Method for path exploration of mobile robot in semi-unknown environment based on ant crawling mechanism

Abstract

The application discloses a mobile robot path exploration method based on an ant crawling mechanism in a semi-unknown environment. A vertical mapping method is used to generate a candidate guide point set on a grid map; on the basis of the candidate guide point set, an ant crawling mechanism is used to generate a guide path; and dynamic obstacle avoidance is performed in the process that the mobile robot travels along the guide path. When the mobile robot predicts that a collision with a dynamic obstacle will occur, the mobile robot takes an obstacle avoidance strategy to perform emergency obstacle avoidance. After the obstacle avoidance behavior is completed, the mobile robot returns to the guide path, and then the mobile robot continues to travel along the guide path, so that the success rate of the mobile robot reaching a terminal point in the semi-unknown environment is improved, the quality of the exploration path of the mobile robot is improved, and the energy consumption required by the mobile robot to explore the path is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of mobile robot path planning, and particularly relates to a mobile robot path exploration method based on an ant crawling mechanism in a semi-unknown environment. BACKGROUND

[0002] With the rapid development of science and technology, mobile robots have been widely used in people's production and life. For example, in the fields of ground autonomous navigation, resource exploration and development, rescue and disaster relief, mobile robots can replace or assist humans to complete various complex work. At the same time, with the continuous expansion of mobile robot application scenarios and service modes, people's requirements for the intelligent degree of mobile robots are also getting higher and higher, and the core technology of the intelligent degree of mobile robots is path planning technology. Path planning refers to the exploration of a high-quality collision-free path from the starting point to the ending point in a complex spatial environment by a mobile robot, and a good path planning technology can effectively improve the utilization rate of resources of the mobile robot, reduce the loss of the mobile robot, and prolong the service life of the mobile robot.

[0003] At present, the mainstream algorithm of mobile robot path planning is divided into two categories: traditional algorithm and intelligent bionics algorithm. Among them, the traditional algorithm mainly includes A* algorithm, artificial potential field method, etc. The intelligent bionics algorithm mainly includes genetic algorithm, ant colony algorithm, particle swarm algorithm, immune algorithm, etc. The traditional algorithm has good performance in simple map environment, but it is not suitable for complex map environment. While the intelligent bionics algorithm has problems such as premature convergence, poor global path quality, and easy to fall into local extreme value. At the same time, semi-unknown environment widely exists in actual application environment, for example, the actual scene of earthquake relief, rescue and disaster relief is usually a semi-unknown environment. Compared with known environment, semi-unknown environment has higher complexity and stronger uncertainty. However, most of the current path planning algorithms are only suitable for known environment, and the applicability for semi-unknown environment is poor. Therefore, it is very necessary to study the path planning technology in semi-unknown environment.

[0004] In actual environment, because the specific environment information of some work area is unknown, and the path planning technology in known environment is mostly not suitable for semi-known environment, which will bring great challenge to the normal work of mobile robot, so a path planning technology in semi-known environment is needed. The path planning technology in semi-known environment refers to the technology that makes mobile robot reach the destination safely, efficiently and quickly by using real-time local information obtained by sensor and corresponding path planning mechanism in the case of lacking a large amount of global information. For the path planning technology in semi-known environment, experts and scholars have carried out in-depth research. Ningbo et al. proposed a path planning algorithm for robot exploration in semi-known environment, which uses finite state machine to iteratively derive the robot's action to plan the environment without prior map available, and realizes complete planning while maintaining efficiency. However, this algorithm does not consider the existence of dynamic obstacles, and the mobile robot may collide with dynamic obstacles. Wang Shirong et al. proposed a terrain-assisted path planning algorithm based on particle swarm optimization. The algorithm integrates terrain recognition strategy and obstacle avoidance strategy in a particle swarm-based path planning algorithm, which can plan a suitable path in semi-known terrain. However, this algorithm is only suitable for semi-known environment with simple obstacle distribution, and is not suitable for semi-known environment with high complexity. Xu Chang et al. proposed a real-time online path planning method based on deep neural network. The method is suitable for chaotic semi-known environment and has significant improvement in efficiency, success rate and path quality. However, this algorithm does not consider special terrains such as concave regions, which will make the mobile robot stagnate or oscillate. SUMMARY

[0005] Therefore, the purpose of the present application is to provide a mobile robot path exploration method in semi-known environment based on ant crawling mechanism, which generates a set of candidate guide points on the grid map by using vertical mapping method; on the basis of the set of candidate guide points, the guide path is generated by using ant crawling mechanism; dynamic obstacle avoidance is performed during the movement of the mobile robot along the guide path. The success rate of the mobile robot reaching the end point in semi-known environment is improved, and the quality of the exploration path of the mobile robot is also improved, and the energy consumption required for the exploration path of the mobile robot is reduced.

[0006] In order to achieve the above purpose, a mobile robot path exploration method in semi-known environment based on ant crawling mechanism, comprising the following steps:

[0007] S1, a spatial environment model is established by using grid method;

[0008] S2, the target point is projected vertically to the non-obstacle region in the known environment to generate a mapping point;

[0009] S3, a plurality of grids are expanded on both sides of the mapping point to generate a candidate area; the grids and obstacle grids beyond the map boundary in the candidate area are deleted, and other grids form a candidate guide point set;

[0010] S4, an ant crawling mechanism is used to generate candidate paths from the current position of the mobile robot to each point in the candidate guide point set; it is checked whether the current candidate guide point set is completely repeated with the historical candidate guide point set, if yes, S5 is entered, otherwise, S6 is entered;

[0011] S5, the mobile robot increases the detection range of the sensor to the maximum, then generates a virtual target point on the left or right side of the target point at an interval of n grid lengths, and replaces the current target point with the virtual target point, and then S2 is entered;

[0012] S6, a path evaluation mechanism is used to evaluate the quality of each candidate path, and the path with the best quality in the candidate path is selected as the guide path, and the candidate guide point corresponding to the guide path is regarded as the current guide point, and it is checked whether there is a virtual target point VG i , if yes, VG i is cancelled and the target point G is restored, and then S7 is entered;

[0013] S7, the mobile robot travels along the guide path to the current guide point i and updates the map environment information in real time, and when the mobile robot predicts that a collision with a dynamic obstacle will occur during the travel, an obstacle avoidance strategy is adopted to perform emergency obstacle avoidance;

[0014] S8, it is judged whether the target point G appears in the known environment, if yes, the mobile robot generates a guide path from the current position to the target point G using the ant crawling mechanism and travels along the path to the target point G; if no, S2 is entered again to enter the cycle;

[0015] Further preferably, in S1, when the spatial environment model is established by using the grid method, the following steps are included:

[0016] S101, the spatial environment is divided into a plurality of square grids of the same size;

[0017] S102, the grids are numbered in sequence from left to right and from top to bottom, starting from 1 to the last grid in the lower right corner;

[0018] S103, each grid is represented by the center point coordinates according to a preset coordinate conversion formula.

[0019] Further preferably, the preset coordinate conversion formula includes

[0020]

[0021] y = n + 0.5 - ceil(m / n) (2)

[0022] m = floor(l - y) * n + ceil(x) (3)

[0023] wherein formula (1) and formula (2) are used to convert the grid number into grid coordinates; formula (3) converts the grid coordinates into grid number; mod() is a remainder function, floor() is a down-rounding function, and ceil() is an up-rounding function; m is the grid number, n is the total column number of the grid, l is the total row number of the grid, (x, y) is the horizontal and vertical coordinates of the grid.

[0024] Further preferably, in S2, the mapping point generated by the vertical projection of the target point to the non-obstacle region in the known environment is calculated using the following formula:

[0025] i0 = G - h1 * l - h2 * l;

[0026] wherein G is the grid number of the target point, h1 is the total row number of the unknown environment in the vertical direction, h2 is the total row number of the obstacle region connected with the unknown environment in the vertical direction, and l is the total grid number of each row.

[0027] Further preferably, in S4, the generation of the candidate path of the mobile robot from the current position to each point in the candidate guide point set using the ant crawling mechanism comprises the following steps:

[0028] S401, initializing pheromone δ μv = c (c > 0) and emptying the path table L, storing the current position c_point of the mobile robot in the path table L, and setting c_point as the current position of the ant;

[0029] S402, the ant searches the path from the current position, calculates the feasibility I μv of transferring from the current grid point to each adjacent non-obstacle grid point according to formula (4), and selects the next grid point according to the maximum feasibility;

[0030]

[0031] wherein I μv is the feasibility of the ant between path point μ and path point v, J(μ) is the next feasible path point set of path point μ, δ μv is the pheromone concentration between path point μ and path point v, is the heuristic information between path point μ and path point v, indicating the reciprocal of the distance between path point μ and v, k is the pheromone heuristic factor, and γ is the expected heuristic factor.

[0032] S403. Store the selected next grid point number in the path table L, and set this grid point as the ant's current position. Determine if the current position number matches the number of the candidate guide point. If they do not match, proceed to S402. If they match, the process ends, and the path in the path table L is the candidate path from the mobile robot's current position c_point to the candidate guide point.

[0033] In a further preferred embodiment, in S6, when using a path evaluation mechanism to evaluate the quality of the selected path, the evaluation indicators for path quality include path reachability, path safety, path length, and path smoothness.

[0034] More preferably, the formula for calculating the path quality is as follows:

[0035] Q = ω １ P f +ω2P s +ω3P l +ω4P m (5)

[0036] Where Q represents path quality, P f P represents path reachability. s P represents path security. l P represents the path length. m ω1 represents path smoothness; ω2, ω3, and ω4 are the weighting coefficients of path reachability, path safety, path length, and path smoothness, respectively, and ω1+ω2+ω3+ω4=1, ω1≥ω2≥ω3≥ω4.

[0037] More preferably, in S7, when the mobile robot predicts an impending collision with a dynamic obstacle, the specific process for predicting the collision is as follows:

[0038] Different collision methods are used to classify collision types into four categories: frontal collision, side collision, stationary collision, and chasing collision;

[0039] Assume the current position of the mobile robot is R. c The coordinates are (x c ,y c ), next position R n The coordinates are (x n ,y n ); Current position of dynamic obstacle O c The coordinates are (r c ,c c ), next position O n The coordinates are (r n ,c n );

[0040] When the coordinates of the mobile robot and the dynamic obstacle satisfy the following relationship (11), it is determined that the mobile robot and the dynamic obstacle have a positive collision with a common collision point;

[0041]

[0042] When the coordinates of the mobile robot and the dynamic obstacle satisfy the following relationship (12), it is determined that the mobile robot and the dynamic obstacle have a positive collision without a common collision point

[0043]

[0044] When the coordinates of the mobile robot and the dynamic obstacle satisfy the following relationship (13), it is determined that the mobile robot and the dynamic obstacle have a lateral collision;

[0045] x n =r n &y n =c n (13)

[0046] When the coordinates of the mobile robot and the dynamic obstacle satisfy the following relationship (14), it is determined that the mobile robot and the dynamic obstacle have a stationary collision;

[0047] x n =r n &y n =c n , r n =r c &c n =c c (14)

[0048] When the coordinates of the mobile robot and the dynamic obstacle satisfy the following relationship (14), it is determined that the mobile robot and the dynamic obstacle have a pursuit collision;

[0049]

[0050] Further preferably, in S7, the taking of the obstacle avoidance strategy for emergency obstacle avoidance is a turning behavior strategy, a stationary waiting strategy, and a local path re-planning strategy.

[0051] Further preferably, the turning behavior strategy is used when the mobile robot perceives that a positive face collision, a positive point collision, or a pursuit collision with the dynamic obstacle is imminent, and the mobile robot avoids the dynamic obstacle by turning left or right.

[0052] The stationary waiting strategy is used when the mobile robot perceives that a lateral collision with the dynamic obstacle is imminent, and the mobile robot avoids the dynamic obstacle by pausing in place.

[0053] Further preferably, the local path re-planning strategy is used to re-plan the local path between the current path point i and the i+2 path point in a way that the mobile robot avoids the dynamic obstacle by using the ant crawl mechanism when the dynamic obstacle is static in the path of the mobile robot; after the mobile robot avoids the dynamic obstacle by using the above-mentioned way, the mobile robot will return to the guide path and then continue to travel along the guide path.

[0054] The mobile robot path exploration method based on the ant crawl mechanism in the semi-unknown environment has the following beneficial effects: first, the path evaluation mechanism can improve the success rate of the mobile robot reaching the destination in the semi-unknown environment, and can also improve the quality of the exploration path of the mobile robot and reduce the energy consumption required for the exploration path of the mobile robot. Second, the ant crawl mechanism can effectively deal with various complex obstacle environments such as concave regions, and ensure that the mobile robot will not be in a static or oscillation state. Third, the mobile robot can freely switch between the guide path and the obstacle avoidance strategy, which can not only ensure that the mobile robot can successfully avoid dynamic obstacles, but also ensure the quality of the mobile robot travel path. Fourth, the collision types between the mobile robot and the dynamic obstacle are discussed and the corresponding detection method is provided. The detection method can greatly reduce the probability of collision between the mobile robot and the dynamic obstacle. Fifth, the obstacle avoidance strategy makes full use of the behavior obstacle avoidance and local path re-planning strategy to deal with various potential collision situations. The obstacle avoidance strategy can greatly improve the obstacle avoidance time of the mobile robot and reduce the obstacle avoidance cost of the mobile robot. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 The working principle diagram of the mobile robot path exploration method based on the ant crawl mechanism in the semi-unknown environment is provided.

[0056] Figure 2 The schematic diagram of establishing a space environment model by using a grid method is provided.

[0057] Figure 3 (a) The schematic diagram of the forward collision between the mobile robot and the dynamic obstacle with a common point is provided.

[0058] Figure 3 (b) The schematic diagram of the forward collision without a common point is provided.

[0059] Figure 3 (c) The schematic diagram of the lateral collision between the mobile robot and the dynamic obstacle is provided.

[0060] Figure 3(d) a schematic diagram of the pursuit collision of the mobile robot with the dynamic obstacle provided by the present application. DETAILED DESCRIPTION

[0061] The present application is further described in detail by the accompanying drawings and specific embodiments.

[0062] As Figure 1 shown, the mobile robot path exploration method in semi-unknown environment based on the ant crawling mechanism provided by the present application includes the following steps in sequence:

[0063] S1, a spatial environment model is established by using the grid method (as Figure 2 shown). The grid method refers to dividing the entire two-dimensional spatial environment into a plurality of square grids of the same size. Among them, the white grid A represents the free area, the black grid B represents the static obstacle, the gray grid C represents the dynamic obstacle, and the black grid D represents the unknown environment area. The map is numbered from left to right and from top to bottom, starting from 1 to add numbers to the grids until the last grid in the lower right corner. The coordinates of each grid in the grid map are represented by the center point coordinates, and the conversion calculation between the grid number and the grid coordinates is:

[0064]

[0065] y = n + 0.5 - ceil(m / n) (2)

[0066] m = floor(l-y)*n + ceil(x) (3)

[0067] Among them, formula (1), (2) converts the grid number into the grid coordinates. Formula (3) converts the grid coordinates into the grid number. mod() is the remainder function, floor() is the floor function, and ceil() is the ceiling function. In the formula, m is the grid number, n is the total number of columns of the grid, l is the total number of rows of the grid, and (x, y) is the horizontal and vertical coordinates of the grid.

[0068] S2: The target point G point (virtual target point VG i point) is vertically projected to the non-obstacle area in the known environment to generate a mapping point i0; i0 = G-h1*l-h2*l, where G is the grid number of the target point, h1 is the total number of rows of the unknown environment in the vertical direction, h2 is the total number of rows of the obstacle area connected with the unknown environment in the vertical direction, and l is the total number of grids per row;

[0069] S3: Expand n grid lengths on both sides of i0 to generate a selected area, so that the selected area contains 2n+1 grids. Remove the grids and obstacle grids in the selected area that exceed the map boundary, and the remaining grids are the selected guide point set i_set.

[0070] S4: generate the candidate paths from the current position c_point of the mobile robot to each point in the set i_set by using the ant crawling mechanism. Check if the current set i_set is completely identical to the historical set i_set. If yes, go to step 5, otherwise go to step 6;

[0071] S5: increase the detection range of the sensor to the maximum, and then generate the virtual goal points VG i on the left or right side of the point G with the interval of n grids. Replace the point G temporarily, and go to S2; i i S6: evaluate the quality of each candidate path by using the path evaluation mechanism, select the best path in the candidate paths as the guide path, and the candidate guide point corresponding to the guide path is the current guide point i. Check if the virtual goal point VG i exists. If yes, remove the virtual goal point VG i and restore the point G;

[0072] S7: the mobile robot travels along the guide path to the current guide point i and updates the map environment information in real time. When the mobile robot predicts that a collision with a dynamic obstacle will occur during the travel, take the obstacle avoidance strategy to avoid the obstacle;

[0073] It should be noted that the function of step S4 is to check if the mobile robot is in a stagnant state due to the terrain. Normally, if the mobile robot is not in a stagnant state, go to step S6, and there is no need to remove the virtual goal point. If the mobile robot is in a stagnant state, go to step S5 to generate a virtual goal point, which will replace the original goal point to return to step S2, and execute until step S4. If the mobile robot is not in a stagnant state, go to step S6. At this time, the virtual goal point is involved in the process instead of the original goal point. Therefore, it is necessary to remove the virtual goal point and restore the original goal point.

[0074] S7: the mobile robot travels along the guide path to the current guide point i and updates the map environment information in real time. When the mobile robot predicts that a collision with a dynamic obstacle will occur during the travel, take the obstacle avoidance strategy to avoid the obstacle;

[0075] S8: determine if the point G exists in the known environment. If the point G exists in the known environment, the mobile robot generates the guide path from the current position c_point to the point G by using the ant crawling mechanism and travels along the path to the point G, and the process ends. If the point G does not exist in the known environment, go to S2;

[0076] In the above implementation process, S7 and S8 are parallel steps, and there is no strict restriction on the sequence of implementation.

[0077] ​In the implementation process, the specific process of generating a candidate path from the current position c_point of the mobile robot to the candidate guide point in S4 by using the ant crawling mechanism is as follows:

[0078] S401, initializing pheromone δ μv =c(c>0) and emptying the path table L. The current position c_point of the mobile robot is stored in the path table L, and c_point is set as the current position of the ant;

[0079] S402, the ant searches a path from the current position, and the feasibility I of transferring from the current grid point to each adjacent unobstructed grid point is calculated according to formula (4) μv , and the next grid point is selected according to the maximum feasibility;

[0080]

[0081] wherein I μv is the feasibility of the ant between the path point μ and the path point v, J(μ) is the set of next feasible path points of the path point μ, δ μv is the pheromone concentration between the path point μ and the path point v, is the heuristic information between the path point μ and the path point v, indicating the reciprocal of the distance between the path point μ and the path point v, k is the pheromone heuristic factor, and γ is the expected heuristic factor.

[0082] S403, the number of the selected next grid point is stored in the path table L, and the grid point is set as the current position of the ant. It is judged whether the number of the current position is consistent with the number of the candidate guide point. If not, go to S402. If consistent, the process is ended, and the path in the path table L is the candidate path from the current position c_point of the mobile robot to the candidate guide point;

[0083] In the implementation process, the specific process of evaluating the quality of the candidate path by using the path evaluation mechanism in S6 is as follows:

[0084] In order to enable the mobile robot to explore a feasible and high-quality path in a semi-unknown environment, a path evaluation mechanism based on multiple performance indexes is designed. The evaluation indexes of the path quality include path accessibility, path safety, path length and path smoothness. The calculation formula of the path quality is shown in formula (5)

[0085] Q=ω1P f +ω2P s +ω3P l +ω4P m (5)

[0086] wherein Q represents the path quality, P f represents the path accessibility, Ps denotes path safety, P l denotes path length, P m denotes path smoothness. ω1, ω2, ω3, ω4 are weight coefficients of path accessibility, path safety, path length, path smoothness respectively and ω1+ω2+ω3+ω4=1.

[0087] For the above four path quality evaluation indexes, the Delphi empowerment method is used to evaluate the weight coefficients of each path index, so as to obtain a set of better weight coefficients. First, the importance order of the four indexes is specified according to the actual demand. Considering the particularity of the semi-unknown environment, the importance order of the four indexes is: path accessibility, path safety, path length, path smoothness, that is, the accessibility and safety of the exploration path are given priority, and then the path length and path smoothness are considered, so ω1≥ω2≥ω3≥ω4.

[0088] The importance ratio is defined as shown in equation (6):

[0089]

[0090] I1=1.3, I2=1.9, I3=1.6. That is, in a semi-unknown environment, the importance of path accessibility is slightly higher than that of path safety, the importance of path safety is much higher than that of path length, and the importance of path length is higher than that of path smoothness. Then ω1, ω2, ω3, ω4 can be solved as: k ≥1, I k the greater the more important the former than the latter.

[0091] The importance ratio is defined. When I k =1, it means that the importance of the former and the latter is the same; when I k =1.3, it means that the importance of the former is slightly higher than that of the latter; when I k =1.6, it means that the importance of the former is higher than that of the latter; when I k =1.9, it means that the importance of the former is much higher than that of the latter.

[0092] According to the defined importance ratio I k , the calculation process of ω4 is shown in equations (7)-(9).

[0093]

[0094]

[0095]

[0096] The present application specifies I1=1.3, I2=1.9, I3=1.6. That is, in a semi-unknown environment, the importance of path accessibility is slightly higher than that of path safety, the importance of path safety is much higher than that of path length, and the importance of path length is higher than that of path smoothness. Then ω1, ω2, ω3, ω4 can be solved as:

[0097]

[0098] The weight coefficients of the path reachability, the path security, the path length and the path smoothness are 0.4117, 0.3171, 0.1669 and 0.1043 respectively.

[0099] In the implementation process, when the mobile robot predicts that a collision with the dynamic obstacle will occur, the specific process of taking an obstacle avoidance strategy for emergency obstacle avoidance in step seven is as follows:

[0100] The present application divides the collision types into four categories according to different collision modes: forward collision, lateral collision, stationary collision and pursuit collision. Among them, the forward collision is divided into forward point collision and forward surface collision. The forward collision refers to the face-to-face collision situation of the mobile robot and the dynamic obstacle due to the collinear and opposite motion directions. Assuming that the coordinates of the current position R c of the mobile robot are (x c , y c ), the coordinates of the next position R n are (x n , y n ), the coordinates of the current position O c of the dynamic obstacle are (r c , c c ), and the coordinates of the next position O n are (r n , c n ).

[0101] If the mobile robot and the dynamic obstacle have a common collision point when the forward collision occurs, this collision situation is called forward point collision (as shown in (a)). Figure 3

[0102]

[0103] If the mobile robot and the dynamic obstacle do not have a common collision point when the forward collision occurs, that is, the next step of the mobile robot and the dynamic obstacle is the mutual travel path point. In this case, although there is no collision point, the mobile robot and the dynamic obstacle will also collide, and at this time, the collision situation is judged as forward surface collision (as shown in (b)). Figure 3

[0104]

[0105] The lateral collision refers to the situation that the mobile robot and the dynamic obstacle do not have a common line of motion, but at a certain moment, they have a common collision point (as shown in (c)). Figure 3

[0106] x n ==r n &&y n ==c​​​n (13)

[0107] Stagnant collision refers to the case that the dynamic obstacle happens to stay on the guide path due to some reason, resulting in the mobile robot and the obstacle will definitely collide. That is

[0108] x n == r n && y n == c n , r n == r c && c n == c c (14)

[0109] Pursuit collision refers to the case that the mobile robot and the dynamic obstacle move collinearly and in the same direction, but due to the speed mismatch between the two, a common collision point is generated at a certain moment (as shown in (d)). That is Figure 3

[0110]

[0111] The present application designs three kinds of obstacle avoidance strategies, which are turning behavior strategy, in-place waiting strategy and local path re-planning strategy. The turning behavior strategy refers to the case that when the mobile robot perceives that it will collide with the dynamic obstacle in a forward face collision, a forward point collision or a pursuit collision, the mobile robot avoids the dynamic obstacle by turning left or right.

[0112] The in-place waiting strategy refers to the case that when the mobile robot perceives that it will collide with the dynamic obstacle in a lateral collision, the mobile robot avoids the dynamic obstacle by pausing in place.

[0113] The local path re-planning strategy refers to the case that when the dynamic obstacle stagnates on the path of the robot, the robot re-plans the local path between the current path point (the i-th path point) and the i+2-th path point in the way of avoiding the dynamic obstacle by using the ant crawling mechanism. After the mobile robot avoids the dynamic obstacle in the above way, the mobile robot will return to the guide path and continue to travel along the guide path.

[0114] Therefore, when the mobile robot and the dynamic obstacle are about to collide in a forward collision or a pursuit collision, the turning behavior strategy is adopted to avoid; when the mobile robot and the dynamic obstacle are about to collide in a lateral collision, the in-place waiting strategy is adopted to avoid; when the mobile robot and the dynamic obstacle are about to collide in a stagnant collision, the local path re-planning strategy is adopted to avoid.

[0115] ​Obviously, the above embodiments are only examples for clearly illustrating the present application and are not intended to limit the present application. Based on the above description, other different forms of changes or variations can be made by those skilled in the art. Here, all the embodiments are not required to be enumerated. The changes or variations derived from the above are still within the protection scope of the present application.

Claims

1. A method for path exploration of a mobile robot in a semi-unknown environment based on an ant crawling mechanism, characterized in that, The method comprises the following steps: S1, a grid method is used to establish a space environment model; S2, a target point is projected vertically to a non-obstacle region in a known environment to generate a mapping point; S3, a plurality of grids are expanded on the left and right sides of the mapping point to generate a candidate region; In the candidate region, the grids beyond the map boundary and the obstacle grids are deleted, and the other grids form a candidate guide point set; S4, an ant crawling mechanism is used to generate a candidate path from the current position of the mobile robot to each point in the candidate guide point set; it is checked whether there is a complete repetition of the current candidate guide point set and the historical candidate guide point set, if there is, S5 is entered, otherwise, S6 is entered; S5, the mobile robot increases the detection range of the sensor to the maximum, then generates a virtual target point on the left or right side of the target point at an interval of n grid lengths, and replaces the current target point with the virtual target point, and then S2 is entered; S6, using a path evaluation mechanism to evaluate the quality of each candidate path, selecting the path with the best quality in the candidate paths as the guide path, regarding the candidate guide point corresponding to the guide path as the current guide point, checking whether there is a virtual target point VG i , if there is, then canceling VG i , and restoring the target point G point, and turning to S7; S7, the mobile robot travels along the guide path to the current guide point i and updates the map environment information in real time; during the travel, when the mobile robot predicts that a collision with a dynamic obstacle will occur, an obstacle avoidance strategy is adopted to perform emergency obstacle avoidance; S8, it is judged whether a target point G appears in the known environment, if the target point G appears in the known environment, the mobile robot generates a guide path from the current position to the target point G by using the ant crawling mechanism and travels along the path to the target point G; if the target point G does not appear in the known environment, S2 is entered again to enter the loop.

2. The path exploration method for mobile robots in semi-unknown environments based on the ant crawling mechanism according to claim 1, characterized in that, In S1, when the grid method is used to establish the space environment model, the following steps are included: S101, the space environment is divided into a plurality of square grids of the same size; S102, the grids are numbered in sequence from left to right and from top to bottom, starting from 1 until the last grid at the bottom right corner is numbered; S103, the grids are represented by the center point coordinates according to a preset coordinate conversion formula.

3. The mobile robot path exploration method in a semi-unknown environment based on an ant crawling mechanism according to claim 2, characterized in that, The preset coordinate conversion formula includes y = n + 0.5 - ceil(m / n) (2) m = floor(l-y)*n + ceil(x) (3) Wherein, formula (1) and formula (2) are used to convert the grid number into grid coordinates; formula (3) converts the grid coordinates into grid numbers; mod() is a remainder function, floor() is a down rounding function, and ceil() is an up rounding function; m is the grid number, n is the total number of grid columns, l is the total number of grid rows, and (x, y) is the horizontal and vertical coordinates of the grid.

4. The path exploration method for mobile robots in semi-unknown environments based on the ant crawling mechanism according to claim 1, characterized in that, In S2, the mapping point generated by projecting the target point vertically to the non-obstacle region in the known environment is calculated using the following formula: i0 = G-h1*l-h2*l; Wherein, G is the grid number of the target point, h1 is the total number of rows of the unknown environment in the vertical direction, h2 is the total number of rows of the obstacle region connected with the unknown environment in the vertical direction, and l is the total number of grids per row.

5. The path exploration method for mobile robots in semi-unknown environments based on the ant crawling mechanism according to claim 1, characterized in that, In S4, the ant crawling mechanism is used to generate a candidate path from the current position of the mobile robot to each point in the candidate guide point set, which includes the following steps: S401, initialize pheromone δ μv = c (c > 0) and empty the path table L, store the current position c_point of the mobile robot in the path table L, and set c_point as the current position of the ant; S402、the ant sets out from the current position to search for a path, and calculates the feasibility I of transferring from the current grid point to each adjacent unobstructed grid point according to formula (4) μv and selects the next grid point according to the maximum feasibility where I μv is the feasibility from path point μ to path point v, J(μ) is the next feasible path point set of path point μ, δ μv is the pheromone concentration from path point μ to path point v, is the heuristic information from path point μ to path point v, indicating the reciprocal of the distance between path point μ and v, k is the pheromone heuristic factor, and γ is the expected heuristic factor. S403, store the number of the selected next grid point into the path table L, and set the grid point as the current position of the ant; determine whether the number of the current position is consistent with the number of the selected guide point, if not, go to S402; if yes, the flow ends, and the path in the path table L is the selected path from the current position c_point of the mobile robot to the selected guide point.

6. The path exploration method for mobile robots in semi-unknown environments based on the ant crawling mechanism according to claim 1, characterized in that, In S6, when the path evaluation mechanism is used to evaluate the quality of the selected path, the evaluation indexes of the path quality include path accessibility, path safety, path length, and path smoothness.

7. The mobile robot path exploration method in a semi-unknown environment based on an ant crawling mechanism according to claim 6, characterized in that, The calculation formula of the path quality is as follows: Q = ω1P f + ω2P s + ω3P l + ω4P m (5) wherein Q represents path quality, P f represents path reachability, P s represents path safety, P l represents path length, P m represents path smoothness; ω1, ω2, ω3, ω4 are weight coefficients of path reachability, path safety, path length, path smoothness respectively and ω1+ω2+ω3+ω4=1, ω1≥ω2≥ω3≥ω4.

8. The mobile robot path exploration method in a semi-unknown environment based on an ant crawling mechanism according to claim 1, characterized in that, In S7, when the mobile robot predicts that a collision with the dynamic obstacle will occur, the specific process of the prediction of the collision is as follows: Different collision modes are divided into four types: forward collision, lateral collision, stationary collision, and pursuit collision. Assume the current position of the mobile robot is R. c The coordinates are (x c y c ), next position R n The coordinates are (x n y n ); Current position of dynamic obstacle O c The coordinates are (r c c c ), next position O n The coordinates are (r n c n ); When the coordinates of the mobile robot and the dynamic obstacle satisfy the following relationship (11), it is determined that the mobile robot and the dynamic obstacle have a forward collision with a common collision point. When the coordinates of the mobile robot and the dynamic obstacle satisfy the following relationship (12), it is determined that the mobile robot and the dynamic obstacle have a forward collision without a common collision point. When the coordinates of the mobile robot and the dynamic obstacle satisfy the following relationship (13), it is determined that the mobile robot and the dynamic obstacle have a lateral collision. x n = r n & y n = c n (13) When the coordinates of the mobile robot and the dynamic obstacle satisfy the following relationship (14), it is determined that the mobile robot and the dynamic obstacle have a stationary collision. x n = r n & y n = c n , r n = r c & c n = c c (14) When the coordinates of the mobile robot and the dynamic obstacle satisfy the following relationship (15), it is determined that the mobile robot and the dynamic obstacle have a pursuit collision. 。 9. The mobile robot path exploration method in a semi-unknown environment based on an ant crawling mechanism according to claim 1, characterized in that, In S7, the obstacle avoidance strategy is used for emergency obstacle avoidance, and the obstacle avoidance strategy includes a turning behavior strategy, a stationary waiting strategy, and a local path re-planning strategy.

10. The mobile robot path exploration method in a semi-unknown environment based on an ant crawling mechanism according to claim 9, characterized in that, The turning behavior strategy is used for avoiding the dynamic obstacle by left turning or right turning when the mobile robot perceives that a forward surface collision, a forward point collision, or a pursuit collision will occur. The stationary waiting strategy is used for avoiding the dynamic obstacle by stationary waiting when the mobile robot perceives that a lateral collision will occur. The local path re-planning strategy is used for avoiding the dynamic obstacle by re-planning a local path between the current path point i and the i+2th path point when the dynamic obstacle is stationary on the path of the mobile robot. After the mobile robot avoids the dynamic obstacle by the above-mentioned methods, the mobile robot will return to the guide path and continue to travel along the guide path.

Images