CN116880507A - Cooperative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraint - Google Patents
Cooperative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraint Download PDFInfo
- Publication number
- CN116880507A CN116880507A CN202311039966.1A CN202311039966A CN116880507A CN 116880507 A CN116880507 A CN 116880507A CN 202311039966 A CN202311039966 A CN 202311039966A CN 116880507 A CN116880507 A CN 116880507A
- Authority
- CN
- China
- Prior art keywords
- robot
- task
- robots
- paths
- individual
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000005457 optimization Methods 0.000 claims abstract description 16
- 108090000623 proteins and genes Proteins 0.000 claims description 39
- 230000035772 mutation Effects 0.000 claims description 10
- 238000010845 search algorithm Methods 0.000 claims description 9
- 238000006467 substitution reaction Methods 0.000 claims description 3
- 238000002474 experimental method Methods 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000013138 pruning Methods 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
The invention discloses a cooperative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraint, which comprises the following steps: inputting starting points to multiple robots in a continuous map; adding a plurality of task points into the continuous map; performing multi-task distribution on the multiple robots according to an optimization algorithm; the multiple robots plan paths conforming to kinematic constraints according to the assigned tasks. The invention combines the optimization algorithm with the path planning of the multiple robots with kinematic constraint, so that the multiple robot system can finish a plurality of tasks, the number of conflict times and the path searching time are reduced, and the safety of the system is improved by combining with the actual system; the conflict among robots is effectively reduced, and the path cost of the multi-robot system is reduced; experiments further show that the method ensures the execution of path planning of multiple robots under the condition that multiple tasks exist, greatly shortens the time for solving the path of the system on the premise of normal solving, and improves the solving speed of the system.
Description
Technical Field
The invention relates to a cooperative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraint, and belongs to the field of multi-robot path planning.
Background
The problem of multi-robot task allocation and path planning is to allocate a specified number of tasks to multiple robots and plan a collision-free path for a given starting point and allocated task target points.
The research results of multi-robot task allocation and path planning can solve the problems in practical engineering in a plurality of fields, such as warehouse logistics, airport transportation and the like. However, for a transportation scene with kinematic constraint, the multi-robot system under the common grid map cannot meet the scene requirement, so that it is necessary to study and consider the multi-robot task allocation and path planning method with kinematic constraint.
Disclosure of Invention
The invention provides a cooperative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraint, which is used for acquiring paths of multiple robots while allocating tasks.
The technical scheme of the invention is as follows:
according to an aspect of the present invention, there is provided a cooperative scheduling method of multi-task allocation and multi-robot path planning with kinematic constraint, including: inputting starting points to multiple robots in a continuous map; adding a plurality of task points into the continuous map; performing multi-task distribution on the multiple robots according to an optimization algorithm; the multiple robots plan paths conforming to kinematic constraints according to the assigned tasks.
The multi-robot multi-task allocation according to the optimization algorithm comprises the following steps: iteratively evolving the starting points and the task points of the multiple robots in the continuous map according to the overall fitness function to obtain task points distributed by each robot; the overall fitness function is the sum of the path fitness function and the conflict fitness function.
The path fitness function is a path length sum which is obtained from the starting point of the robot to each task point by using a standard A-type algorithm.
The conflict fitness function is the sum of penalty values generated by robots meeting intersection judgment conditions.
The iterative evolution is carried out on the starting points and the task points of the multiple robots in the continuous map according to the overall fitness function to obtain the task points distributed by each robot, and the method comprises the following steps:
according to the random mode, N task points in the continuous map are distributed to M robots, each robot and the task point distributed by the robot form a gene, the M robot genes are used as a population individual, and P individuals are generated by P times of random distribution according to the mode;
carrying out mutation cross operation on P individuals in sequence, after the operation is finished, carrying out individual sorting according to the overall fitness value of a single individual, selecting Q individuals with the overall fitness value at the front for reservation, and expanding the population scale into P individuals according to the Q individuals reserved; and iterating according to the variation cross operation process until reaching a final condition, and reserving and outputting the individual with the minimum overall fitness value to obtain the task points distributed by each robot.
The mutation crossover operation in each iteration process is specifically as follows:
mutation operation: judging each gene in an individual, if the probability of random generation is greater than a first preset probability, randomly selecting a task point sequence number A in a task set range, changing the task point sequence number B of a robot corresponding to the randomly selected current gene into the task point sequence number A, and replacing the task point sequence number A of the robot corresponding to the gene with the task point sequence number A with the task point sequence number B;
crossover operation: judging each gene in the individual, and if the probability of random generation is greater than the second preset probability, replacing all task point serial numbers allocated to robots corresponding to the current gene with all task point serial numbers allocated to robots corresponding to a randomly selected gene in the same individual; wherein the two genes involved in the substitution operation are different genes.
The single individual overall fitness value acquisition process comprises the following steps:
solving the path length of the robot starting point sequentially reaching each allocated task point through a standard A-type algorithm according to the task points allocated by the robot in each individual, and obtaining the path fitness value of the single individual;
determining the number of the intersections of the robot paths in each individual according to the intersection judgment conditions; according to the number of the intersection points, a conflict fitness value of robots in each individual is obtained; the conflict fitness value of the robots in each individual in turn is obtained, and the conflict fitness of the individual is obtained;
and obtaining the overall fitness value of the single individual according to the path fitness value of the single individual and the conflict fitness of the single individual.
The intersection judgment conditions specifically include: and if the intersection point occurs between any two robot paths and the distance between the intersection point and the position point of the two robot paths before the intersection point is in a preset range, the intersection point count is increased by 1.
The multi-robot plans a path conforming to kinematic constraint according to the distributed tasks, and specifically comprises the following steps: the multiple robots use an improved vehicle-like collision search algorithm to plan paths conforming to kinematic constraints according to the assigned tasks.
The multi-robot uses an improved vehicle-like collision search algorithm to plan a path conforming to kinematic constraints according to the assigned tasks, comprising: and sequentially planning all robots from the self starting point according to task points by utilizing a mixed A algorithm to obtain paths, judging whether collision exists on the paths, storing the paths of the robots without collision, searching the first collision among the paths for the paths with collision, judging the priority condition of the robot with the first collision, storing the paths of the robots with the highest priority, adding path constraint to the robots with the highest priority to reprogram the paths, continuing searching the next collision among the paths until no collision exists among the paths of all robots, and integrating to obtain the paths of all robots conforming to the kinematic constraint.
According to another aspect of the present invention, there is provided a co-scheduling system for multi-tasking and multi-robot path planning with kinematic constraints, comprising the modules of the method of any of the above.
According to another aspect of the present invention, there is provided a processor for running a program, wherein the program when run performs the co-scheduling method of multi-tasking with kinematic constraints and multi-robot path planning as described in any of the above.
According to another aspect of the present invention, there is provided a computer readable storage medium, the computer readable storage medium including a stored program, wherein when the program is run, the apparatus in which the computer readable storage medium is located is controlled to perform the collaborative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraints as described in any one of the above.
The beneficial effects of the invention are as follows: the invention combines the optimization algorithm with the path planning of the multiple robots with kinematic constraint, so that the multiple robot system can finish a plurality of tasks, the number of conflict times and the path searching time are reduced, and the safety of the system is improved by combining with the actual system; the conflict among robots is effectively reduced, and the path cost of the multi-robot system is reduced; experiments further show that the method ensures the execution of path planning of multiple robots under the condition that multiple tasks exist, greatly shortens the time for solving the path of the system on the premise of normal solving, and improves the solving speed of the system.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a schematic diagram of path fitness function interpretation;
FIG. 3 is a diagram illustrating a collision fitness function interpretation;
FIG. 4 is a schematic view of PCL-CBS pruning;
FIG. 5 is a diagram of an alternative example of a display;
FIG. 6 is a diagram showing the experimental results of the present invention compared with other algorithms.
Detailed Description
The invention will be further described with reference to the drawings and examples, but the invention is not limited to the scope.
Example 1: as shown in fig. 1-6, according to an aspect of the embodiments of the present invention, there is provided a cooperative scheduling method of multi-task allocation and multi-robot path planning with kinematic constraint, including: inputting starting points to multiple robots in a continuous map; adding a plurality of task points into the continuous map; performing multi-task distribution on the multiple robots according to an optimization algorithm; the multiple robots plan paths conforming to kinematic constraints according to the assigned tasks. The plurality of task points form a task set, so that different task points in the task set cannot have the same position points in the continuous map, and the task points distributed by different robots cannot be the same.
Further, the multi-robot performs multi-task allocation according to an optimization algorithm, specifically: iteratively evolving the starting points and the task points of the multiple robots in the continuous map according to the overall fitness function to obtain task points and task point sequences distributed by each robot; the overall fitness function is the sum of the path fitness function and the conflict fitness function.
As shown in fig. 2 and 3, the left part of fig. 2 is a continuous map, a blank part in the continuous map represents a continuous space where the robot works, a black rectangle represents an obstacle, and a dotted rectangle represents a task point which the robot needs to execute; in order to better show the positions of robots in the continuous section, the continuous space is displayed in a gridding manner, for example, the right side of fig. 2 is the continuous space after the gridding, and two black line segments represent paths of different robots solved by the standard a. The rasterization is to extract coordinate points from a continuous map, and the coordinate points in a certain range are used as a unit cell.
Further, the iterative evolution is performed on the multiple robot starting points and task points in the continuous map according to the overall fitness function to obtain task points allocated to each robot, including:
the following is a multi-task allocation description with the task points as robot multiples:
according to a random mode, N task points in a continuous map are evenly distributed to M robots, each robot and the task point distributed by the robot form a gene, the M robot genes are used as a population individual, and P individuals are generated by P times of random distribution according to the mode;
carrying out mutation cross operation on P individuals in sequence, after the operation is finished, carrying out individual sorting according to the overall fitness value of a single individual, selecting Q individuals with the overall fitness value at the front for reservation, and expanding the population scale into P individuals according to the Q individuals reserved; iterating according to the variation cross operation process until reaching a termination condition (in the embodiment of the invention, the set termination condition is that the iteration times are 50), and reserving and outputting an individual with the minimum overall fitness value to obtain task points distributed by each robot; wherein, the whole fitness value is earlier than Q < P of the individual with small whole fitness value, and P-Q individuals are generated by randomly selecting the individual from Q individuals, so as to realize the expansion of the Q individuals into P individuals.
Wherein n=l×m; wherein L, M, N is a positive integer and L >1.
Further, the mutation crossover operation in each iteration process is specifically:
mutation operation: judging each gene in an individual, if the probability of random generation is greater than a first preset probability, randomly selecting a task point sequence number A in a task set range, changing the task point sequence number B of a robot corresponding to the randomly selected current gene into the task point sequence number A, and replacing the task point sequence number A of the robot corresponding to the gene with the task point sequence number A with the task point sequence number B; based on the operation, the task which still satisfies the distribution of different robots can not be the same after the mutation operation is performed.
Crossover operation: judging each gene in the individual, and if the probability of random generation is greater than the second preset probability, replacing all task point serial numbers allocated to robots corresponding to the current gene with all task point serial numbers allocated to robots corresponding to a randomly selected gene in the same individual; wherein the two genes involved in the substitution operation are different genes; for example, there are 1 individual in a population, and the individual has three genes: and when the front gene is the first gene and the other gene is the second gene, replacing all task point serial numbers corresponding to the first gene and all task point serial numbers corresponding to the second gene. In the embodiment of the invention, the first preset probability and the second preset probability are both 0.5.
Further, the individual overall fitness value obtaining process includes: solving the path length of the robot starting point sequentially reaching each allocated task point through a standard A-type algorithm according to the task points allocated by the robot in each individual, and obtaining the path fitness value of the single individual; determining the number of the intersections of the robot paths in each individual according to the intersection judgment conditions; according to the number of the intersection points, a conflict fitness value of robots in each individual is obtained; the conflict fitness value of the robots in each individual in turn is obtained, and the conflict fitness of the individual is obtained; and obtaining the overall fitness value of the single individual according to the path fitness value of the single individual and the conflict fitness of the single individual.
Further, the path fitness function is a sum of path lengths from the robot starting point to each task point, which is obtained by using a standard a-x algorithm.
Further, the conflict fitness function is a sum of penalty values generated by robots meeting intersection judgment conditions.
In the above, the path fitness function of the individual genes involved:conflict fitness function for individual genes: />Path fitness function for individual individuals: /> Conflict fitness function for individual individuals: />Individual individualsOverall fitness function: f (n) =f conflict (n)+F dist (n); wherein (1)>Representing the ith robot a i Path fitness value s of (2) i Representing the ith robot a i Starting point of->Representing the ith robot a i Assigned 1 st task point, m represents the i-th robot a i The total number of assigned task points; />Representing solving the ith robot a according to standard a-algorithm i A path length from the start point to the 1 st task point; />Representing the ith robot a i Conflict fitness value of +.>Representing the ith robot a i The number of the intersection points of the paths, p, represents the punishment value of the robot intersection points (in the embodiment of the invention, the punishment value is set to be 1/10 of the side length of the continuous map, and the continuous map is square); f (F) dist (n) represents the path fitness value of the nth individual, and the total number of individuals is P; f (F) conflict (n) represents the conflicting fitness value of the nth individual, and F (n) represents the overall fitness value of the nth individual. When the standard a-algorithm is adopted to obtain the path, a gridding map is adopted, and for the robot, the position of the center point of the two rear wheels is taken as the grid position.
Further, the intersection judgment condition specifically includes: an intersection point appears between any two robot paths, and the distance between the intersection point and the position point of the two robot paths before the intersection point is in a preset range, and the intersection point count is increased by 1; wherein the previous location point may be a task point or a starting point. As shown in fig. 3, for example, in the left part of fig. 3, a robot person actually collides, a dotted rectangle indicates a task that the robot needs to perform, a solid rectangle indicates the robot, and in the right part of fig. 3, a circle portion indicates an intersection between robot paths. In the embodiment of the invention, the distance between the intersection point and the position point of the two robot paths before the intersection point is set to be equal, the intersection point count is increased by 1, for the working environment with a plurality of robots, for the current robot, intersection point judgment is sequentially carried out on the current robot and other robot paths except the current robot, and the intersection point count is increased by 1 when the judgment meets the condition.
Further, the multi-robot plans a path conforming to kinematic constraint according to the allocated task, specifically: the multi-robot uses an Improved vehicle collision search algorithm (Improved CL-CBS, PCL-CBS) to plan paths conforming to kinematic constraint according to the allocated tasks; among them, CL-CBS (Car-like Conflict based search).
Further, the multi-robot uses an improved vehicle-like collision search algorithm to plan paths conforming to kinematic constraints according to the assigned tasks, comprising:
the multi-task distribution step is carried out on the multi-robots according to an optimization algorithm to obtain the task point sequence distributed by the robots;
and (3) sequentially planning all robots from the self starting point according to task points by utilizing a hybrid A algorithm to obtain paths, judging whether collision exists on the paths, storing the paths of the robots without collision, searching the first collision among the paths for the paths with collision, judging the priority condition of the robot with the first collision, storing the paths of the robots with the highest priority, adding path constraint to the robot with the highest priority, re-planning the paths, continuing searching the next collision among the paths until no collision exists among the paths of all robots, and integrating to obtain the paths of all robots conforming to the kinematic constraint kinematic constraint.
Further, the robot priority condition specifically includes: the distance between the robot with collision and the next task point is higher, and the robot with the closer distance has higher priority.
By applying the technical scheme, multiple robots can be used for planning paths uniformly, and path constraint is added according to the occurring path conflict and the robot priority. That is, according to the distance between the robot with conflict and the task point, adding constraint to the robot with longer distance, when the ith robot a is completed i After the path planning of the robot is completed and no conflict exists, the path information is added to the path set until the path planning schemes of all robots without conflict are obtained.
According to another aspect of the embodiments of the present invention, there is provided a cooperative scheduling system for multi-task allocation and multi-robot path planning with kinematic constraint, including the modules of the method of any one of the above, specifically including: a module for inputting starting points for multiple robots in a continuous map; a module for adding a plurality of task points to the continuous map; a module for multitasking the multiple robots according to an optimization algorithm; a module for planning a path conforming to the kinematic constraint by the multiple robots according to the assigned tasks.
According to another aspect of the embodiments of the present invention, there is provided a processor for running a program, wherein the program when running performs the collaborative scheduling method of multi-task allocation with kinematic constraints and multi-robot path planning described in any one of the above.
According to another aspect of the embodiment of the present invention, there is provided a computer readable storage medium, where the computer readable storage medium includes a stored program, and when the program runs, controls a device where the computer readable storage medium is located to execute the cooperative scheduling method of the multi-task allocation with kinematic constraint and the multi-robot path planning according to any one of the above.
An alternative embodiment of the invention is described in detail below:
as shown in fig. 5, in the barrier-free map of 50m×50m, the numbers of two robots are 1,2, the number of tasks is set to 4, and the sizes of the robots are 2×3m. The robot movement speed was 2m/s. It is known that when t=0 at the initial time, the robots are at the respective start positions, and the black boxes marked with numerals in the figure indicate robots present in the map, and the dotted boxes indicate task points that the robots need to complete. The path planning steps are as follows:
step 1, a group of multiple robots a= { a consisting of 2 robots are present in a given continuous map 1 ,a 2 Task set t= { T of 4 task points 1 ,t 2 ,…,t 4 Each robot has a start point and two task points, the position of the black rectangle marked with the number at the time t=0 represents the start point, and the dotted rectangle marked with the number represents the task point.
And 2, introducing an optimization algorithm to distribute tasks before planning paths by a group of robots, wherein the optimization algorithm adopts a genetic algorithm architecture, and the multi-task distribution and the multi-robot path planning have more tasks, so that the multi-robot can be distributed with multiple tasks through the optimization algorithm distribution, and the effects of reducing the path length and the system running time are achieved. Because the robot randomly distributes tasks, the defects of overlarge path length and more rescheduling times of the robot can be brought to obtain a collision-free path, and based on the method, the path length of the robot executing the tasks and the collision times possibly occurring are reduced by adopting the path fitness function and the collision fitness function, and the effect is further improved by further matching with an improved vehicle-like collision search algorithm planning method.
Step 3, planning according to the allocated tasks, wherein the executed task sequence is the sequence obtained by allocation, and carrying out unified path planning on all robots; the conflict processing is involved in the planning process, the system adds constraint according to the priority of the robot to further plan the conflict occurring in the robot path planning, and finally all conflict-free paths planned by the robot are integrated into a complete planning scheme; the conflict judging process comprises the following steps: all robots solve paths by using a mixed A-algorithm, judge whether collision exists among the robots, and judge the priority condition of the robots if collision occurs: the distance between the robot with the conflict and the next task point is higher, and the robot with the closer distance has higher priority; the rule that the hybrid a algorithm chooses the extension node is to minimize the cost value consumed by the path, minimize the time, and satisfy the kinematic constraint. As shown in fig. 4, when the robot No. 1 plans the path, the robot No. 2 collides with the robot No. 2, after judging the priorities of the robot No. 1 and the robot No. 2, the robot No. 1 re-plans the path due to the lower priority, and the robot No. 2 directly passes through the collision area.
As shown in fig. 5 below, a specific path planning procedure is given:
before path planning starts, tasks 1 and 4 are randomly allocated to the No. 1 robot, tasks 2 and 3 are randomly allocated to the No. 2 robot, and after the task planning, tasks 1 and 3 are allocated to the No. 1 robot, and tasks 2 and 4 are allocated to the No. 2 robot.
Path extension of robot a 1: at time t=0, robot a1 is at a starting point (18,2,0) (18 means that the X-axis coordinate of robot a1 is at 18m of the continuous map, 2 means that the Y-axis coordinate of robot a1 is at 2m of the continuous map, 0 means that the deflection angle of robot a1 is 0 degrees, and the same is followed by) obtaining the position of an expansion node (20,2,0) according to the hybrid a-algorithm, which expansion node (20,2,0) satisfies that it does not collide with other robots at time t=1, and so on, which expansion node is (22,2.7,5.5) at time t=2, which expansion node is (23.5,4,5.5) at time t=3, which expansion node is (24.5,5.9,4.9) at time t=4, which expansion node is (24.3,7.9,4.2) at time t=5, which expansion node is (23.4,9.5,4.2) at time t=6, which expansion node is (23, 11, -1.57), and so on at time t=7, robot a1 completes task No. 1; at this time, robot a1 continues to complete task number 3, with its expansion node at time t=8 (22.3,9,5.4), its expansion node at time t=9 (24, 10,6.1), its expansion node at time t=10 (26.1,9.7,0.5), its expansion node at time t=11 (28,8.6,0.5), its expansion node at time t=12 (29,7.5,9.5), its expansion node at time t=13 (31.5,6.3,0.5), its expansion node at time t=14 (33.3,5.3,0.5), its expansion node at time t=15 (35.2,4.1,0.5), its expansion node at time t=16 (37,3,0.5), its expansion node at time t=17 (38.7,2,0.5), its expansion node at time t=18 (40.6,0.8,0.5), its expansion node at time t=19 (42.6,0.4,6.1), its expansion node at time t=20 (43.7,0.8,5.7), its expansion node at time t=21 (45.5,2,5.7), its expansion node at time t=22 (46.4,2.5,5.7), its expansion node at time t=22 (48,3,0), and this robot a1 completes task number 3.
Then the path of the other robot a2 expands: at time t=0, robot a2 is at the starting point (14, 17,1.57), the position of the expansion node (14.7, 15,0.8) is obtained according to the hybrid a-algorithm, the expansion node (14.7, 15,0.8) thereof satisfying at time t=1 and not colliding with other robots, and so on, the expansion node (16.5, 14,0.1) thereof at time t=2, the expansion node (18.5, 14.4,5.7) thereof at time t=3, the expansion node (19.8, 16, 5) thereof at time t=4, the expansion node (19.8, 18.1,4.3) thereof at time t=5, the expansion node (18.9, 20.1,4.3) thereof at time t=6, the expansion node (18.1, 22,4.3) thereof at time t=7, the expansion node (17.3, 24,4.3) thereof at time t=8, the expansion node (17.3, 26, 5) thereof at time t=9, the expansion node (17.9, 28, 5) thereof at time t=10), the expansion node (19.8, 16, 5) thereof at time t=11), the expansion node (19.8, 18.1,4.3) thereof at time t=6, the expansion node (18.9, 20.1,4.3) thereof at time t=6, the expansion node (18.3, 22,4.3) thereof at time t=8, expansion node (17.3, 24,4.3) at time t=8, expansion node (17.3, expansion node (17.35) at time t=8, expansion node (35) at time t=8, 35) at time t=8, expansion node (35) at time t=8, 35, expansion node (35) at time t=2) and 2,3.5 At time t=18, its expansion node is (20, 39,3.14), so far robot a2 has completed task No. 2; at this time, robot a2 continues to complete task No. 4, its expansion node is (22.1, 39,3.14) at time t=19, its expansion node is (24.2, 39,3.14) at time t=20, its expansion node is (28.2, 38.3,3.8) at time t=21, its expansion node is (29.9, 37,3.8) at time t=22, its expansion node is (30.4, 36.5,3.8) at time t=23, its expansion node is (32.4, 35.8,3.1) at time t=24, its expansion node is (34.3, 36.6,2.4) at time t=25, its expansion node is (34.6, 36.9,2.3) at time t=26, and its expansion node is (34, 35,1.57) at time t=27, so that robot a2 completes task No. 4.
By applying the technical scheme, if the robots randomly select tasks, the path length and the number of re-planning times are increased, and after task planning, the robots travel according to paths explored by the system, so that the path intersection between the robots is reduced, the path length is reduced, and meanwhile, the probability of collision is reduced, so that the rapidity and the safety of the system are improved.
Further, the performance of the present invention was verified in conjunction with experimental data:
1. selecting a map of a typical graph calculation example 50m by 50m, taking the execution time as a path planning termination condition (60 s in the embodiment), randomly selecting 2 robots and 10 and 20 tasks for ten experiments respectively, wherein the running average time is 0.007,0.032 seconds respectively; the maximum finishing times were 96.657 and 203.809 seconds, respectively, as shown in table 1.
TABLE 1
2. Selecting a typical warehouse map 50m and a map 100m and 100m, and randomly selecting 8,10,14 tasks and 2 robots from the map 50m and 50m respectively; and 10, 20, 30 tasks and 5 robots are randomly selected from 100 m-100 m maps respectively, ten times of comparison experiments of the optimization algorithm, the improved vehicle-like conflict search algorithm (GA+PCL-CBS) and the random task allocation, the vehicle-like conflict search algorithm (R+CL-CBS) and the optimization algorithm and the vehicle-like conflict search (GA+CL-CBS) are carried out on each map, and the execution time is taken as a termination condition (60 s in the embodiment). As can be seen from FIG. 6 (the number of tasks on the abscissa and the time/s on the ordinate), the performance of the method of the present invention, GA+CL-CBS, is far better than that of R+CL-CBS, and the system running time can be greatly shortened by comparing the method of the present invention with GA+CL-CBS with a small increase in the maximum finishing time.
While the present invention has been described in detail with reference to the drawings, the present invention is not limited to the above embodiments, and various changes can be made without departing from the spirit of the present invention within the knowledge of those skilled in the art.
Claims (10)
1. A collaborative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraints, comprising:
inputting starting points to multiple robots in a continuous map;
adding a plurality of task points into the continuous map;
performing multi-task distribution on the multiple robots according to an optimization algorithm;
the multiple robots plan paths conforming to kinematic constraints according to the assigned tasks.
2. The collaborative scheduling method for multi-task allocation with kinematic constraints and multi-robot path planning according to claim 1, wherein the multi-robot performing multi-task allocation according to an optimization algorithm comprises: iteratively evolving the starting points and the task points of the multiple robots in the continuous map according to the overall fitness function to obtain task points distributed by each robot; the overall fitness function is the sum of the path fitness function and the conflict fitness function.
3. The collaborative scheduling method for multi-tasking with kinematic constraint and multi-robot path planning according to claim 2 wherein the path fitness function is a sum of path lengths from a robot origin to each task point using standard a-x algorithm.
4. The collaborative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraints of claim 2, wherein the collision fitness function is a sum of penalty values generated between robots in compliance with intersection judgment conditions.
5. The collaborative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraint according to claim 2, wherein iteratively evolving a plurality of robot starting points and task points in a continuous map according to an overall fitness function to obtain task points allocated to each robot comprises:
according to the random mode, N task points in the continuous map are distributed to M robots, each robot and the task point distributed by the robot form a gene, the M robot genes are used as a population individual, and P individuals are generated by P times of random distribution according to the mode;
carrying out mutation cross operation on P individuals in sequence, after the operation is finished, carrying out individual sorting according to the overall fitness value of a single individual, selecting Q individuals with the overall fitness value at the front for reservation, and expanding the population scale into P individuals according to the Q individuals reserved; and iterating according to the variation cross operation process until reaching a final condition, and reserving and outputting the individual with the minimum overall fitness value to obtain the task points distributed by each robot.
6. The collaborative scheduling method for multi-task allocation with kinematic constraint and multi-robot path planning according to claim 5, wherein the mutation crossover operation in each iteration process is specifically:
mutation operation: judging each gene in an individual, if the probability of random generation is greater than a first preset probability, randomly selecting a task point sequence number A in a task set range, changing the task point sequence number B of a robot corresponding to the randomly selected current gene into the task point sequence number A, and replacing the task point sequence number A of the robot corresponding to the gene with the task point sequence number A with the task point sequence number B;
crossover operation: judging each gene in the individual, and if the probability of random generation is greater than the second preset probability, replacing all task point serial numbers allocated to robots corresponding to the current gene with all task point serial numbers allocated to robots corresponding to a randomly selected gene in the same individual; wherein the two genes involved in the substitution operation are different genes.
7. The collaborative scheduling method for multi-tasking with kinematic constraint and multi-robot path planning according to claim 5, wherein the single individual overall fitness value acquisition process comprises:
solving the path length of the robot starting point sequentially reaching each allocated task point through a standard A-type algorithm according to the task points allocated by the robot in each individual, and obtaining the path fitness value of the single individual;
determining the number of the intersections of the robot paths in each individual according to the intersection judgment conditions; according to the number of the intersection points, a conflict fitness value of robots in each individual is obtained; the conflict fitness value of the robots in each individual in turn is obtained, and the conflict fitness of the individual is obtained;
and obtaining the overall fitness value of the single individual according to the path fitness value of the single individual and the conflict fitness of the single individual.
8. The collaborative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraints according to claim 7, wherein the intersection judgment conditions are specifically: and if the intersection point occurs between any two robot paths and the distance between the intersection point and the position point of the two robot paths before the intersection point is in a preset range, the intersection point count is increased by 1.
9. The collaborative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraints according to claim 1, wherein the multi-robot plans paths conforming to kinematic constraints according to the allocated tasks, specifically: the multiple robots use an improved vehicle-like collision search algorithm to plan paths conforming to kinematic constraints according to the assigned tasks.
10. The collaborative scheduling method for multi-task allocation with kinematic constraints and multi-robot path planning according to claim 9, wherein the multi-robot uses an improved vehicle-like collision search algorithm to plan paths that conform to kinematic constraints according to the allocated tasks, comprising:
and sequentially planning all robots from the self starting point according to task points by utilizing a mixed A algorithm to obtain paths, judging whether collision exists on the paths, storing the paths of the robots without collision, searching the first collision among the paths for the paths with collision, judging the priority condition of the robot with the first collision, storing the paths of the robots with the highest priority, adding path constraint to the robots with the highest priority to reprogram the paths, continuing searching the next collision among the paths until no collision exists among the paths of all robots, and integrating to obtain the paths of all robots conforming to the kinematic constraint.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311039966.1A CN116880507A (en) | 2023-08-17 | 2023-08-17 | Cooperative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraint |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311039966.1A CN116880507A (en) | 2023-08-17 | 2023-08-17 | Cooperative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraint |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116880507A true CN116880507A (en) | 2023-10-13 |
Family
ID=88260679
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311039966.1A Pending CN116880507A (en) | 2023-08-17 | 2023-08-17 | Cooperative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraint |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116880507A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117494919A (en) * | 2023-11-13 | 2024-02-02 | 广州力生机器人技术有限公司 | Path planning method and device based on multi-robot collaborative stacking operation |
-
2023
- 2023-08-17 CN CN202311039966.1A patent/CN116880507A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117494919A (en) * | 2023-11-13 | 2024-02-02 | 广州力生机器人技术有限公司 | Path planning method and device based on multi-robot collaborative stacking operation |
| CN117494919B (en) * | 2023-11-13 | 2024-04-19 | 广州力生机器人技术有限公司 | Path planning method and device based on multi-robot collaborative stacking operation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103413209B (en) | Many client many warehouses logistics distribution routing resources | |
| CN107895225B (en) | Multi-Agent conflict-free cooperative task allocation method | |
| CN109960544B (en) | Task parallel scheduling method based on data driving type agile satellite | |
| CN114895690B (en) | Robot path planning method and system based on dynamic weighting and thermodynamic diagram algorithm | |
| Bogyrbayeva et al. | A reinforcement learning approach for rebalancing electric vehicle sharing systems | |
| CN109409773B (en) | Dynamic planning method for earth observation resources based on contract network mechanism | |
| CN110807236A (en) | Warehouse logistics simulation system based on multiple robots | |
| Liu et al. | Multi-task-oriented vehicular crowdsensing: A deep learning approach | |
| CN107093046A (en) | Unmanned dispensing vehicle method for allocating tasks, system and unmanned dispensing vehicle | |
| CN116880507A (en) | Cooperative scheduling method for multi-task allocation and multi-robot path planning with kinematic constraint | |
| CN108268039A (en) | The paths planning method and system of mobile robot | |
| CN102903263A (en) | Method and device used for removing flight conflicts and based on packet mode | |
| CN113867358B (en) | Intelligent path planning method for multi-unmanned vehicle collaborative traversal task | |
| CN113743874A (en) | Optimal scheduling method for emergency material logistics distribution | |
| CN116720642A (en) | Method and system for optimizing path of cooperative distribution of vehicle and unmanned aerial vehicle | |
| Hani et al. | Simulation based optimization of a train maintenance facility | |
| CN113408988B (en) | Scheduling optimization method for warehouse system, electronic equipment and storage medium | |
| Xu et al. | Dynamic spare point application based coordination strategy for multi-AGV systems in a WIP warehouse environment | |
| CN113960969A (en) | Logistics storage scheduling method and system based on big data | |
| CN116185035B (en) | Unmanned cluster dynamic task allocation method and system based on improved bionic wolf clusters | |
| Wang et al. | Research on Hybrid Real-Time Picking Routing Optimization Based on Multiple Picking Stations. | |
| CN116382331A (en) | Multi-unmanned aerial vehicle rapid scheduling method, device, equipment and readable storage medium | |
| CN115344050A (en) | Stacker path planning method based on improved clustering algorithm | |
| CN115456268A (en) | Guide roller manufacturing resource optimal allocation method, device, equipment and medium | |
| CN115409448A (en) | Storage scheduling decision method and device, terminal equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |