CN109115226B - Route planning method for avoiding multi-robot conflict based on jumping point search - Google Patents
Route planning method for avoiding multi-robot conflict based on jumping point search Download PDFInfo
- Publication number
- CN109115226B CN109115226B CN201811017251.5A CN201811017251A CN109115226B CN 109115226 B CN109115226 B CN 109115226B CN 201811017251 A CN201811017251 A CN 201811017251A CN 109115226 B CN109115226 B CN 109115226B
- Authority
- CN
- China
- Prior art keywords
- node
- point
- conflict
- nodes
- agents
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 72
- 230000009191 jumping Effects 0.000 title claims abstract description 69
- 238000010845 search algorithm Methods 0.000 claims abstract description 56
- 238000001514 detection method Methods 0.000 claims abstract description 12
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 230000004888 barrier function Effects 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000013138 pruning Methods 0.000 claims description 3
- 235000008694 Humulus lupulus Nutrition 0.000 claims description 2
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 3
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 1
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0287—Control of position or course in two dimensions specially adapted to land vehicles involving a plurality of land vehicles, e.g. fleet or convoy travelling
- G05D1/0289—Control of position or course in two dimensions specially adapted to land vehicles involving a plurality of land vehicles, e.g. fleet or convoy travelling with means for avoiding collisions between vehicles
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/20—Instruments for performing navigational calculations
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Automation & Control Theory (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Aviation & Aerospace Engineering (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)
Abstract
The invention belongs to the field of artificial intelligence, and particularly relates to a multi-robot conflict avoidance path planning method based on skip point search; the method specifically comprises the following steps: 1. preprocessing a map into a data map; 2. performing path planning for all agents by using a Search framework of a CBS (Conflict-based Search) and combining an unconstrained jumping point Search algorithm; 3. and traversing the multi-branch tree to perform conflict detection, constraint addition and constrained path planning on all the agents. The route planning and searching algorithm for avoiding conflict based on the jumping point search improves the searching efficiency of the algorithm for planning all the agent conflict-free route sequences by greatly reducing the number of the expansion nodes in the route planning process, directly searches the real distance from the current node to the terminal point by using the EfftiveG table when the route is searched again in the process of processing conflict, reduces the time for planning all the agent conflict-free route sequences by the system again, and ensures that the shortest route searched is closer to the real route.
Description
Technical Field
The invention belongs to the field of artificial intelligence, and particularly relates to a multi-robot conflict avoidance path planning method based on skip point search.
Background
Currently, multiple robots are widely used in multiple fields, such as: logistics storage, search and rescue, mine exploration, mine clearance, and a wide variety of other dangerous or boring tasks, wherein the robots not only need to bypass some obstacle areas during the task, but also need to ensure that no collision occurs between the robots. As known from the prior art, each robot can be regarded as a proxy, and then rapidly planning a collision-free optimal path for multiple robots is a multi-proxy path planning problem.
When only one agent exists, the problem can be used for rapidly planning the optimal path by algorithms such as an A-star algorithm, a Dijkstra algorithm, a Floyed algorithm and the like. But when there are multiple agents, there is a problem of multi-agent resource conflict. Multi-agent cooperative routing, MAPE, is the planning of non-conflicting paths for multiple agents from a current state to a target state. At present, the multi-agent cooperation way-finding problem is widely researched in multiple fields.
For the MAPF problem, the basic idea is to plan conflict-free paths for all agents, and take some conflict resolution for the conflicting paths to resolve the conflicts. The Cooperative A-star algorithm stores unit paths in a multi-agent environment by using a storage table formed by a reservation table and time, and avoids conflicts by continuously searching the reservation table of each unit. Where "reservation table" refers to a container or a collection used to store all the proxy path sequences. However, in order to avoid deadlock, Sharon et al proposed a collision tree-based multi-agent collaborative optimal routing algorithm in 2015, which effectively solves the collisions between agents by means of a multi-branch tree processing method, returns no solution when all agents do not have a collision-free avoided path sequence set, and returns an optimal collision-free path sequence result when there is a collision-free path sequence result, but this method uses a-algorithm in the bottom level search to generate more extension nodes in the process of searching paths, which reduces the search efficiency of the system for planning collision-free path sequences of all agents, wherein an extension node is in the process of searching a single agent path sequence, the next node to be traversed starting from the starting point. Through retrieval, the existing patents on the path planning method for collision avoidance do not relate to a jump point search algorithm, which is different from the method adopted by the invention.
Disclosure of Invention
The invention aims to provide a multi-robot conflict avoidance path planning method based on hop point search, which can solve the problems of higher time complexity of a multi-agent conflict avoidance search algorithm and lower system planning conflict-free path efficiency.
A multi-robot conflict avoidance path planning method based on jumping point search specifically comprises the following steps:
and 3, traversing the multi-branch tree to perform conflict detection, constraint addition and constrained path planning on all the agents.
The multi-robot conflict avoidance path planning method based on the jumping point search specifically comprises the following steps in step 1:
step 1.1, inputting map scene information, wherein the map scene information is that the dimension of a 0-1 two-dimensional matrix containing barrier information in the actual scene of the map is M x N, the value of each node is 0 or 1, the barrier is represented by 1, and the non-barrier area is represented by 0;
step 1.2, marking all main jumping points in the map; judging whether each node has a forced neighbor in the basic direction according to a method of circular traversal from left to right and from top to bottom for all nodes in the map, if so, setting the node as a main hop and adding the node into a main hop set Primary; otherwise, continuously traversing other nodes; the forced neighbor is that when the distance from the current node to a certain point p is calculated, if the neighbor node of p is an obstacle, the walking distance from the current node to the node p is influenced, and the current node has the forced neighbor;
step 1.3, on the basis of the step 1.2, traversing the basic directions of all nodes, namely the east-west-north direction, if the current node has a main jumping point in the basic direction, the node is a direct jumping point and is added into a direct jumping point set Straight; otherwise, continuously traversing other nodes;
step 1.4, traversing the Diagonal directions of all nodes on the basis of the step 1.3, if a main jumping point or a direct jumping point exists in the Diagonal direction of the current node or in the horizontal and vertical directions of the nodes in the Diagonal direction, setting the main jumping point or the direct jumping point as the Diagonal jumping point, and adding the main jumping point or the direct jumping point into a set Diagon of the Diagonal jumping points; otherwise, continuously traversing other nodes;
step 1.5, traversing all nodes, recording the distance between each node and a main hop point or a direct hop point in the basic direction and the diagonal direction of each node, and adding distance information of a distance wall for the directions of the current nodes if the 8 directions of the current nodes do not contain distance information, namely the information of the distance between each node and the main hop point, the direct hop point or the diagonal hop point;
and step 1.6, outputting a data map, wherein each node in the data map comprises distance information of a main jumping point, a direct jumping point and a diagonal jumping point in the basic direction and the diagonal direction.
After a data map is obtained, reverse searching is firstly carried out on all agents by using a hop search algorithm, the distances from all extended nodes in a path sequence to an original end point are stored, then an optimal path is planned for all agents in a forward direction again by using the hop search algorithm, the paths of all the agents are stored on a multi-branch tree root node, and then step 2 is carried out.
The multi-robot conflict avoidance path planning method based on the jumping point search specifically comprises the following steps of:
step 2.1, adding a starting point into an open table, wherein the open table is used for storing nodes which are not traversed currently;
step 2.2, judging whether the open table is empty, if not, continuing the step 2.3, otherwise, continuing the step 2.7;
step 2.3, selecting the node with the minimum F value from the open table as the current node, removing the node from the open table, and adding the node into the close table; g represents the minimum distance from the starting node to the current node, and H represents the distance from the current node to the end point; the open table is used for storing nodes which are not expanded yet, and the close table is used for storing nodes which are expanded already; the close table refers to the traversed expansion nodes, and the subsequent steps cannot be expanded again; the extended nodes refer to all neighbor nodes to be traversed from a starting point in the process of searching a single proxy path sequence;
step 2.4, obtaining a successor node set S of the current node according to the obtained data map, wherein the successor node refers to a jumping point set which can be reached by the current node in the basic direction and the diagonal direction;
step 2.5, updating information of the successor nodes, updating information such as G values and father nodes and the like for all the successor nodes, and storing the G value of each node into an Effect G table;
step 2.6, judging whether the end point is in an open table, if so, entering step 2.7, otherwise, returning to step 2.1;
and 2.7, storing the path sequence of the planned agent in the root node of the multi-branch tree.
In the path planning method for avoiding multi-robot collision based on hop search, the method for updating the information of the subsequent node in the step 2.5 specifically comprises the following steps:
step 2.5.1, determining a successor node in the basic direction of the current node, judging the relationship between the current node and a terminal, if the terminal is not in the basic direction or the diagonal direction and is not in the middle of a key hop and the current node, taking all key hops as the successor node, otherwise, adding the terminal, finally updating a father node of the successor node as the current node, and updating a G value;
step 2.5.2, determining a successor node in the diagonal direction of the current node, if the end point is on the straight line in the diagonal direction and is not between the current node and the key jump point, adding the next jump point of the current node as the successor node into an open table, otherwise, adding the end point into the open table, finally updating a father node of the successor node as the current node, and updating a G value;
and 2.5.3, determining a successor node between the basic direction of the current node and the diagonal direction, if the end point is in the middle of the basic direction and the diagonal direction, judging whether the successor node in the diagonal direction has a forced node in the basic direction, if so, adding a skip point of the successor node as the successor node into an open table, finally updating a father node of the successor node as the current node, and updating the value G.
The multi-robot conflict avoidance path planning method based on the jumping point search specifically comprises the following steps of:
step 3.1, traversing the multi-branch tree root node to perform conflict detection on the agent path sequence, taking every two agents as a group when detecting path conflict, respectively performing vertex conflict and edge conflict detection on paths between a group of agents by using time nodes, and if conflict exists, adding vertex constraint and edge constraint to the paths, wherein the vertex constraint form is (a)i,(vx,vy) T), representing agent aiCan not be in position (v) at time tx,vy) At least one of (1) and (b); while an edge conflict represents two agents that cannot act in the same direction, i.e., agent aiAt time t, from position m to position n, and agent ajWhen the position n reaches the position m and the two agents conflict with each other, the two agents are added with edge constraint, namely (a)iN, t +1) and (a)jM, t +1), after detecting the conflict and generating the constraint, replanning the path for the agent with the conflict path by using a constrained hop search algorithm, wherein the algorithm specifically comprises the following steps;
3.2, aiming at the agents generating the conflict, firstly adding the initial points of the agents into an open table;
3.3, judging whether the open table is empty, if so, performing the step 3.10, otherwise, continuing the step 3.4;
step 3.4, selecting the node with the minimum F value from the open table as the current node, removing the node from the open table, and adding the node into the close table, wherein the H value is obtained by directly traversing the Effect G table, so that the time of planning all the conflict-free path sequences of the agents by the system is reduced, and the shortest path searched is closer to a real path;
step 3.5, obtaining a subsequent set S of the current node according to the data map;
step 3.6, constraint condition judgment, namely judging whether each subsequent node meets the constraint condition of the agent, if so, not adding an open table and continuously judging the subsequent node; if not, continuing to step 3.7;
step 3.7, updating information such as G values and father nodes and the like for all subsequent nodes, wherein the specific method is the same as that in the step 2.5;
step 3.8, adding a waiting state to the current node with the subsequent node to prevent the conflict-free paths from being planned only by waiting among the agents with conflict, thereby not only effectively solving the conflict among the agents, but also providing a plurality of choices of the conflict-free paths for the agents, if the priorities among the clients are different, returning a path sequence set that the agent with higher priority passes through the conflict point preferentially when the conflict occurs; such as: if the agent A has higher priority than the agent B, namely when the agent A conflicts with the agent B, the agent A needs to pass through first, the agent B needs to wait, and finally a conflict-free path sequence set with which the agent A passes through conflict points first is selected; on the contrary, if the agent B has higher priority than the agent A, a conflict-free path sequence set with the agent B passing through the conflict point first is finally selected; the invention does not consider the priority problem among the agents, and the final result returns the path sequence set with the minimum sum of the path sequence time of all the agents;
step 3.9, judging whether the end point is in an open table, if so, entering step 3.10, otherwise, returning to step 3.4;
step 3.10, storing the path sequence of the agent in the current node of the multi-branch tree, wherein the path sequence is a leaf node of the multi-branch tree;
and 3.11, after the steps 3.2-3.10 are completed, traversing all the nodes of the multi-branch tree, wherein each branch node of the multi-branch tree is a leaf node, comparing the sum of the path time of all the agents of all the leaf nodes, and finally returning a path sequence set with the minimum sum of the time.
In the path planning method for avoiding multi-robot collision based on jumping point search, the constraint condition judgment method in step 3.6 specifically comprises the following steps:
step 3.6.1, compare tSubsequent operationAnd tConstrainingIf t isSubsequent operation＜tConstrainingIf yes, returning to false; i.e. tSubsequent operationTime, t, representing the succeeding nodeConstrainingTime representing the point of constraint;
step 3.6.2, if tSubsequent operation＝tConstrainingContinuously judging the agent ID and coordinate information of the subsequent node and the constraint point, if all the agent ID and coordinate information are the same, returning true, otherwise, returning false;
step 3.6.3, if tSubsequent operation＝tConstrainingFirstly, calculating the coordinate information of a coordinate point ConstrainT between the current node and the subsequent node at the same moment according to the time of the ConstrainT point, then continuously judging the coordinate information of the ConstrainT and the ConstrainT point and the agent ID information, if all the coordinate information and the agent ID information are the same, returning true, otherwise, returning false.
The search algorithm based on the CBS is mainly divided into two layers of search algorithms including a bottom layer search algorithm and a top layer search algorithm, wherein the bottom layer search algorithm refers to a path sequence result for planning a single agent, and the number of nodes expanded in the search process can be effectively reduced; the top-level search algorithm can avoid detecting conflicts and generate corresponding constraints, and finally, a conflict-free path sequence is planned.
The detection conflict of the top-level search algorithm in the search algorithm based on the CBS comprises the vertex conflict and the edge conflict between detection agent path sequences, wherein the vertex conflict means that two agents cannot be positioned at the same position at the same time, and the added vertex constraint means that the constraint form is (a)i,(vx,vy) T), representing agent aiAt time tCan not be in position plane coordinate (v)x,vy) At least one of (1) and (b); while an edge conflict represents two agents that cannot act in the same direction, i.e., agent aiAt time t, from position m to position n, and agent ajWhen the position n reaches the position m and the two agents conflict with each other, the two agents are added with edge constraint, namely (a)iN, t +1) and (a)j,m,t+1)。
The search algorithm based on the CBS comprises a tree building process, firstly, a root node of a multi-branch tree is created, path sequences of all agents planned by the skip point search algorithm for the first time are stored in the root node, and the current node of the multi-branch tree is the root node; and then when the conflict between the agents is detected, if the conflict is detected, newly creating a multi-branch tree node which is used as a child node of the current tree node, wherein the current tree node is the child node, then storing the re-planned agent path sequence result in the current tree node, and circulating the process until the paths traversed among all the agents have no conflict, namely the current tree node is a leaf node, and the completion of the traversal means that the tree building process of the multi-branch tree is finished.
The jumping point search algorithm is a path planning method by pruning part of invalid redundant extension nodes, and the jumping point refers to an effective extension node which is selectively extended from the current point in the map.
The invention has the beneficial effects that:
the invention provides a path planning and searching algorithm for avoiding conflict based on hop searching, which improves the searching efficiency of the algorithm for planning all agent conflict-free path sequences by a system by greatly reducing the number of expansion nodes in the path planning process. The method for judging whether the successor node meets the constraint according to the time node and the method for adding the waiting state to the current node with the extensible successor node, which are provided by the invention, not only can effectively solve the conflict between the agents, but also can provide a plurality of choices of conflict-free path sequence sets for the agents, so that the agents can select a conflict-free path according to different requirements.
Drawings
FIG. 1 is a flow chart of a multi-agent conflict path planning algorithm based on a skip point search;
FIG. 2 is a flow chart of pre-processing map data information for a multi-agent conflict path planning method based on a skip point search;
FIG. 3 is an example diagram of pre-processing map data information for a multi-agent conflict path planning method based on a skip point search;
FIG. 4 is a flow chart of a path planning algorithm based on an unconstrained jumping point search algorithm;
FIG. 5 is a flow chart of a method for updating information of a successor node based on an unconstrained jumping point search algorithm;
FIG. 6 is an example diagram of a path planning algorithm based on an unconstrained hop search algorithm;
FIG. 7 is a flow chart of path planning based on a constrained jump point search algorithm;
FIG. 8 is a flowchart of a method for determining whether a constraint condition based on a constrained skip point search algorithm is satisfied;
fig. 9 is a diagram of an example of path planning based on a constrained hop search algorithm.
Detailed Description
The invention is further described below with reference to the accompanying drawings:
a multi-agent conflict avoidance path planning method based on jump point search JPS (jump point search) aims to solve the problems of higher time complexity of a multi-agent conflict avoidance search algorithm and lower system planning conflict-free path efficiency. The main idea is to rapidly and accurately plan a collision-free path for multiple agents and finally output an optimal path with the minimum sum of path time of all agents by utilizing a CBS (Conflict-based Search) Search framework and combining a bottom layer hop Search algorithm according to input map scene information.
The CBS-based search algorithm is mainly divided into two layers of search algorithms, including a bottom layer search algorithm and a top layer search algorithm, wherein the bottom layer search algorithm refers to a path sequence result for planning a single agent, and the number of expanded nodes in the search process can be effectively reduced; and the top-level search algorithm aims to avoid detecting conflicts, generate corresponding constraints and finally plan a conflict-free path sequence.
Wherein the detection conflict of the aforementioned top-level search algorithm is further characterized by: it includes detecting vertex and edge conflicts between the sequence of proxy paths. The vertex conflict means that two agents cannot be in the same position at the same time, and the added vertex constraint means that the constraint form is (a)i,(vx,vy) T), representing agent aiCannot be in the position plane coordinate (v) at time tx,vy) To (3). While the aforementioned edge conflict represents that two agents cannot proceed in the same direction, i.e., agent aiAt time t, from position m to position n, and agent ajWhen the position n reaches the position m and the two agents conflict with each other, the two agents are added with edge constraint, namely (a)iN, t +1) and (a)j,m,t+1)。
The aforementioned CBS-based search algorithm is further characterized by: the CBS-based search algorithm comprises a tree building process, firstly, a root node of a multi-branch tree is built, path sequences of all agents planned by the skip point search algorithm for the first time are stored in the root node, and the current node of the multi-branch tree is the root node; and then when the conflict between the agents is detected, if the conflict is detected, a multi-branch tree node is newly created and used as a child node of the current tree node, the current tree node is the child node, and then the replanned agent path sequence result is stored in the current tree node. And circulating the process until the paths traversed to all the agents have no conflict, namely the current tree node is a leaf node, and the completion of the traversal means that the tree building process of the multi-branch tree is finished.
The aforementioned skip point search algorithm is a path planning method by pruning off partially invalid redundant extension nodes. The skip point refers to an effective extension node that is selectively extended from a current point in the map.
Fig. 1 is an overall implementation flow of a path planning method for collision avoidance based on a skip point search algorithm. Firstly, a map scene is processed as a data map, that is, a jump point which can be directly and critically located in the basic direction and the diagonal direction of each node in the map is processed for each node in the map, that is, a jump point which can be reached in the basic direction and the diagonal direction of each node when each node is used as a current node. Secondly, an optimal path is drawn by using a jumping point search algorithm according to the requirements of a starting point and an end point of each agent, when the path is planned for the first time, the starting point and the end point are searched reversely, the real distance between a node in the path sequence process and the end point is recorded, namely an Effect G table, so that the Effect G table can be directly consulted when other nodes of the multi-branch tree are subjected to constrained search, the path planning time is further reduced, and the path sequence is stored in the root nodes of the multi-branch tree. Then, traversing the root nodes of the multi-branch tree, detecting whether all the agent path sequences in the root nodes contain vertex conflicts and edge conflicts, if the agent path sequences contain conflict path sequences, generating new multi-branch tree nodes, adding corresponding constraints to each agent, and finally planning an optimal path for the agents by using a constrained hop search algorithm. And finally returning the time series and shortest path series results of all the agents until the leaf nodes are traversed, namely no conflict exists.
Fig. 2 shows a flow chart for preprocessing a data map. The method specifically comprises the following steps:
step 1.1, inputting map scene information, wherein the map scene information is that the dimension of a 0-1 two-dimensional matrix containing barrier information in the actual scene of the map is M x N, the value of each node is 0 or 1, the barrier is represented by 1, and the non-barrier area is represented by 0;
step 1.2, marking all main jumping points in the map; judging whether each node has a forced neighbor in the basic direction according to a method of circular traversal from left to right and from top to bottom for all nodes in the map, if so, setting the node as a main hop and adding the node into a main hop set Primary; otherwise, continuously traversing other nodes; the forced neighbor is that when the distance from the current node to a certain point p is calculated, if the neighbor node of p is an obstacle, the walking distance from the current node to the node p is influenced, and the current node has the forced neighbor;
step 1.3, on the basis of the step 1.2, traversing the basic directions of all nodes, namely the east-west-north direction, if the current node has a main jumping point in the basic direction, the node is a direct jumping point and is added into a direct jumping point set Straight; otherwise, continuously traversing other nodes;
step 1.4, traversing the Diagonal directions of all nodes on the basis of the step 1.3, if a main jumping point or a direct jumping point exists in the Diagonal direction of the current node or in the horizontal and vertical directions of the nodes in the Diagonal direction, setting the main jumping point or the direct jumping point as the Diagonal jumping point, and adding the main jumping point or the direct jumping point into a set Diagon of the Diagonal jumping points; otherwise, continuously traversing other nodes;
step 1.5, traversing all nodes, recording the distance between each node and a main hop point or a direct hop point in the basic direction and the diagonal direction of each node, and adding distance information of a distance wall for the directions of the current nodes if the 8 directions of the current nodes do not contain distance information, namely the information of the distance between each node and the main hop point, the direct hop point or the diagonal hop point;
and step 1.6, outputting a data map, wherein each node in the data map comprises distance information of a main jumping point, a direct jumping point and a diagonal jumping point in the basic direction and the diagonal direction.
In fig. 3, a 6 × 6 map scene is shown, which includes a certain obstacle, and the distance values from the base direction and the diagonal direction to the key points of each node are marked according to the above steps, and all data information of the data map is shown in the map.
Fig. 4 shows a flow chart for planning optimal paths for all agents using the hop search algorithm. The method specifically comprises the following steps:
step 2.1, adding the starting point into an open table, wherein the open table is used for storing nodes which are not traversed currently, and the planned path can be more accurate by the reverse searching process of the starting point and the end point in the first step of using skip point searching;
step 2.2, judging whether the open table is empty, if not, continuing the step 2.3, otherwise, continuing the step 2.7;
step 2.3, selecting the node with the minimum F value from the open table as the current node, removing the node from the open table, and adding the node into the close table; g represents the minimum distance from the starting node to the current node, and H represents the distance from the current node to the end point; the open table is used for storing nodes which are not expanded yet, and the close table is used for storing nodes which are expanded already; the close table refers to the traversed expansion nodes, and the subsequent steps cannot be expanded again; the extended nodes refer to all neighbor nodes to be traversed from a starting point in the process of searching a single proxy path sequence;
step 2.4, obtaining a successor node set S of the current node according to the obtained data map, wherein the successor node refers to a jumping point set which can be reached by the current node in the basic direction and the diagonal direction;
step 2.5, updating information of the successor nodes, updating information such as G values and father nodes and the like for all the successor nodes, and storing the G value of each node into an Effect G table;
step 2.6, judging whether the end point is in an open table, if so, entering step 2.7, otherwise, returning to step 2.1;
and 2.7, storing the path sequence of the planned agent in the root node of the multi-branch tree.
Fig. 5 is a method for updating information of a successor node included in step 2.5 in fig. 4, which mainly extends the successor node from three directions, determines its successor node according to the basic direction of the current node in the basic direction, and determines the relationship between the current node and the end point, if the end point is not in the middle of the key jump point and the current node in the basic direction or diagonal direction, all the key jump points are used as the successor node, otherwise, the end point is added, and finally the parent node and the G value of the successor node are updated. And if the end point is on the straight line in the diagonal direction and is not between the current node and the key jump point, adding the next jump point of the current node as a subsequent node into the open table, and otherwise, adding the end point into the open table. If the end point is in the middle of the basic direction and the diagonal direction, judging whether a subsequent node in the diagonal direction has a forced node in the basic direction, if so, adding the skip point of the subsequent node as the subsequent node into an open table, and updating a parent node and a G value of the subsequent node.
In fig. 6, a 6 × 6 map scene is simulated, and agent a and agent B have respective starting points and end points, and an optimal path sequence is planned according to the obtained data map and an unconstrained jumping point search algorithm.
Fig. 7 illustrates a flow chart for planning a collision-free path using a hop search under constraint when a collision is detected between the proxy path sequences. The method specifically comprises the following steps:
step 3.1, traversing the multi-branch tree root node to perform conflict detection on the agent path sequence, taking every two agents as a group when detecting path conflict, respectively performing vertex conflict and edge conflict detection on paths between a group of agents by using time nodes, and if conflict exists, adding vertex constraint and edge constraint to the paths, wherein the vertex constraint form is (a)i,(vx,vy) T), representing agent aiCan not be in position (v) at time tx,vy) At least one of (1) and (b); while an edge conflict represents two agents that cannot act in the same direction, i.e., agent aiAt time t, from position m to position n, and agent ajWhen the position n reaches the position m and the two agents conflict with each other, the two agents are added with edge constraint, namely (a)iN, t +1) and (a)jM, t +1), after detecting a conflict and generating a constraint, utilizing the constraint on the agents of the conflicting pathThe jumping point searching algorithm carries out path re-planning, and the algorithm is concretely as the following steps;
3.2, aiming at the agents generating the conflict, firstly adding the initial points of the agents into an open table;
3.3, judging whether the open table is empty, if so, performing the step 3.10, otherwise, continuing the step 3.4;
step 3.4, selecting the node with the minimum F value from the open table as the current node, removing the node from the open table, and adding the node into the close table, wherein the H value can be obtained by directly traversing the Effect G table, so that the time of planning all the conflict-free path sequences of the agents by the system is reduced, and the shortest path searched is closer to a real path;
step 3.5, obtaining a subsequent set S of the current node according to the data map;
step 3.6, constraint condition judgment, namely judging whether each subsequent node meets the constraint condition of the agent, if so, not adding an open table and continuously judging the subsequent node; if not, continuing to step 3.7;
step 3.7, updating information such as G values and father nodes and the like for all subsequent nodes, wherein the specific method is the same as that in the step 2.5;
step 3.8, adding a waiting state to the current node with the subsequent node to prevent the conflict-free paths from being planned only by waiting among the agents with conflict, thereby not only effectively solving the conflict among the agents, but also providing a plurality of choices of the conflict-free paths for the agents, if the priorities among the clients are different, returning a path sequence set that the agent with higher priority passes through the conflict point preferentially when the conflict occurs; such as: if the agent A has higher priority than the agent B, namely when the agent A conflicts with the agent B, the agent A needs to pass through first, the agent B needs to wait, and finally a conflict-free path sequence set with which the agent A passes through conflict points first is selected; on the contrary, if the agent B has higher priority than the agent A, a conflict-free path sequence set with the agent B passing through the conflict point first is finally selected; the invention does not consider the priority problem among the agents, and the final result returns the path sequence set with the minimum sum of the path sequence time of all the agents;
step 3.9, judging whether the end point is in an open table, if so, entering step 3.10, otherwise, returning to step 3.4;
step 3.10, storing the path sequence of the agent in the current node of the multi-branch tree, wherein the path sequence is a leaf node of the multi-branch tree;
and 3.11, after the steps 3.2-3.10 are completed, traversing all the nodes of the multi-branch tree, wherein each branch node of the multi-branch tree is a leaf node, comparing the sum of the path time of all the agents of all the leaf nodes, and finally returning a path sequence set with the minimum sum of the time.
Fig. 8 is a flowchart of a constraint condition determining method of the multi-agent collision avoiding method based on the skip point search algorithm included in step 3.6 of fig. 7, and mainly determines whether the constraint condition is satisfied according to a time node. If the time of the current jumping point is less than the time of the constraint point, returning false; if the time of the current jumping point is the time of the constraint point, continuously judging the agent ID and the coordinate, if the time of the current jumping point is the time of the constraint point, returning true, and otherwise, returning false; if the time of the current jump point is larger than the time of the constraint point, firstly calculating the coordinate of the current node at the same time according to the time of the constraint point, then continuously judging the coordinate and the ID of the intelligent agent, if the coordinate and the ID of the intelligent agent are the same, returning true, otherwise, returning false. In fig. 9, agent a and agent B have vertex collisions at node D (at time t 3, agent a and agent B both appear at position D), then by letting agent B wait at point E for a grid of time, agent a and agent B do not collide at node D and the path is better.
Claims (9)
1. A multi-robot conflict avoidance path planning method based on hop search is characterized by comprising the following steps:
step 1, preprocessing a map into a data map;
step 2, utilizing a Search framework of CBS (Conflict-based Search) and combining an unconstrained jumping point Search algorithm to plan paths for all agents;
the step 2 specifically comprises the following steps:
step 2.1, adding a starting point into an open table, wherein the open table is used for storing nodes which are not traversed currently;
step 2.2, judging whether the open table is empty, if not, continuing the step 2.3, otherwise, continuing the step 2.7;
step 2.3, selecting the node with the minimum F value from the open table as the current node, removing the node from the open table, and adding the node into the close table; g represents the minimum distance from the starting node to the current node, and H represents the distance from the current node to the end point; the open table is used for storing nodes which are not expanded yet, and the close table is used for storing nodes which are expanded already; the close table refers to the traversed expansion nodes, and the subsequent steps cannot be expanded again; the extended nodes refer to all neighbor nodes to be traversed from a starting point in the process of searching a single proxy path sequence;
step 2.4, obtaining a successor node set S of the current node according to the obtained data map, wherein the successor node refers to a jumping point set which can be reached by the current node in the basic direction and the diagonal direction;
step 2.5, updating information of the successor nodes, updating information such as G values and father nodes and the like for all the successor nodes, and storing the G value of each node into an Effect G table;
step 2.6, judging whether the end point is in an open table, if so, entering step 2.7, otherwise, returning to step 2.1;
step 2.7, storing the planned path sequence of the agent in a multi-branch tree root node;
and 3, traversing the multi-branch tree to perform conflict detection, constraint addition and constrained path planning on all the agents.
2. The route planning method for multi-robot collision avoidance based on hop search as claimed in claim 1, wherein said step 1 specifically comprises the steps of:
step 1.1, inputting map scene information, wherein the map scene information is that the dimension of a 0-1 two-dimensional matrix containing barrier information in the actual scene of the map is M x N, the value of each node is 0 or 1, the barrier is represented by 1, and the non-barrier area is represented by 0;
step 1.2, marking all main jumping points in the map; judging whether each node has a forced neighbor in the basic direction according to a method of circular traversal from left to right and from top to bottom for all nodes in the map, if so, setting the node as a main hop and adding the node into a main hop set Primary; otherwise, continuously traversing other nodes; wherein, forced neighbor means that when the distance from the current node to a certain point p is calculated, if the neighbor node of p is an obstacle, the walking distance from the current node to the node p is influenced, and the current node has forced neighbor;
step 1.3, on the basis of the step 1.2, traversing the basic directions of all nodes, namely the east-west-north direction, if the current node has a main jumping point in the basic direction, the node is a direct jumping point and is added into a direct jumping point set Straight; otherwise, continuously traversing other nodes;
step 1.4, traversing the Diagonal directions of all nodes on the basis of the step 1.3, if a main jumping point or a direct jumping point exists in the Diagonal direction of the current node or in the horizontal and vertical directions of the nodes in the Diagonal direction, setting the main jumping point or the direct jumping point as the Diagonal jumping point, and adding the main jumping point or the direct jumping point into a set Diagon of the Diagonal jumping points; otherwise, continuously traversing other nodes;
step 1.5, traversing all nodes, recording the distance between each node and a main hop point or a direct hop point in the basic direction and the diagonal direction of each node, and adding distance information of a distance wall for the directions of the current nodes if the 8 directions of the current nodes do not contain distance information, namely the information of the distance between each node and the main hop point, the direct hop point or the diagonal hop point;
and step 1.6, outputting a data map, wherein each node in the data map comprises distance information of a main jumping point, a direct jumping point and a diagonal jumping point in the basic direction and the diagonal direction.
3. The route planning method for avoiding multi-robot collision based on hop search as claimed in claim 1, wherein after the data map is obtained, reverse search is performed on all agents by using a hop search algorithm and the distances from all extended nodes in the route sequence to the original end point are stored, then the optimal route is planned forward again for all agents by using the hop search algorithm, the routes of all agents are stored on the multi-branch tree root node, and then step 2 is performed.
4. The route planning method for multi-robot collision avoidance based on hop search as claimed in claim 1, wherein the method for updating the information of the subsequent node in step 2.5 specifically comprises the following steps:
step 2.5.1, determining a successor node in the basic direction of the current node, judging the relationship between the current node and a terminal, if the terminal is not in the basic direction or the diagonal direction and is not in the middle of a key hop and the current node, taking all key hops as the successor node, otherwise, adding the terminal, finally updating a father node of the successor node as the current node, and updating a G value;
step 2.5.2, determining a successor node in the diagonal direction of the current node, if the end point is on the straight line in the diagonal direction and is not between the current node and the key jump point, adding the next jump point of the current node as the successor node into an open table, otherwise, adding the end point into the open table, finally updating a father node of the successor node as the current node, and updating a G value;
and 2.5.3, determining a successor node between the basic direction of the current node and the diagonal direction, if the end point is in the middle of the basic direction and the diagonal direction, judging whether the successor node in the diagonal direction has a forced node in the basic direction, if so, adding a skip point of the successor node as the successor node into an open table, finally updating a father node of the successor node as the current node, and updating the value G.
5. The route planning method for multi-robot collision avoidance based on hop search as claimed in claim 1, wherein said step 3 specifically comprises the steps of:
step 3.1, traversing the multi-branch tree root node to perform conflict detection on the agent path sequence, taking every two agents as a group when detecting path conflict, respectively performing vertex conflict and edge conflict detection on paths between a group of agents by using time nodes, and if conflict exists, adding vertex constraint and edge constraint to the paths, wherein the vertex constraint form is (a)i,(vx,vy) T), representing agent aiCan not be in position (v) at time tx,vy) At least one of (1) and (b); while an edge conflict represents two agents that cannot act in the same direction, i.e., agent aiAt time t, from position m to position n, and agent ajWhen the position n reaches the position m and the two agents conflict with each other, the two agents are added with edge constraint, namely (a)iN, t +1) and (a)jM, t +1), after detecting the conflict and generating the constraint, replanning the path for the agent with the conflict path by using a constrained hop search algorithm, wherein the algorithm specifically comprises the following steps;
3.2, aiming at the agents generating the conflict, firstly adding the initial points of the agents into an open table;
3.3, judging whether the open table is empty, if so, performing the step 3.10, otherwise, continuing the step 3.4;
step 3.4, selecting the node with the minimum F value from the open table as the current node, removing the node from the open table, and adding the node into the close table, wherein the H value is obtained by directly traversing the Effect G table;
step 3.5, obtaining a subsequent set S of the current node according to the data map;
step 3.6, constraint condition judgment, namely judging whether each subsequent node meets the constraint condition of the agent, if so, not adding an open table and continuously judging the subsequent node; if not, continuing to step 3.7;
step 3.7, updating information such as G values and father nodes and the like for all subsequent nodes, wherein the specific method is the same as that in the step 2.5;
step 3.8, adding a waiting state to the current node with the subsequent node;
step 3.9, judging whether the end point is in an open table, if so, entering step 3.10, otherwise, returning to step 3.4;
step 3.10, storing the path sequence of the agent in the current node of the multi-branch tree, wherein the path sequence is a leaf node of the multi-branch tree;
and 3.11, after the steps 3.2-3.10 are completed, traversing all the nodes of the multi-branch tree, wherein each branch node of the multi-branch tree is a leaf node, comparing the sum of the path time of all the agents of all the leaf nodes, and finally returning a path sequence set with the minimum sum of the time.
6. The method for path planning for multi-robot collision avoidance based on hop search of claim 5, wherein the method for determining constraint conditions in step 3.6 specifically comprises the following steps:
step 3.6.1, compare tSubsequent operationAnd tConstrainingIf t isSubsequent operation＜tConstrainingIf yes, returning to false; i.e. tSubsequent operationTime, t, representing the succeeding nodeConstrainingTime representing the point of constraint;
step 3.6.2, if tSubsequent operation＝tConstrainingContinuously judging the agent ID and coordinate information of the subsequent node and the constraint point, if all the agent ID and coordinate information are the same, returning true, otherwise, returning false;
step 3.6.3, if tSubsequent operation＝tConstrainingFirstly, calculating the coordinate information of a coordinate point ConstrainT between the current node and the subsequent node at the same moment according to the time of the ConstrainT point, then continuously judging the coordinate information of the ConstrainT and the ConstrainT point and the agent ID information, if all the coordinate information and the agent ID information are the same, returning true, otherwise, returning false.
7. The method for path planning for multi-robot collision avoidance based on hop search as claimed in claim 1, wherein: the CBS-based search algorithm is mainly divided into two layers of search algorithms, including a bottom layer search algorithm and a top layer search algorithm, wherein the bottom layer search algorithm refers to a path sequence result for planning a single agent, the top layer search algorithm can avoid detecting conflicts and generate corresponding constraints, and finally a conflict-free path sequence is planned.
8. The path planning method for multi-robot collision avoidance based on hop search of claim 1, wherein the search algorithm based on CBS comprises a tree building process, first creating a root node of a multi-branch tree, and storing the path sequences of all agents planned by the hop search algorithm for the first time into the root node, where the current node of the multi-branch tree is the root node; and then when the conflict between the agents is detected, if the conflict is detected, newly creating a multi-branch tree node which is used as a child node of the current tree node, wherein the current tree node is the child node, then storing the re-planned agent path sequence result in the current tree node, and circulating the process until the paths traversed among all the agents have no conflict, namely the current tree node is a leaf node, and the completion of the traversal means that the tree building process of the multi-branch tree is finished.
9. The method of claim 1, wherein the hop search algorithm is a path planning method by pruning out partially invalid redundant extension nodes, and the hop is a valid extension node that is selectively extended from a current point in the map.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811017251.5A CN109115226B (en) | 2018-09-01 | 2018-09-01 | Route planning method for avoiding multi-robot conflict based on jumping point search |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811017251.5A CN109115226B (en) | 2018-09-01 | 2018-09-01 | Route planning method for avoiding multi-robot conflict based on jumping point search |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109115226A CN109115226A (en) | 2019-01-01 |
| CN109115226B true CN109115226B (en) | 2021-11-19 |
Family
ID=64861717
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811017251.5A Active CN109115226B (en) | 2018-09-01 | 2018-09-01 | Route planning method for avoiding multi-robot conflict based on jumping point search |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109115226B (en) |
Families Citing this family (33)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110006429A (en) * | 2019-03-20 | 2019-07-12 | 智慧航海（青岛）科技有限公司 | A kind of unmanned boat path planning method based on depth optimization |
| CN110006430B (en) * | 2019-03-26 | 2021-05-04 | 智慧航海（青岛）科技有限公司 | Optimization method of track planning algorithm |
| CN111641729B (en) * | 2019-05-23 | 2021-03-30 | 北京航空航天大学 | Inter-domain path identification prefix conflict detection and decomposition method based on prefix tree |
| CN110428688B (en) * | 2019-08-09 | 2021-03-02 | 郑州铁路职业技术学院 | Rail transit vehicle overhauls real standard system of simulation based on virtual scene |
| CN110530390A (en) * | 2019-09-16 | 2019-12-03 | 哈尔滨工程大学 | A kind of non-particle vehicle path planning method under narrow environment |
| CN110975290B (en) * | 2019-11-20 | 2023-09-05 | 中国人民解放军国防科技大学 | Path planning method and system based on pattern database |
| CN111104471B (en) * | 2019-11-20 | 2023-03-31 | 中国人民解放军国防科技大学 | Mode database information compression method and system based on jumping point path search |
| CN110967015B (en) * | 2019-11-20 | 2021-11-12 | 中国人民解放军国防科技大学 | Path planning method and system |
| CN110975291B (en) * | 2019-11-20 | 2023-11-10 | 中国人民解放军国防科技大学 | Path extraction method and system |
| CN110975288B (en) * | 2019-11-20 | 2023-08-29 | 中国人民解放军国防科技大学 | Geometric container data compression method and system based on jump point path search |
| CN110849373B (en) * | 2019-11-28 | 2023-07-21 | 中国航空工业集团公司沈阳飞机设计研究所 | Real-time route re-planning method for man-machine |
| CN111238519B (en) * | 2020-01-06 | 2022-05-03 | 华侨大学 | Multi-unmanned vehicle road finding method based on topological map and conflict elimination strategy |
| CN111510427B (en) * | 2020-03-06 | 2022-02-11 | 杜晓楠 | Method for mitigating path creation attack in I2P network system, computer-readable storage medium, and I2P network system |
| CN111422741B (en) * | 2020-03-24 | 2022-02-11 | 苏州西弗智能科技有限公司 | Method for planning movement path of bridge crane |
| CN111736524A (en) * | 2020-07-17 | 2020-10-02 | 北京布科思科技有限公司 | Multi-robot scheduling method, device and equipment based on time and space |
| CN111829526B (en) * | 2020-07-23 | 2022-05-10 | 中国人民解放军国防科技大学 | Distance map reconstruction and jumping point path planning method based on anti-collision radius |
| CN112229419B (en) * | 2020-09-30 | 2023-02-17 | 隶元科技发展(山东)有限公司 | Dynamic path planning navigation method and system |
| CN112731929A (en) * | 2020-12-23 | 2021-04-30 | 浙江大学 | Ackerman model-based mobile robot obstacle avoidance path planning method |
| CN112835364B (en) * | 2020-12-30 | 2022-05-13 | 浙江大学 | Multi-robot path planning method based on conflict detection |
| CN113074728B (en) * | 2021-03-05 | 2022-07-22 | 北京大学 | Multi-AGV path planning method based on jumping point routing and collaborative obstacle avoidance |
| CN112947475A (en) * | 2021-03-22 | 2021-06-11 | 山东大学 | Laser navigation forklift type AGV vehicle-mounted system and method |
| CN112915541B (en) * | 2021-04-07 | 2022-09-23 | 腾讯科技（深圳）有限公司 | Jumping point searching method, device, equipment and storage medium |
| CN113311829B (en) * | 2021-05-11 | 2022-04-08 | 北京理工大学 | Multi-robot path planning method based on dynamic time window conflict search |
| CN113505922B (en) * | 2021-07-07 | 2023-08-01 | 福州大学 | Truss assembly sequence planning method based on parallel tree search |
| CN113568411B (en) * | 2021-07-30 | 2023-10-27 | 深圳市新盒科技有限公司 | Machine trolley scheduling system for track map |
| CN113804197A (en) * | 2021-07-30 | 2021-12-17 | 深圳市新盒科技有限公司 | Multi-machine trolley path planning method based on complex track map |
| CN113515129B (en) * | 2021-08-23 | 2022-02-11 | 哈尔滨理工大学 | Bidirectional skip point search unmanned vehicle path planning method based on boundary search |
| CN113885567B (en) * | 2021-10-22 | 2023-08-04 | 北京理工大学 | Collaborative path planning method for multiple unmanned aerial vehicles based on conflict search |
| CN114326621B (en) * | 2021-12-25 | 2023-11-14 | 长安大学 | Group intelligent airport consignment car scheduling method and system based on layered architecture |
| CN114564023B (en) * | 2022-03-11 | 2022-11-08 | 哈尔滨理工大学 | Jumping point search path planning method under dynamic scene |
| CN114705196B (en) * | 2022-06-07 | 2022-08-30 | 湖南大学 | Self-adaptive heuristic global path planning method and system for robot |
| CN115507858B (en) * | 2022-11-24 | 2023-03-03 | 青岛中德智能技术研究院 | Single-robot and multi-robot driving path navigation method |
| CN116486907B (en) * | 2023-01-10 | 2024-04-30 | 湖南工商大学 | Protein sequence tag sequencing method based on A star algorithm |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103557867A (en) * | 2013-10-09 | 2014-02-05 | 哈尔滨工程大学 | Three-dimensional multi-UAV coordinated path planning method based on sparse A-star search (SAS) |
| CN105955280A (en) * | 2016-07-19 | 2016-09-21 | Tcl集团股份有限公司 | Mobile robot path planning and obstacle avoidance method and system |
| CN106774347A (en) * | 2017-02-24 | 2017-05-31 | 安科智慧城市技术（中国）有限公司 | Robot path planning method, device and robot under indoor dynamic environment |
| CN108268971A (en) * | 2017-12-06 | 2018-07-10 | 腾讯科技（深圳）有限公司 | Searching method, device, processor and the electronic device in path |
-
2018
- 2018-09-01 CN CN201811017251.5A patent/CN109115226B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103557867A (en) * | 2013-10-09 | 2014-02-05 | 哈尔滨工程大学 | Three-dimensional multi-UAV coordinated path planning method based on sparse A-star search (SAS) |
| CN105955280A (en) * | 2016-07-19 | 2016-09-21 | Tcl集团股份有限公司 | Mobile robot path planning and obstacle avoidance method and system |
| CN106774347A (en) * | 2017-02-24 | 2017-05-31 | 安科智慧城市技术（中国）有限公司 | Robot path planning method, device and robot under indoor dynamic environment |
| CN108268971A (en) * | 2017-12-06 | 2018-07-10 | 腾讯科技（深圳）有限公司 | Searching method, device, processor and the electronic device in path |
Non-Patent Citations (3)
| Title |
|---|
| Adding Heuristics to Conflict-Based Search for Multi-Agent Path Finding;Ariel Felner .et al;《Proceedings of the International Conference on Automated Planning and Scheduling》;20180615;第83-87页 * |
| ICBS: Improved Conflict-Based Search Algorithm for Multi-Agent Pathfinding;Eli Boyarski .et al;《PROCEEDINGS OF THE TWENTY-FOURTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE (IJCAI)》;20150706;第740-746页 * |
| Optimization Using Boundary Lookup Jump Point Search;Jason Traish .et al;《IEEE TRANSACTIONS ON COMPUTATIONAL INTELLIGENCE AND AI IN GAMES》;20150410;第8卷(第3期);第268-277页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109115226A (en) | 2019-01-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109115226B (en) | Route planning method for avoiding multi-robot conflict based on jumping point search | |
| CN109976350B (en) | Multi-robot scheduling method, device, server and computer readable storage medium | |
| CN111811514B (en) | Path planning method based on regular hexagon grid jump point search algorithm | |
| CN107167154B (en) | Time window path planning conflict solution method based on time cost function | |
| CN111532641B (en) | Parallel path planning method for automatic guide vehicle in storage sorting | |
| US10365110B2 (en) | Method and system for determining a path of an object for moving from a starting state to an end state set avoiding one or more obstacles | |
| CN107913039B (en) | Block selection method and device for cleaning robot and robot | |
| CN113074728B (en) | Multi-AGV path planning method based on jumping point routing and collaborative obstacle avoidance | |
| CN112229419B (en) | Dynamic path planning navigation method and system | |
| CN114510056A (en) | Stable moving global path planning method for indoor mobile robot | |
| CN114489062B (en) | Workshop logistics-oriented multi-automatic guided vehicle distributed dynamic path planning method | |
| CN114815802A (en) | Unmanned overhead traveling crane path planning method and system based on improved ant colony algorithm | |
| CN114756034B (en) | Robot real-time obstacle avoidance path planning method and device | |
| CN115237135A (en) | Mobile robot path planning method and system based on conflict | |
| CN112327890A (en) | Underwater multi-robot path planning based on WHCA algorithm | |
| CN111427341B (en) | Robot shortest expected time target searching method based on probability map | |
| CN115047880A (en) | Intelligent path planning method for robot in unknown dynamic environment | |
| CN105698796B (en) | A kind of method for searching path of multirobot scheduling system | |
| CN115685982A (en) | Navigation path planning method based on connected graph and iterative search | |
| CN116414139B (en) | Mobile robot complex path planning method based on A-Star algorithm | |
| CN116560360A (en) | Method and system for planning real-time dynamic path of medical care robot facing complex dynamic scene | |
| CN112764413B (en) | Robot path planning method | |
| CN114564023A (en) | Jumping point search path planning method under dynamic scene | |
| JP3755268B2 (en) | Automatic guided vehicle control device and automatic guided vehicle control method | |
| CN116907490A (en) | Multi-robot path planning method based on conflict search |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |