Blind obstacle avoidance path planning method

Abstract

The application discloses a blind person obstacle avoidance path planning method, which comprises the following steps: firstly, a three-dimensional information grid map is established; secondly, a starting point and an ending point are determined, and a heuristic function is established; thirdly, a crossing cost and a turning cost are introduced into the heuristic function; fourthly, a gravitational potential field and a repulsive potential field are established; fifthly, the heuristic function is weighted by using the gravitational potential field and the repulsive potential field; and finally, the path planning is performed by using the heuristic function to obtain a blind person obstacle avoidance path. The blind person path planning method disclosed by the application can effectively overcome the defects that the existing planning algorithm cannot solve the crossing problem, effectively reduce the crossing phenomenon and the turning phenomenon, and maximally avoid the problem that the planned path is too close to the obstacle, and fully meet the obstacle avoidance demand of the blind person.

Description

Technical Field

[0001] This invention relates to the field of guide technology, and more specifically to a method for planning obstacle avoidance paths for the blind. Background Technology

[0002] In today's era of rapid technological advancement, further attention should be paid to how to use intelligent systems and algorithms to meet the needs of blind people during their travels. The most crucial aspect of blind people's travel is obstacle avoidance; however, current obstacle avoidance technologies are mostly applied to drones and autonomous vehicles, with limited research on obstacle avoidance for the blind. Drones and autonomous vehicles need to plan paths as smoothly as possible, using drive control to change attitude in real time. For blind people, however, smoother paths mean more obstacle avoidance information, and real-time attitude changes are complex and unsuitable for them.

[0003] In obstacle avoidance path planning for the blind, ensuring their safety is paramount. Due to the lack of visual information, guidance should avoid narrow areas and minimize turns, while ensuring smooth path planning even when encountering steps. Although the A-Star algorithm is a highly efficient direct search method for finding the shortest path in static road networks and is currently a well-performing obstacle avoidance planning algorithm, its output is not suitable for blind pedestrians because it does not take into account their specific circumstances.

[0004] Therefore, how to design an obstacle avoidance path planning method specifically for blind people is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] In view of this, the present invention addresses the shortcomings of existing obstacle avoidance algorithms that cannot meet the obstacle avoidance requirements of blind people, and provides a path planning method for obstacle avoidance for blind people. This method comprehensively considers the problems of crossing steps, turning, and narrow areas, and meets the obstacle avoidance needs of blind people to the greatest extent.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A method for obstacle avoidance path planning for the blind includes the following steps:

[0008] S1. Obtain the 3D information of the scene, and after filtering and filling, establish a 3D information raster map;

[0009] S2. Determine the starting point and ending point of the path on the grid map. The point with the minimum cost in the path is the parent node, and the eight-directional neighborhood around the parent node is the child node.

[0010] S3. Establish a heuristic function based on the starting point, the ending point, the parent node, and the child node;

[0011] The heuristic function includes the actual cost value from the starting point to the child node, and the estimated cost value from the child node to the ending point;

[0012] The actual cost from the starting point to the child node includes: the movement cost from the starting point to the parent node and the cost from the parent node to the child node. The cost from the parent node to the child node includes: the basic walking cost from the parent node to the child node, the crossing cost from the parent node to the child node, and the turning cost from the parent node to the child node.

[0013] According to the artificial potential field method, the normalized gravitational potential field and repulsive potential field of the child node are calculated, and the estimated cost value from the child node to the endpoint is weighted by the gravitational potential field, and the cost value from the parent node to the child node is weighted by the repulsive potential field to obtain the weighted heuristic function.

[0014] S4. Traverse multiple parent nodes in the path according to the weighted heuristic function, and end when the destination is reached. Obtain the obstacle avoidance path for the blind based on the traversed parent nodes.

[0015] Preferably, the three-dimensional information of the scene is obtained using binocular stereo vision;

[0016] Preferably, the estimated cost from the child node to the endpoint is the Euclidean distance from the child node to the endpoint;

[0017] Preferably, the crossing cost from the parent node to the child node is set to n times the free region cost. When it is impossible to bypass or the bypass cost exceeds n times the free region cost, crossing is selected.

[0018] Preferably, it is determined whether the child node, the parent node, and the parent node's parent node are on a straight line. If so, the turning cost is set to 0; otherwise, a set turning cost is added.

[0019] Preferred,

[0020] The normalized child nodes

[0021] Wherein, the maximum value of the gravitational potential field is the gravitational potential field between the starting point and the ending point;

[0022] Preferably, the gravitational potential field value of the child node is the product of the gravitational gain coefficient and the distance between the current child node and the endpoint, wherein the gravitational gain coefficient is defined according to requirements;

[0023] Preferred,

[0024] The normalized child nodes

[0025]

[0026] Wherein, the maximum value of the repulsive potential field is twice the repulsive potential field between two adjacent grid points;

[0027] Preferably, the repulsive potential field value of the child node is the product of the repulsive gain coefficient and the distance between the current child node and the obstacle;

[0028] The distance between the child node and the obstacle is obtained by superimposing the direction vectors of the repulsive potential field generated by the obstacle on the child node to obtain the combined repulsive potential field and then taking the modulus.

[0029] Preferably, when the child nodes in the forward direction of the parent node are both adjacent to obstacles on the left and right, the repulsive potential field value of the child node is set to the maximum value of the repulsive potential field.

[0030] As can be seen from the above technical solution, the present invention discloses a path planning method for obstacle avoidance for the blind. Compared with the prior art, the present invention adds crossing cost and turning cost to the heuristic function, so that the planned path minimizes crossing and turning, which fully solves the defects of the prior art in that it cannot solve the crossing and turning problems, and facilitates the actual travel of the blind. For the problem of being too close to the obstacle, the artificial potential field method is introduced to establish an attractive potential field and a repulsive potential field, and the heuristic function is weighted to avoid the problem of the planned path being too close to the obstacle to the greatest extent, thus meeting the obstacle avoidance needs of the blind.

[0031] Another advantage of this invention is that the three-dimensional information of the scene obtained by this invention, after being filtered and filled, creates a three-dimensional information grid map that can distinguish between obstacles and traversable obstacles, thereby providing a basis for subsequent path planning. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0033] Figure 1 The attached figure is a block diagram of the obstacle avoidance path planning method of the present invention;

[0034] Figure 2 The attached figure shows a three-dimensional information grid map established in the obstacle avoidance path planning method of the present invention;

[0035] Figure 3 The attached figure is a flowchart of the obstacle avoidance path planning method of the present invention for solving the crossing problem;

[0036] Figure 4 The attached figure is a schematic diagram of the path after introducing the crossing cost in the obstacle avoidance path planning method of the present invention;

[0037] Figure 5 The attached figure is a flowchart of the turning problem algorithm in the obstacle avoidance path planning method of the present invention;

[0038] Figure 6 The attached figure is a schematic diagram of the path after introducing turning costs in the obstacle avoidance path planning method of the present invention;

[0039] Figure 7 The attached figure shows the algorithm flow for the narrow area problem in the obstacle avoidance path planning method of the present invention;

[0040] Figure 8 The attached figure is a schematic diagram of the path after the introduction of the narrow area algorithm in the obstacle avoidance path planning method of the present invention. Detailed Implementation

[0041] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] This invention discloses a method for obstacle avoidance path planning for the blind, which mainly includes the following steps:

[0043] S1. Obtain the 3D information of the scene, and after filtering and filling, establish a 3D information raster map;

[0044] S2. Determine the starting and ending points of the path on the grid map. The point with the minimum cost in the path is the parent node, and the eight-directional neighborhood around the parent node is the child node.

[0045] S3. Create a heuristic function based on the start point, end point, parent node, and child nodes.

[0046] The heuristic function includes the actual cost from the starting point to the child nodes, and the estimated cost from the child nodes to the ending point;

[0047] The actual cost from the starting point to the child node includes: the cost of moving from the starting point to the parent node and the cost of moving from the parent node to the child node.

[0048] The cost from parent node to child node specifically includes: the basic traversal cost from parent node to child node, the crossing cost from parent node to child node, and the turning cost from parent node to child node;

[0049] Based on the artificial potential field method, the normalized gravitational and repulsive potential fields of the child nodes are calculated. The gravitational potential field is used to weight the estimated cost from the child node to the endpoint, and the repulsive potential field is used to weight the cost from the parent node to the child node, thus obtaining the weighted heuristic function.

[0050] S4. Traverse the parent nodes in the path according to the weighted heuristic function, and end when the destination is reached. Obtain the obstacle avoidance path for the blind based on the traversed parent nodes.

[0051] In summary, the obstacle avoidance path planning method for the blind disclosed in this invention, such as Figure 1 This includes basic path planning algorithms, improved algorithms for crossing obstacles, improved algorithms for turning penalty problems, and improved algorithms for narrow areas. By combining these algorithms, the path planning problem for blind people during obstacle avoidance can be solved.

[0052] The above path planning method will be further explained below with reference to the accompanying drawings.

[0053] First, stereo vision is used to acquire 3D information of the scene; then, the acquired data is filtered by height and filled by mapping to create a 3D information raster map.

[0054] In one embodiment, the raster map established by the present invention is as follows: Figure 2 As shown, black squares represent insurmountable obstacles, gray squares represent traversable obstacles, the bottom squares represent the starting point of the path, and the top squares represent the ending point of the path.

[0055] Furthermore, the path planning method disclosed in this invention is used to plan on the established three-dimensional information grid map to obtain obstacle avoidance paths for the blind.

[0056] Specifically, the starting and ending points of the path are determined on the grid map. Starting from the starting point, the search is performed in the neighborhood. The node with the lowest search cost is taken as the parent node, and the parent node is taken as the current node. The search continues in the neighborhood in the eight directions. At this time, the neighborhood in the eight directions is the child node.

[0057] In one embodiment, an open list and a close list are established. The open list contains the nodes to be searched, i.e., child nodes, and the close list contains the parent nodes, i.e., the nodes with the lowest cost found are added to the close list. Starting from the origin as the initial parent node, the search is performed on its eight-directional neighborhood. When the destination appears in the open list, the planning is complete. Based on the parent nodes added to the close list, the path with the lowest cost from the origin to the destination can be obtained.

[0058] When searching for the node with the minimum cost value, a heuristic function established according to the present invention is used. Specifically, the heuristic function of the present invention includes two parts: the actual cost value from the starting point to the child node and the estimated cost value from the child node to the end point. At the same time, according to the artificial potential field method, the normalized gravitational potential field and repulsive potential field of the child node are calculated, and the estimated cost value from the child node to the end point is weighted by the gravitational potential field, and the cost value from the parent node to the child node is weighted by the repulsive potential field.

[0059] In one embodiment, the heuristic function F(n) is calculated as follows:

[0060] F(n) = G(n) + H(n)

[0061] Where G(n) represents the actual cost value from the starting point to the child node, and H(n) represents the estimated cost value from the child node to the end point;

[0062] Furthermore, the estimated cost from the child node to the destination is the Euclidean distance from the child node to the destination, calculated using the following formula:

[0063]

[0064] Where (x1, y1) are the raster coordinates of the child node, and (x2, y2) are the raster coordinates of the endpoint.

[0065] In one embodiment, when calculating distance, if the original calculation is used, the cost of moving forward, backward, left, and right is 1, and the cost of moving diagonally is √2 (approximately 1.141, or about 1.4). The program is more complicated to calculate decimals and irrational numbers. To facilitate the calculation, the cost is multiplied by 10, that is, the cost value of the four directions of the grid forward, backward, left, and right is set to 10, and the cost value of the other four directions is set to 14.

[0066] The cost from the starting point to the child node specifically includes: the movement cost from the starting point to the parent node and the cost from the parent node to the child node. The cost from the parent node to the child node further includes: the basic traversal cost from the parent node to the child node, the crossing cost from the parent node to the child node, and the turning cost from the parent node to the child node. The specific expression can be represented as:

[0067] G(x,y)=G(x p ,y p )+Price base +Price across +Price turn

[0068] In the formula, G(x) p ,y p Price represents the cost of moving from the starting point to the parent node. base Price is the basic cost of walking from the parent node to the child node.across Price is the cost of traversing from a parent node to a child node. turn This represents the cost of turning from the parent node to the child node.

[0069] Regarding the traversal cost, since there are more traversable problems such as steps in the travel scenarios of blind people, and the existing basic A* algorithm cannot effectively handle the traversal problem, it is necessary to improve the algorithm and introduce traversal cost to solve the traversal problem.

[0070] Introducing the cost of crossing requires that the raster map obtained first distinguishes between the areas to be crossed and the areas that cannot be crossed. This problem has been solved in step S1.

[0071] Secondly, the cost of crossing from the parent node to the child node can be set to n times the cost of the free area. When it is impossible to bypass or the cost of bypassing exceeds n times the cost of the free area, crossing is chosen. In other words, when avoiding obstacles, bypassing is adopted if possible, and crossing is planned when it is impossible to bypass or the cost of bypassing is too high. The specific value of n can be flexibly set according to the grid size and the needs of blind people.

[0072] For blind people, it is best to choose a detour. However, since there may be situations where detours may require too much walking or may not be possible, crossing planning is necessary. In one embodiment, the crossing cost is set to 6 times the free area cost. That is, if the crossing area in the direction of travel can be bypassed within 6 free grids, then detour is chosen.

[0073] Refer to the algorithm flowchart for solving the crossing problem, such as... Figure 3 As shown, the process finds the child node to be filtered from the open list, which is the neighborhood of the current node. It determines whether it is an obstacle to cross. If not, it proceeds according to the normal planning process, either going straight or going around. If it is an obstacle to cross, it further determines whether the child node's position is forward. If it is not forward, it returns to the open list and searches for other child nodes to be filtered. If it is forward, it determines whether the crossable area around the child node is greater than 3. If so, it crosses the area and introduces a crossing cost. If it is not greater than 3, it abandons the process and returns to the open list to search for other child nodes to be filtered.

[0074] To determine if there are at least three obstacles around a child node, check if there are more than three obstacles to the right or left of the current node's centerline. Setting the number to three is merely an empirical value; crossing an obstacle requires 2 steps, while going around it requires 8 steps. If the difference in steps is greater than 6, then crossing the obstacle is chosen.

[0075] After introducing the cost of crossing, the planned path diagram is as follows: Figure 4 As shown, in Figure 4In this problem, it is impossible to reach the destination by detouring, so the only option is to cross over. This shows that adding a crossable cost factor can solve the shortcomings of common problems such as stair problems where it is impossible to plan a path. It has good applicability to crossable problems and can solve problems that traditional algorithms cannot plan, such as stair problems.

[0076] Furthermore, regarding turning costs, due to the heuristic minimum cost search characteristic of the A* algorithm, there will be many turning phenomena due to inappropriate cost selection and the influence of the grid environment. In the process of obstacle avoidance path planning for blind people, the path should be as simple as possible to facilitate obstacle avoidance. Therefore, this invention further introduces turning costs on the basis of the above to reduce the number of turns in the path and facilitate better travel for blind people. The specific cost can be flexibly set according to the grid size and the needs of blind people.

[0077] In one embodiment, the algorithm is configured as follows, which can be referred to in detail. Figure 5 First, select the filtered child node from the open list and record the position of the child node and its two parent nodes above it. That is, record the position of the child node, its parent node, and the parent node's parent node. Then, determine whether the three are on the same straight line. If they are, set the turning cost to 0. If not, there is a turn, and add the set turning cost value. Here, the turning cost value is set to 3 times that of the free area.

[0078] After improving the turning problem, further planning and control were carried out based on the crossing problem. The results are shown in the appendix. Figure 6 As shown in the results, this improved algorithm can reduce the number of turns during path planning and is more suitable for obstacle avoidance by blind people.

[0079] Integrating narrow areas into path planning involves using the artificial potential field method to establish gravitational and repulsive potential fields. The heuristic function is then weighted based on these fields to dynamically improve the grid cost, thus addressing the problem of planned paths being too close to obstacles and further ensuring the safety of blind people traveling.

[0080] Specific execution steps, such as Figure 7 As shown, firstly, select the filtered child nodes from the open list and record the starting point. Calculate the gravitational potential field value of the corresponding child node. Then, normalize the gravitational potential field value using the following method:

[0081] Child node normalization

[0082] Among them, the maximum value of the gravitational potential field is the gravitational potential field between the starting point and the ending point.

[0083] Then, the weighting coefficients of the estimated cost value from the child node to the endpoint are weighted using the normalized gravitational potential field value;

[0084] Since the A* programming algorithm plans the direction with low cost, the absolute value of the gravitational potential field can be used directly for calculation without considering its direction. Specifically, the gravitational potential field value is the product of the gravitational gain coefficient and the distance between the current child node and the endpoint. The gravitational gain coefficient can be defined according to the requirements (in this example, the gravitational gain coefficient is set to 15). The method for calculating the potential field in this invention is a simplified method designed to meet the requirements of this algorithm, based on the principle of the potential field and combined with the planning requirements of the A* algorithm.

[0085] Then, the maximum potential field is assumed to be the gravitational potential field between the starting point and the ending point. Therefore, its maximum value is used for normalization to obtain the grid point normalized gravitational potential field. At this time, in order to meet the minimum cost planning requirements of the A* algorithm, the normalization method is to divide the current gravitational potential field value by the maximum value of the gravitational potential field.

[0086] Furthermore, an obstacle list is obtained, and the effective range of the repulsive potential field can be determined based on the grid size. This reduces the computational load of the algorithm and improves the system speed while meeting obstacle avoidance requirements.

[0087] Calculate the repulsive potential field based on the positional relationship between the obstacles within the effective range and the current point. When calculating the resultant repulsive potential field, consider the direction. After the calculation is completed, take the modulus.

[0088] The specific process is as follows: the repulsive potential field is calculated by first calculating the distance between the child node and the obstacle, and then multiplying it by the repulsive gain coefficient to obtain its repulsive potential field. The repulsive gain coefficient can be defined according to the requirements (in this example, the repulsive gain coefficient is set to 5).

[0089] When there are more than one obstacle around a child node, they cannot be simply summed. The direction of the resultant potential field needs to be considered, that is, it needs to be regarded as multiple vectors. When summing multiple vectors, the direction vector of the repulsive potential field generated by the obstacle on the child node can be obtained according to the coordinates of the obstacle and the child node. By superimposing the repulsive potential fields of each obstacle one by one, the resultant repulsive potential field of the child node can be obtained. After obtaining the resultant repulsive potential field, take its modulus to simplify it into the potential field generated by the obstacle.

[0090] The default maximum repulsive potential field value is twice the repulsive potential field between two adjacent grid points. According to the requirements of the A* algorithm's heuristic function, the calculated normalized repulsive potential field needs to be inversely proportional to the distance. Therefore, its maximum value is used for normalization to obtain the normalized repulsive potential field of the grid points. In other words, normalization solves the problem of the repulsive potential field being inversely proportional to the distance required by the A* algorithm. To meet the minimum cost planning requirements of the A* algorithm, the normalization method is as follows:

[0091] Repulse after child node normalization

[0092] To avoid incorrect judgment of zero potential field in narrow areas due to opposite repulsive force directions, it is agreed that if the current node's forward direction is adjacent to obstacles on both the left and right, the node value is set to the maximum value.

[0093] That is, continue to refer to Figure 4 After obtaining the obstacle list, search for neighboring nodes, that is, the obstacles within 3 grids around the child node. Determine whether there are obstacles on the left and right sides of the child node in the direction of the child node's movement. If there are, set the repulsive potential field at the child node to the maximum value. If not, calculate the repulsive potential field value of the current child node normally, and then normalize it to weight the cost value from the parent node to the child node.

[0094] A weighted heuristic function is used for path planning to obtain obstacle avoidance paths for the blind. Specifically, heuristic functions derived from basic path planning algorithms, traversal problem algorithms, turning penalty algorithms, and narrow area algorithms are used to plan the path from the starting point to the destination. This path planning satisfies the blind user's requirements for traversing steps, solving turning problems, and navigating narrow areas. The overall path planning results obtained after applying the path planning method disclosed in this invention are shown in the appendix. Figure 8 As shown in the results, the path planned by this algorithm has good practicality for traversable areas, and the planned path has fewer turns and can avoid narrow areas as much as possible, making the planned path more suitable for blind people to travel. This provides new ideas and methods for the development of blind people's travel system.

[0095] In summary, this invention utilizes the 3D scene information obtained through binocular vision to reconstruct a scene environment grid map. The grid map is then height-divided to distinguish between obstacles and traversable obstacles. Traversal and turning costs are incorporated into the heuristic function, minimizing traversal and turning distances in the planned path, thus facilitating travel for the blind. To address the issue of paths being too close to obstacles, an artificial potential field method is introduced to establish gravitational and repulsive potential fields, which are then weighted in the heuristic function to minimize the problem of planned paths being too close to obstacles, thus meeting the obstacle avoidance needs of the blind.

[0096] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0097] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for obstacle avoidance path planning for the blind, characterized in that, Includes the following steps: S1. Obtain the 3D information of the scene, and after filtering and filling, establish a 3D information raster map; S2. Determine the starting point and ending point of the path on the grid map. The point with the minimum cost in the path is the parent node, and the eight-directional neighborhood around the parent node is the child node. S3. Establish a heuristic function based on the starting point, the ending point, the parent node, and the child node; The heuristic function includes the actual cost value from the starting point to the child node, and the estimated cost value from the child node to the ending point; The actual cost from the starting point to the child node includes: the movement cost from the starting point to the parent node and the cost from the parent node to the child node. The cost from the parent node to the child node includes: the basic walking cost from the parent node to the child node, the crossing cost from the parent node to the child node, and the turning cost from the parent node to the child node. The cost of crossing from the parent node to the child node is set to n times the cost of the free region. When it is impossible to bypass or the cost of bypassing exceeds n times the cost of the free region, crossing is selected. Determine whether the child node, the parent node, and the parent node's parent node are on a straight line. If yes, set the turning cost to 0; otherwise, add the set turning cost. According to the artificial potential field method, the normalized gravitational and repulsive potential fields of the child nodes are calculated. The estimated cost from the child node to the endpoint is weighted using the gravitational potential field, and the cost from the parent node to the child node is weighted using the repulsive potential field, resulting in a weighted heuristic function. ; Wherein, the maximum value of the gravitational potential field is the gravitational potential field between the starting point and the ending point; ; Wherein, the maximum value of the repulsive potential field is twice the repulsive potential field between two adjacent grid points; It is the product of the repulsive force gain coefficient and the distance between the current child node and the obstacle; the distance between the child node and the obstacle is obtained by superimposing the direction vectors of the repulsive potential fields generated by the obstacle on the child node to obtain the combined repulsive potential field and then taking the modulus; S4. Traverse multiple parent nodes in the path according to the weighted heuristic function, and end when the destination is reached. Obtain the obstacle avoidance path for the blind based on the traversed parent nodes.

2. The obstacle avoidance path planning method for blind people according to claim 1, characterized in that, The three-dimensional information of the scene is obtained using binocular stereo vision.

3. The obstacle avoidance path planning method for blind people according to claim 1, characterized in that, The estimated cost from the child node to the destination is the Euclidean distance from the child node to the destination.

4. The obstacle avoidance path planning method for blind people according to claim 1, characterized in that, It is the product of the gravity gain coefficient and the distance between the current child node and the endpoint, whereby the gravity gain coefficient is defined according to requirements.

5. The obstacle avoidance path planning method for blind people according to claim 1, characterized in that, When the child nodes in the direction of the parent node's movement are both adjacent to obstacles on the left and right, the repulsive potential field value of the child node is set to the maximum value of the repulsive potential field.

Images