A dynamic path planning method and system for parking lot multi-vehicle cooperation

Abstract

The application discloses a kind of parking lot multi-vehicle coordination's dynamic path planning method and system, the method includes: S1, road network modeling and initialization, constructs topological road network model;S2, road network time sequence state dynamic update, according to the planning way of other already in the way vehicle in parking lot, determine the node or road section occupation corresponding to each time point of the vehicle in planning period, lay a foundation for subsequent shortest path search and road network state update;S3, dynamic shortest path search, based on road network time sequence state, the shortest path from starting point to target node is calculated using dynamic shortest path search algorithm;S4, trigger type re-planning;S5, path output and navigation guidance, the current optimal path and alternative path information are output to user or vehicle-mounted system.The application explores the dynamic shortest path planning method of the occupation interaction of multi-vehicle in parking lot in detail, can reflect the multi-vehicle operating state inside parking lot in real time, effectively reduce path conflict and local congestion probability, support high-frequency dynamic update, suitable for complex, dense parking lot operating environment, provide important reference for the core path planning of intelligent parking system and automatic parking system.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation technology, specifically to a dynamic path planning method and system for multi-vehicle collaboration in parking lots. Background Technology

[0002] With the continuous increase in car ownership, people's travel methods have been greatly improved, but this has also led to increasingly serious traffic congestion and parking problems. In addition, the construction speed of urban parking facilities cannot keep up with the pace of car growth, resulting in parking lots becoming more crowded and parking space being constantly compressed. Therefore, higher demands are placed on drivers' parking skills.

[0003] Automated parking technology is an intelligent driving assistance technology that can autonomously and accurately complete the parking maneuver of a vehicle into a parking space, replacing the driver. It enables proactive safety and intelligent functions such as finding and recognizing parking spaces, and safely parking the vehicle. This can significantly reduce parking time and avoid unnecessary parking collision risks. Furthermore, the standardized parking practices guided by automated parking systems will undoubtedly save space resources and improve the chaotic parking situation in cities.

[0004] Path planning algorithms for autonomous parking are a core component of autonomous driving technology, responsible for providing the vehicle with the best or feasible path from its origin to its destination. Path planning in autonomous parking systems is mainly divided into two levels: global path planning and local path planning. Global path planning algorithms design the most efficient driving route from the starting point to the destination based on all environmental data; local path planning algorithms plan with limited information and are typically used to handle dynamic and unknown environments or to react instantly to sudden changes in the environment.

[0005] To shorten parking time, avoid unnecessary parking collision risks, and save more space resources, autonomous parking path planning algorithms urgently need to find an efficient and convenient planning and parking guidance method. Therefore, how to scientifically and rationally determine the shortest planned route for each vehicle entering the parking lot to maximize the saving of parking time and parking distance waste has become an important problem that urgently needs to be solved in the field of intelligent transportation path planning. Summary of the Invention

[0006] This invention provides a dynamic path planning method and system for multi-vehicle collaboration in parking lots. By scientifically and rationally finding the shortest parking path for vehicles entering the parking lot, it provides an important reference for the core path planning of intelligent parking systems and automatic parking systems.

[0007] According to the first aspect, one embodiment provides a dynamic path planning method for multi-vehicle collaboration in a parking lot, the method comprising: Based on the physical structure of the parking lot, a parking lot topology network model containing nodes and edges is constructed. The nodes of the network include channel nodes, entrance and exit nodes, and parking space nodes, and the edges represent passable road segments and are initialized. Based on the parking lot topology road network model, the occupancy status of each node or road segment is dynamically updated in real time, and a road network time-series impedance matrix that is dynamically updated over time is constructed according to the occupancy status. Based on the road network time-series impedance matrix that is dynamically updated over time, a dynamic shortest path search algorithm is used to calculate the shortest path from the current location to the available parking space node. Based on the obtained shortest path, the system guides the vehicle's movement and determines whether the vehicle has reached the planned available parking space node based on the vehicle's real-time location. Before reaching the destination, the system continuously triggers a new dynamic shortest path search based on the road network time-series impedance matrix that is dynamically updated over time, so as to achieve timely updates of the shortest path until the vehicle reaches the final available parking space node.

[0008] Furthermore, based on the physical structure of the parking lot, a parking lot topology network model containing nodes and edges is constructed. The network nodes include access nodes, entrance / exit nodes, and parking space nodes, while edges represent passable road segments. This model is then initialized, specifically including: Establish a topological road network based on the overall organizational flow layout structure of the parking lot, and generate a topological road network map of the parking lot; Based on the topological road network map of the parking lot, the path of each main road / segment is divided into multiple nodes according to the actual distance. The entrance, exit and parking space of the parking lot are all abstracted as nodes, and basic weights including road segment length and travel time are initialized.

[0009] Furthermore, based on the parking lot topology network model, the occupancy status of each node or road segment is dynamically updated in real time. A time-series impedance matrix of the road network is constructed based on the occupancy status and is dynamically updated over time. Specifically, this includes: A time-varying road network time-series impedance matrix is ​​constructed. The matrix elements in this matrix represent the connectivity relationships between nodes, passage costs, and dynamic occupancy status. At time t, when there is a passable edge between node i and node j that is not occupied, the matrix elements... Set as the passage cost for the corresponding road segment; when there is no passable edge between node i and node j, or when the corresponding node or road segment is occupied by other vehicles at the current time, the matrix element... Take the unreachable value, that is When a vehicle is allowed to stay and wait at the current node i, the matrix elements... Set to a non-zero waiting cost to indicate that the vehicle remains in a waiting state at the current node.

[0010] Furthermore, based on the dynamically updated road network time-series impedance matrix, a dynamic shortest path search algorithm is used to calculate the shortest path from the current location to the available parking space node, specifically including: Suppose we have a weighted directed graph G=(V,E), where V is a set of nodes containing n nodes, E is a set of edges containing m edges, and (v,w) is the edge in E from v to w. The non-negative weights of edge (v, w) are the matrix elements in the time-series impedance matrix of the road network at time t. Let s be a node in V, and T be a node in V that can be reached from s. The process of finding the shortest path from s to T with the minimum sum of edge weights is as follows: a. Assign three pieces of information to each node v: k(v), d(v), and p(v), where k(v) is a Boolean variable indicating whether the shortest path to node v has been found; d(v) is the upper bound of the currently known shortest path length from s to v; and p(v) is a pointer to the next node of v. Initialize them respectively:

[0011] In the formula: This represents the upper bound of the current known shortest path length from the starting point s to node v; =0 means the distance from the starting point to itself is 0; = This indicates that all other points are initially considered unreachable; This indicates whether the shortest path to node v has been finally determined. Not yet determined; true: confirmed; This represents the pointer to the predecessor node of node v, also called the forward pointer. This indicates that there is currently no predecessor node; b. Scan k(v) = From the nodes, select a node v with the minimum path length, and let:

[0012] In the formula: : Any candidate node; :node Not yet permanently labeled; : From the starting point to Shortest path estimation; Find the minimum value among these candidate nodes; The currently selected optimal node; Add it to the already determined set; c. Detect each k(w) = And a node w adjacent to node v, if:

[0013] Then let:

[0014] In the formula: : with the current node Adjacent undetermined nodes; : The previously known starting point to Shortest path estimation; From the starting point to the current node The determined shortest path length; :time From arrive The right of the border; :via arrive The cost of the new path; If the new path is shorter, then update. ; Simultaneously set ,express The current best predecessor is ; d. Repeat scan steps b and c until k(t) = true; e. Starting from T, traverse backwards from node pointer p to the starting point s to obtain the shortest path solution:

[0015] In the formula: Indicates starting from the origin To the target point The shortest path; :starting point; :end; The first node of the path is the starting point; The last node of the path is the target point; : An intermediate node on the path from the starting point to the ending point; : The predecessor node is .

[0016] Furthermore, the method also includes: It outputs the current optimal route and alternative route information to the user or vehicle system for navigation or automatic parking control in parking lots.

[0017] According to a second aspect, one embodiment provides a dynamic path planning system for multi-vehicle collaboration in a parking lot, the system comprising: The road network modeling and initialization module is used to construct a parking lot topology road network model containing nodes and edges based on the physical structure of the parking lot. The nodes of the road network include channel nodes, entrance and exit nodes, and parking space nodes, and the edges represent passable road segments and are initialized. The road network time-series impedance matrix construction module is used to dynamically update the occupancy status of each node or road segment in real time based on the parking lot topology road network model, and construct a road network time-series impedance matrix that is dynamically updated over time based on the occupancy status. The dynamic shortest path search module is used to calculate the shortest path from the current location to the available parking space node based on the road network time-series impedance matrix that is dynamically updated over time. The replanning module guides vehicles based on the obtained shortest path. It determines whether the vehicle has reached the planned available parking space node based on the vehicle's real-time location. Before reaching the destination, it continuously triggers a new dynamic shortest path search based on the road network time-series impedance matrix that is dynamically updated over time, so as to achieve timely updates of the shortest path until the vehicle reaches the final available parking space node.

[0018] Furthermore, the system also includes: The route output and navigation guidance module is used to output the current optimal route and alternative route information to the user or vehicle system for navigation or automatic parking control in parking lots.

[0019] This invention provides a dynamic path planning method and system for multi-vehicle collaboration in parking lots, which has the following beneficial effects: This invention explores in detail the dynamic shortest path planning method for the occupancy interaction of multiple vehicles in a parking lot, which can reflect the real-time operation status of multiple vehicles in the parking lot, effectively reduce path conflicts and the probability of local congestion, support high-frequency dynamic updates, and is suitable for complex and dense parking lot operating environments, providing an important reference for the core path planning of intelligent parking systems and automatic parking systems. Attached Figure Description

[0020] Figure 1 A flowchart illustrating a dynamic path planning method for multi-vehicle collaboration in a parking lot, as provided in one embodiment of the present invention; Figure 2 The physical layout diagram of a parking lot in a dynamic path planning method for multi-vehicle collaboration in a parking lot provided in one embodiment of the present invention; Figure 3 This is a parking lot traffic path topology diagram provided in a dynamic path planning method for multi-vehicle cooperation in a parking lot according to an embodiment of the present invention. Figure 4 The diagram shows the dynamic change of the road network temporal impedance matrix with time t in a dynamic path planning method for multi-vehicle cooperation in a parking lot, provided as an embodiment of the present invention. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to specific embodiments and accompanying drawings. Similar elements in different embodiments are referred to by associated similar element reference numerals. In the following embodiments, many details are described to facilitate a better understanding of the invention. However, those skilled in the art will readily recognize that some features may be omitted in different situations, or may be replaced by other elements, materials, or methods. In some cases, certain operations related to the present invention are not shown or described in the specification. This is to avoid obscuring the core parts of the invention with excessive description. For those skilled in the art, detailed description of these related operations is not necessary; they can fully understand the related operations based on the description in the specification and general technical knowledge in the art.

[0022] Furthermore, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can be rearranged or adjusted in a manner obvious to those skilled in the art. Therefore, the various orders in the specification and drawings are only for the clear description of a particular embodiment and do not imply a necessary order, unless otherwise stated that a particular order must be followed.

[0023] The first embodiment of this invention provides a dynamic path planning method for multi-vehicle cooperation in parking lots, which will be discussed below. Figure 1 Please provide a detailed explanation.

[0024] S1. Road Network Modeling and Initialization: Based on the physical structure of the parking lot, construct a parking lot topology road network model containing nodes and edges. The nodes of the road network include channel nodes, entrance and exit nodes, and parking space nodes. The edges represent passable road segments and are initialized.

[0025] The above steps specifically include: S1.1 Topological Road Network Model and Point Cloud Data Basics The topology road network model is based on the physical structure of the parking lot. The topology road network is established according to the overall organizational flow layout structure of the parking lot. The topology road network diagram of the parking lot is drawn. The nodes of the road network represent channel nodes, entrance and exit nodes and parking space nodes, and the edges represent passable road segments. Basic weights such as road segment length and passage time are initialized.

[0026] This step involves clarifying the passability relationships between nodes to form a complete road network structure. Point cloud data is used for the preceding road network modeling and initialization, and for extracting the geometry of parking lot access routes. The point cloud transforms the original parking lot scene into several key points, several passable line segments, and several obstacle boundaries, which facilitates the subsequent establishment of the node set V and edge set E. By fitting the lane centerline to the point cloud, the edge weight set can be further obtained. First, vector line segments are generated based on point cloud data. Principal component analysis is used to determine the main direction of the point cloud distribution in the 3D point cloud under the world coordinate system, and the corresponding vector line segment combinations for each lane are fitted. These line segments will serve as the basic passage paths between nodes. Then, node connection rules are established to determine the connectivity between skeleton nodes and auxiliary nodes, eliminating paths blocked by obstacles such as pillars and walls. Simultaneously, for special areas such as parking spaces, virtual arc segments can be used to construct connections, i.e., a portion of the road segment corresponding to the parking space is extracted as a virtual arc segment to establish the topological relationship between the parking space and the main road network, facilitating precise vehicle parking. The point cloud layer describes the real spatial geometry. The topology layer describes the passage relationships between nodes.

[0027] S 1.2 Physical Layout Diagram The physical layout diagram shows the distribution of parking spaces on the ground and the various components of the parking lot, including the parking lot entrance (the starting point for vehicles to enter), the parking lot exit (the ending point for vehicles to leave), the passage area (the main road for vehicles), and the specific location of the parking spaces. The following example of a parking lot will be used as a supplementary explanation.

[0028] like Figure 2Parking lot physical layout diagram. This is the planar distribution of parking spaces in the parking lot. The elements correspond to the parking lot scene: the left `X1` is the parking lot entrance (the starting point for vehicles to enter); the right `X2` is the parking lot exit (the ending point for vehicles to leave). Passage areas: `S1` and `S2` are the main vehicular passages (the main roads for vehicles to travel within the parking lot), which are the core paths connecting the various parking spaces and the entrances and exits. Parking space areas: `Pa~Pf` are specific parking spaces, distributed around the passages (`S1` / `S2`): `Pa` is located next to the passage above `S1`, `Pb` is next to the passage to the upper right of `S1`; `Pc` and `Pe` are next to the passage to the left of `S1`, `Pd` is next to the passage to the right of `S1`; `Pf` is next to the passage below `S2`.

[0029] S1.3 Topology Graph Establishment The main steps in establishing the topology are as follows: The parking lot traffic path topology is an abstract network model of the parking lot. First, the physical layout is transformed into "nodes + paths," that is, the specific location distribution of parking spaces and the main traffic roads are transformed into parking space nodes and path nodes, respectively. Each parking space corresponds to one parking space node, and each main road / segment is divided into multiple path nodes according to the actual distance (the node interval is the distance of one parking space or a fixed 1-5m). Second, the entrance and exit of the parking lot are also transformed into the starting node for vehicles to enter and the ending node for vehicles to leave the parking lot. Third, the path formed by connecting each node in the topology is called an edge. The weight of the edge is determined by the passage cost / passage expense, which can be the travel distance / estimated time. Finally, the road network topology of the parking lot is formed.

[0030] Specific examples are as follows, such as Figure 3 Parking lot traffic path topology diagram: This is an abstract network model of the parking lot, transforming "physical space" into "nodes + paths" for navigation or path calculation. Node definitions: `X1` (entrance), `X2` (exit): the starting / ending point for vehicles entering and exiting; `S1` / `S2` (channel nodes): representing the main channels within the parking lot (which can be understood as the "intersection" or "segment" of channels); `Pa~Pf` (parking space nodes): representing specific parking spaces. Edges and weights: Each edge corresponds to the traffic path of a vehicle between two nodes, and the numbers on the edges represent the traffic cost.

[0031] The core function of a parking lot scenario. The topology map is the basic model of an intelligent parking system. For example, after a vehicle enters from `X1`, the system can use the topology map to calculate the "shortest path to the available parking space `Pe`" (`X1→S1→S2→Pe`); it can also calculate the "fastest route from `Pf` to `X2`" (`Pf→X2`).

[0032] S2. Dynamic update of road network time sequence status: Based on the parking lot topology road network model, the occupancy status of each node or road segment is dynamically updated in real time, and a road network time sequence impedance matrix that is dynamically updated over time is constructed according to the occupancy status.

[0033] In this embodiment, based on the planned routes, known location status, and estimated arrival times of vehicles already en route in the parking lot, the occupancy status of each vehicle at different time levels within the planning period is deduced, and the occupancy status is mapped to the road network time-series impedance matrix. The road network time-series impedance matrix dynamically changes with the time levels, and the passage weight of the same node or road segment is allowed to be different in different time levels. Nodes or road segments occupied by vehicles are marked as high impedance or impassable in the corresponding time level. After the occupancy is lifted, they are restored to the basic passable state in subsequent time levels, thereby forming a road network passable state covering the planning period, providing a basis for subsequent dynamic shortest path search. Specifically, as follows: S2.1 Road Network Time Sequence Status Dynamic Update The road network time-series status update includes: discretizing the parking lot topology road network according to a preset time step; determining the occupancy status of each node or road segment in the corresponding time layer based on the planned vehicle position status at each time point; and mapping the occupancy status to a passage weight or passability status. The basic principle is to update the impedance matrix at a fixed time step (e.g., 1 second) to quantify the time-series occupancy status of each moving vehicle on the road network nodes. In the road network, when there is a passable connection between two nodes, a basic passage weight is assigned. The basic passage weight can be set to 1 to represent the distance unit between adjacent nodes (it can take any value), and 1 represents passability (in this case, a distance unit can be set to 2m or 3m). If a node is occupied by a vehicle at a certain moment, then at that moment, this node is not connected to other nodes in the topology road network, and the impedance value related to this node in the impedance matrix is ​​assigned a high impedance. If a road network node is occupied by multiple vehicles, the relevant link impedance value in the road network impedance matrix of each occupied node at this moment will be assigned a high impedance or marked as impassable. When a vehicle is allowed to stay and wait at the current node i, the corresponding diagonal element will be... Set to a non-zero waiting cost to indicate that the vehicle remains in a waiting state at the current node.

[0034] S2.2 Parking Lot Example (Dynamic Update of Time-Sequence Impedance Matrix) Example (dynamic update of timing impedance matrix) is as follows: Suppose there is a directed graph with n nodes, and the edges and their weights are as follows: At time t, if the nodes are connected, then the weight is... (Size is 1, corresponding to an actual node spacing of 1m); at time t, there are no direct connections between nodes, so the weight is... (for The time-series impedance matrix changes dynamically with time t. It also changes in real time. At time t=1, only one car enters the parking lot, so all paths can be chosen. Therefore, the time-series impedance matrix is ​​composed of the initial path nodes and directly connected edges. At time t=2, the second car enters the parking lot. The first car occupies the shortest path planned for the second car. Assume that entrance 1 is only connected to node 2, node 2 is only connected to node 3, node 3 is only connected to node 4, and so on. All nodes in the parking lot entrance section are individually connected.

[0035] Therefore, by occupying node 2, the timing impedance matrix of the second vehicle is connected to node 2. = If other conditions are unaffected, they remain unchanged; the weight of the edge directly connected to the node that is occupied is determined by the node's position. = Updating the time-series impedance matrix at this point yields: The network time-series impedance matrix at time t:

[0036]

[0037] At time t=2, the weights between the nodes are as follows:

[0038] At time t=3, the second car is still in the parking lot, following its path to a parking space and occupying node 1. The first car does not occupy the adjacent node of the shortest path planned by the second car. Node 2 is empty at this time, and the first car occupies node 3. Therefore, the time-series impedance matrix of the second car is connected to node 2. Restore the weights to their original values; leave other weights unchanged if they have no effect. Update the time-series impedance matrix as follows:

[0039]

[0040] At time t=3, the weights between the nodes are as follows:

[0041] At time t=4, the third car enters the parking lot. At this time, the second car is still in the parking lot, following its path to a parking space. The first car does not occupy the adjacent node of the shortest path planned by the second car. Node 2 is now occupied by the second car. Therefore, the time-series impedance matrix of the second car is connected to node 2. Node 4 is currently occupied by the first vehicle, and the corresponding time-series impedance matrix... If other parameters remain unchanged, the time-series impedance matrix is ​​updated to obtain:

[0042]

[0043] At time t=4, the weights between the nodes are as follows:

[0044] like Figure 4 The timing impedance matrix is ​​dynamically updated at each time step to generate impedance matrices corresponding to time steps t=2, t=3, and t=4, which represent the occupancy status of each node at that time step.

[0045] Similarly, the network time-series impedance matrix will continue to be updated in real time according to time t.

[0046] S3. Dynamic Shortest Path Search: Based on the road network time-series impedance matrix that is dynamically updated over time, the dynamic shortest path search algorithm is used to calculate the shortest path from the current location to the available parking space node.

[0047] In this embodiment, based on the temporal state of the road network, a dynamic shortest path search algorithm is used to calculate the shortest path from the starting point to the target node. The shortest path calculation allows the vehicle to remain parked at any node, thereby calculating the optimal travel path from the starting point to the target node by comprehensively comparing the temporal costs of continuing to travel versus waiting in place, while satisfying the occupancy constraint. The starting point is the parking lot entrance, and the target node, i.e., the destination, is the target parking space selected by the user.

[0048] In this embodiment, a speed prediction matrix is ​​added to the traditional Dijkstra shortest path algorithm to simulate the dynamic changes in road segment speed, and finally outputs the path with the optimal time (rather than the path with a fixed distance).

[0049] The core of the dynamic Dijkstra's algorithm lies in introducing the time dimension into traditional pathfinding. Unlike static algorithms that only consider distance, the dynamic algorithm considers the time when vehicles arrive at each node. Real-time query of the road network time-series impedance matrix at that moment This allows them to avoid nodes that are already occupied by other vehicles.

[0050] Find the shortest path candidate set and update the latest road occupancy status every 1-5 seconds (in real time) to update the shortest path candidate set. Construct the passability status of each node of the parking lot road network at each time according to the predetermined time granularity, and obtain the vehicle location and operation status in the parking lot in real time to form a road network time-series status model covering the entire planning cycle; The above steps specifically include: S3.1 Dynamic Road Network Model Selection Dynamic road network models are mainly divided into: time-sharing road network models and real-time road network models. The time-sharing road network model is as follows:

[0051] In the formula Represents a dynamic road network map; Represents a set of nodes; in: The i-th node, Total number of nodes; This represents the set of edges, which is the road connection relationship between nodes. Indicates from node To the node One of the edges; Represents a set of time periods, where: The kth time period, m: the total number of time periods; This represents the set of edge weights, that is, the weight of each edge at different time periods. : The weight of the edge from node i to node j during the k-th time interval.

[0052] The real-time road network model is as follows:

[0053] In the formula Represents a real-time road network map; Represents a set of nodes; in: The i-th node, Total number of nodes; This represents the set of edges, which is the road connection relationship between nodes. Indicates from node To the node One of the edges; T is a time series set, which divides the entire day into different points in time and is updated in real time at a frequency of . t: the t-th moment; s: the total number of updates in one day; This represents the set of weights in real time. At time t, the weight of the edge from node i to node j is continuously updated based on real-time data.

[0054] As shown in equations (1) and (2), both the time-segmented Dijkstra and real-time Dijkstra algorithms use real-time traffic information to establish dynamic traffic models for analysis. The time-segmented road network model uses real-time road information, with time intervals typically measured in hours, resulting in relatively large intervals. The real-time road network model uses the traffic information update frequency as the dividing standard, with time interval A typically being 5 minutes. This time interval can be further adjusted to 1-5 seconds, thus resulting in more refined time segmentation and denser information data.

[0055] Based on the above analysis, the real-time road network model is clearly more refined and better reflects the instantaneous changes in traffic conditions.

[0056] S3.2 Dijkstra's Algorithm and Principles The dynamic Dijkstra algorithm is also a shortest path algorithm based on a greedy strategy. Its other basic principles are the same as Dijkstra's algorithm. Its main idea is to construct a path tree by increasing the path length point by point, so as to obtain the shortest path from the root node (i.e. the specified starting point) of the tree to all other nodes.

[0057] Let set S store the destinations of the shortest paths that have been found. Initially, set S contains only one starting point. . For each shortest path found thereafter ( ,…, ), then Add them to set S until all vertices are added to S.

[0058] Introduce an auxiliary vector d, whose components... , indicating the current location found from the starting point to other vertices The shortest path length. Its initial state is: if from to If there is an arc, then Let be the weight on the arc, otherwise let Let the first shortest path be (∞). , ), where k satisfies

[0059] Then the next shortest path (with the destination as...) ), or ( , ), or ( , , Generally, assuming S is the set storing the endpoints of the shortest paths already found, then the length of the intermediate node of the next shortest path is:

[0060] After finding a shortest path each time, its destination is... Add to set S, then for all other vertices Modify it

[0061] In the formula, It is an arc ( , The weights on ) . The value of t is determined by distance The number of links is represented by 't', which corresponds to the number of links. This can be viewed using a static Dijkstra's algorithm. distance There are several links, mainly depending on the... As the endpoint, with If there are several intervals, then t corresponds to... and The number of links between them.

[0062] The above algorithm will generate the shortest path from the starting point to all other vertices. For single-vehicle path planning, it is only necessary to calculate one shortest path from the starting point to the destination. Therefore, the algorithm can be simply modified so that it terminates when the destination of the shortest path is found to be the destination of the target path.

[0063] Based on the above principles, let there be a weighted directed graph G=(V,E), where V is a vertex set containing n vertices (nodes), E is an arc set containing m arcs (edges), and (v,w) is the arc in E from v to w. Let s be the non-negative weights of arc (v, w). Let s be a vertex in V, and T be a vertex in V that can be reached from s. The process of finding the shortest path from s to T with the minimum sum of arc weights is as follows: (1) Assign three pieces of information to each vertex v: k(v), d(v), and p(v), where k(v) is a Boolean variable indicating whether the shortest path to vertex v has been found; d(v) is the upper bound of the currently known shortest path length from s to v; and p(v) is a pointer to the back vertex of v, and initialized to:

[0064] In the formula This represents the upper bound of the current known shortest path length from the starting point s to node v; =0, the distance from the starting point to itself is 0; = All other points are initially considered unreachable; This indicates whether the shortest path to node v has been finalized.

[0065] Not yet determined; true: confirmed; This represents the pointer to the predecessor node of node v, also called the forward vertex pointer; This indicates that there is currently no predecessor node; (2) Scan k(v) = From the vertices, select a vertex v with the minimum path length, and let...

[0066] In the formula : Any candidate node; :node Not yet permanently labeled; : From the starting point to Shortest path estimation; Find the minimum value among these candidate nodes; The currently selected optimal node; : Add it to the established set.

[0067] (3) Detect each k(w) = And a vertex w adjacent to vertex v, if satisfying

[0068] Then let

[0069] In the formula : with the current node Adjacent undetermined nodes; : The previously known starting point to Shortest path estimation; From the starting point to the current node The determined shortest path length; :time From arrive The edge weights (i.e., the corresponding impedance values ​​in the road network time-series impedance matrix). :via arrive The cost of the new path; If the new path is shorter, then update. ; Simultaneously set ,express The current best predecessor is ; (4) Repeat the scan operations (2) and (3) until k(t) = true.

[0070] (5) Starting from T, traverse backwards from vertex pointer p to the starting point s to obtain the shortest path solution:

[0071] From the above formula, we can obtain the final shortest path.

[0072] In the formula Indicates starting from the origin To the target point The shortest path; :starting point; :end; The first node of the path is the starting point; The last two nodes of the path are the target points; : An intermediate node on the path from the starting point to the ending point; : The predecessor node is .

[0073] S3.3 Example of Dynamic Dijkstra's Algorithm (1) Background setting assumptions Map model based on Figure 2 , Figure 3 The system includes nodes: entrance X1 and parking space node P. The distance between each node is one step, and the distance between nodes is 1 second, which corresponds to a distance of 1m. At time t=0, the first car a1 enters the parking lot from entrance X1. The target parking space of car a1 is the parking space to the left of parking space node P. At this time, the road network topology map and the instantaneous impedance matrix are generated based on the occupancy of the road network nodes in the parking lot. Then, Dijkstra's algorithm is performed to search for the shortest path at this time and proceed according to it.

[0074] (2) At time t=1, car a1 travels in the parking lot according to the shortest path obtained at time t=0. At this time, based on the occupancy of the path nodes in the parking lot, the road network topology map at time t=1 is generated again, and Dijkstra's algorithm is performed again to search for the shortest path at this time and travels according to it.

[0075] (3) After a period of time, such as at time t=10, the second car a2 arrives at the entrance and is about to enter the parking lot. The target parking space of car a2 is parking space P. Car a1 is moving according to the shortest path generated at each time. At this time, it is at node x10. x10 is located at the entrance of the left vertical passage. Then, at time t=10, the road network node topology diagram generated by car a2 should not include node x10 and other nodes directly connected to x10. That is, the node at the entrance of the passage in the road between the left passages is not included in the topology diagram. In the topology road network diagram generated at this time, it finds its own shortest path and moves in parallel.

[0076] (4) After a period of time, such as at t=20, car a1 reaches the exit of the left passage road, and at this time the second car a2 reaches the entrance of the left passage road. At this time, the node at the exit of the left passage is not included in the generated road network node topology graph, and other nodes directly connected to this node do not exist in the topology graph. At this time, car a2 has two paths to choose from in the road network topology graph: 1. Travel along the entrance X1 to exit X2 passage towards the upper right exit X2 of the parking lot, reach the entrance of the right vertical passage, and then travel down to finally reach the target node P; 2. Travel along the left vertical passage. After car a1 reaches the target node, the road network topology graph is updated again. All the forward road network nodes of car a2 are in an idle state. At this time, all the nodes in the road network topology graph are connected. After reaching the exit of the left vertical passage, car a2 can travel to the right to finally reach the target node P. By comparing the two driving schemes, the shortest path, path 2, is obtained. By traveling along the shortest path, the target node P is finally reached. If different situations arise, such as the previous vehicle occupying a node for a long time, resulting in only one path being available in the subsequent road network topology, or after comparison, it is found that the vehicle waits too long in place in another method, indicating that this path is not the shortest path, then the shortest path is selected and the vehicles proceed in parallel, thus satisfying the dynamic programming requirements for multiple vehicles.

[0077] S4. Triggered Replanning: Based on the obtained shortest path, guide the vehicle to travel. Determine whether the vehicle has reached the planned vacant parking space node based on the vehicle's real-time location. Before reaching the destination, continuously trigger the re-dynamic shortest path search based on the road network time-series impedance matrix that is dynamically updated over time, so as to achieve timely updates of the shortest path, until the vehicle reaches the final vacant parking space node.

[0078] In this embodiment, the vehicle travels along the path obtained by the shortest path search in S3. The road network topology map and the time-series impedance matrix are updated every moment according to the vehicle's position status in the parking lot. That is, changes in the node occupancy status, road segment occupancy status, or target parking space occupancy status will be immediately reflected in the updated road network topology map and time-series impedance matrix at this moment. Alternatively, if the waiting time of the current vehicle at a certain node exceeds a preset threshold, the dynamic shortest path search is re-triggered to update the optimal travel path of the current vehicle from the current position to the target node and guide the vehicle to continue traveling.

[0079] The threshold for replanning is set to 1 second, meaning it is updated every second, since the shortest path is calculated at every moment during the process.

[0080] S5, Route Output and Navigation Guidance, outputs the current optimal route and alternative route information to the user or vehicle system for navigation or automatic parking control in parking lots.

[0081] 3D navigation implementation: Based on the 3D digital twin model of the parking lot, the system renders a highlighted path (supporting perspective switching: first person / overall view), giving drivers a 3D view of the scene and marking key facilities such as elevator entrances and payment machines in the surrounding area.

[0082] It should be noted that for a detailed description of the dynamic path planning system for multi-vehicle collaboration in parking lots provided in the embodiments of the present invention, please refer to the relevant descriptions provided in the embodiments of the present invention, which will not be repeated here.

[0083] The above specific examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the ideas of this invention. It is impossible to exhaustively list all possible embodiments here; all obvious variations or modifications derived from the technical solutions of this invention are still within the protection scope of this invention.

Claims

1. A dynamic path planning method for multi-vehicle cooperation in a parking lot, characterized in that, The method includes: Based on the physical structure of the parking lot, a parking lot topology network model containing nodes and edges is constructed. The nodes of the network include channel nodes, entrance and exit nodes, and parking space nodes, and the edges represent passable road segments and are initialized. Based on the parking lot topology road network model, the occupancy status of each node or road segment is dynamically updated in real time, and a road network time-series impedance matrix that is dynamically updated over time is constructed according to the occupancy status. Based on the road network time-series impedance matrix that is dynamically updated over time, a dynamic shortest path search algorithm is used to calculate the shortest path from the current location to the available parking space node. Based on the obtained shortest path, the system guides the vehicle's movement and determines whether the vehicle has reached the planned available parking space node based on the vehicle's real-time location. Before reaching the destination, the system continuously triggers a new dynamic shortest path search based on the road network time-series impedance matrix that is dynamically updated over time, so as to achieve timely updates of the shortest path until the vehicle reaches the final available parking space node.

2. The dynamic path planning method for multi-vehicle cooperation in a parking lot as described in claim 1, characterized in that, Based on the physical structure of the parking lot, a parking lot topology network model is constructed, including nodes and edges. The network nodes include access nodes, entrance / exit nodes, and parking space nodes, and the edges represent passable road segments. The model is initialized, specifically including: Establish a topological road network based on the overall organizational flow layout structure of the parking lot, and generate a topological road network map of the parking lot; Based on the topological road network map of the parking lot, the path of each main road / segment is divided into multiple nodes according to the actual distance. The entrance, exit and parking space of the parking lot are all abstracted as nodes, and basic weights including road segment length and travel time are initialized.

3. The dynamic path planning method for multi-vehicle cooperation in a parking lot as described in claim 1, characterized in that, Based on the parking lot topology network model, the occupancy status of each node or road segment is dynamically updated in real time. A time-series impedance matrix of the road network is constructed based on the occupancy status and is updated over time. Specifically, this includes: A time-varying road network time-series impedance matrix is ​​constructed. The matrix elements in this matrix represent the connectivity relationships between nodes, passage costs, and dynamic occupancy status. At time t, when there is a passable edge between node i and node j that is not occupied, the matrix elements... Set as the passage cost for the corresponding road segment; when there is no passable edge between node i and node j, or when the corresponding node or road segment is occupied by other vehicles at the current time, the matrix element... Take the unreachable value, that is When a vehicle is allowed to stay and wait at the current node i, the matrix elements... Set to a non-zero waiting cost to indicate that the vehicle remains in a waiting state at the current node.

4. The dynamic path planning method for multi-vehicle cooperation in a parking lot as described in claim 1, characterized in that, Based on the dynamically updated road network time-series impedance matrix, a dynamic shortest path search algorithm is used to calculate the shortest path from the current location to the available parking space node, specifically including: Suppose we have a weighted directed graph G=(V,E), where V is a set of nodes containing n nodes, E is a set of edges containing m edges, and (v,w) is the edge in E from v to w. The non-negative weights of edge (v, w) are the matrix elements in the time-series impedance matrix of the road network at time t. Let s be a node in V, and T be a node in V that can be reached from s. The process of finding the shortest path from s to T with the minimum sum of edge weights is as follows: a. Assign three pieces of information to each node v: k(v), d(v), and p(v), where k(v) is a Boolean variable indicating whether the shortest path to node v has been found; d(v) is the upper bound of the currently known shortest path length from s to v; and p(v) is a pointer to the next node of v. Initialize them respectively: In the formula: This represents the upper bound of the current known shortest path length from the starting point s to node v; =0 means the distance from the starting point to itself is 0; = This indicates that all other points are initially considered unreachable; This indicates whether the shortest path to node v has been finally determined. Not yet determined; true: confirmed; This represents the pointer to the predecessor node of node v, also called the forward pointer. This indicates that there is currently no predecessor node; b. Scan k(v) = From the nodes, select a node v with the minimum path length, and let: In the formula: : Any candidate node; :node Not yet permanently labeled; : From the starting point to Shortest path estimation; Find the minimum value among these candidate nodes; The currently selected optimal node; Add it to the already determined set; c. Detect each k(w) = And a node w adjacent to node v, if: Then let: In the formula: : with the current node Adjacent undetermined nodes; : The previously known starting point to Shortest path estimation; From the starting point to the current node The determined shortest path length; :time From arrive The right to the side; :via arrive The cost of the new path; If the new path is shorter, then update. ; Simultaneously set ,express The current best predecessor is ; d. Repeat scan steps b and c until k(t) = true; e. Starting from T, traverse backwards from node pointer p to the starting point s to obtain the shortest path solution: In the formula: Indicates starting from the origin To the target point The shortest path; :starting point; :end; The first node of the path is the starting point; The last node of the path is the target point; : An intermediate node on the path from the starting point to the ending point; : The predecessor node is .

5. The dynamic path planning method for multi-vehicle cooperation in a parking lot as described in claim 1, characterized in that, The method further includes: It outputs the current optimal route and alternative route information to the user or vehicle system for navigation or automatic parking control in parking lots.

6. A dynamic path planning system for multi-vehicle cooperation in a parking lot, characterized in that, The system includes: The road network modeling and initialization module is used to construct a parking lot topology road network model containing nodes and edges based on the physical structure of the parking lot. The nodes of the road network include channel nodes, entrance and exit nodes, and parking space nodes, and the edges represent passable road segments and are initialized. The road network time-series impedance matrix construction module is used to dynamically update the occupancy status of each node or road segment in real time based on the parking lot topology road network model, and construct a road network time-series impedance matrix that is dynamically updated over time based on the occupancy status. The dynamic shortest path search module is used to calculate the shortest path from the current location to the available parking space node based on the road network time-series impedance matrix that is dynamically updated over time. The replanning module guides vehicles based on the obtained shortest path. It determines whether the vehicle has reached the planned available parking space node based on the vehicle's real-time location. Before reaching the destination, it continuously triggers a new dynamic shortest path search based on the road network time-series impedance matrix that is dynamically updated over time, so as to achieve timely updates of the shortest path until the vehicle reaches the final available parking space node.

7. A dynamic path planning system for multi-vehicle cooperation in a parking lot according to claim 6, characterized in that, The system also includes: The route output and navigation guidance module is used to output the current optimal route and alternative route information to the user or vehicle system for navigation or automatic parking control in parking lots.

Images