user to enter the changed attribute value of the current grid point. The changed attribute value range is between 0 and 1. The changed grid point attribute value is judged. If the changed attribute value is 1, the current grid point is added to the obstacle list. Finally, the multi-attribute map and obstacle map information are updated in real time to replan the path.

[0040] Figure 5The figure shows the process of manually setting the attribute value of the grid point (20, 28) in the map to 1, updating the grid point to the obstacle grid and replanning the path. (a) is the original generated path, (b) is the process of changing the attribute value of the grid point, and (c) is the path replanned after the attribute value of the grid point is changed, the obstacle map and the attribute value map are updated. In the figure, the black grid point is the obstacle, the blue grid point is the starting point, the yellow grid point is the end point, the white, green, and red grid points are the grid points with attribute values ​​between 0-0.33, 0.34-0.66, and 0.67-1 (represented as excellent environment points, good environment points, and bad environment points), and the blue path is the algorithm-generated path.

[0041] Figure 6 The figure shows the process of setting the grid point with bad environment in the map to the grid point with good environment. After selecting the grid point (8, 15) with bad environment in (a), the prompt box shown in (b) pops up. Enter the attribute value you want to change in the prompt box. The attribute value map and obstacle map are updated as shown in (c). In the figure, the grid point (8, 15) changes from red to white, indicating that the attribute value has been successfully modified.

[0042] In the step three, the specific steps include:

[0043] When the LPA* dynamic programming method expands the node, it will n The node with the lowest estimated cost among the 4-16 neighboring nodes is added to the priority queue. n Expand to neighboring node S n+1 The cost is d. After considering the impact of attribute values ​​on path planning, the attribute value cost Cost is introduced. env .d and Cost env Calculated by the following formula:

[0044]

[0045] Cost env =d*e

[0046] Where x1 and y1 are the current nodes S n The coordinates of x2 and y2 are the neighboring nodes S of the current node. n+1 The coordinates of the node, e is the neighboring node S of the current node n+1 The attribute value of can be modified through interactive methods in step 2.

[0047] When searching for neighboring nodes, the distance cost d from the current node to the next neighboring node and the attribute value cost Cost env As the total cost, guide the path planning process to search for nodes with low attribute values. The total cost Cost is calculated by the following formula:

[0048] Cost=d+Cost env

[0049] In the process of node expansion, nodes with the smallest total cost are preferentially selected for expansion, so as to effectively guide the search process to bypass grid points with high attribute values, ensuring that the path planning process is more in line with the actual environment requirements.

[0050] Figure 7 (a) shows the search results of the LPA* algorithm without introducing attribute value costs. Figure 7 In (b), the path generated after the attribute value cost is introduced passes through more points with good environments and points with bad environments, has a higher total attribute value, and is more optimized.

Claims

1. A dynamic programming method of attribute value interaction LPA* for a disinfection robot, characterized in that: Includes steps: Step 1: Obtain attribute values ​​of grid points from a variety of environmental data and construct a multi-attribute map; Step 2: After constructing the multi-attribute grid map, define the attributes of the grid points interactively; Step 3: Combine the LPA* dynamic programming method and apply the grid point attribute values ​​to the dynamic path planning process.

2. The attribute value interaction LPA* dynamic programming method for the disinfection robot according to claim 1, characterized in that: In step 1, each grid point of the multi-attribute map can store multiple attribute information. For multiple attributes, first determine the weight of each attribute. The main execution steps are: the first step is to normalize all attributes, the second step is to calculate the weight of each individual attribute map, the third step is to calculate the attribute value of each grid point, and the fourth step is to complete the construction of the multi-attribute map.

3. The attribute value interaction LPA* dynamic programming method for the disinfection robot according to claim 1 is characterized in that: In step 2, the grid point attributes are defined interactively, including but not limited to changing, adding, and deleting the grid point attribute values ​​through mouse clicks, keyboard input, and other interactive methods, as well as real-time updates of the obstacle map and attribute map. When interacting with the grid points in the grid map, the following judgment is performed: if the grid point is an obstacle, delete the obstacle; if the grid point is not an obstacle, a prompt box pops up to prompt the input of the grid point attribute value to be modified.

4. The attribute value interaction LPA* dynamic programming method for the disinfection robot according to claim 1, characterized in that: In step three, in the process of searching multiple neighboring nodes, the search is performed in combination with the attribute value cost, including but not limited to 4 neighboring nodes, 8 neighboring nodes, 16 neighboring nodes and other search methods. Mainly, when searching for neighboring nodes, the distance cost from the current node to the next neighboring node and the attribute value cost are used as the total cost, guiding the path planning process to search for the node with the lowest total cost. The present invention takes into account multiple attribute value factors in the path planning process, and uses the distance cost in the traditional search and the attribute value cost proposed by the present invention as the total cost, guiding the algorithm to search for nodes with low attribute values.

Images