CN112797995A - Vehicle emergency navigation method with space-time characteristic situation information - Google Patents
Vehicle emergency navigation method with space-time characteristic situation information Download PDFInfo
- Publication number
- CN112797995A CN112797995A CN202011495841.6A CN202011495841A CN112797995A CN 112797995 A CN112797995 A CN 112797995A CN 202011495841 A CN202011495841 A CN 202011495841A CN 112797995 A CN112797995 A CN 112797995A
- Authority
- CN
- China
- Prior art keywords
- road network
- network
- road
- vehicle
- nodes
- 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.)
- Granted
Links
Images
Classifications
-
- 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
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/34—Route searching; Route guidance
- G01C21/3407—Route searching; Route guidance specially adapted for specific applications
- G01C21/343—Calculating itineraries, i.e. routes leading from a starting point to a series of categorical destinations using a global route restraint, round trips, touristic trips
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Automation & Control Theory (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a vehicle emergency navigation method with space-time characteristic situation information. The method achieves the effect of simplification while ensuring the integrity of the road network structure, thereby improving the quick update of situation information and realizing a quick mapping mechanism of a path; and finally, on a vehicle path planning layer, combining the processing results of the two layers to the road network, providing a heuristic dynamic path planning algorithm aiming at urban emergency, and realizing the planning of the vehicle emergency route by the recommended evacuation area and the analysis of the situation information of the road. The invention enables the vehicle to drive from the emergency area to the safe area in the shortest time, achieves good vehicle evacuation effect and enables the route planned by the emergency navigation of the vehicle to be more accurate and reliable.
Description
Technical Field
The invention belongs to the field of urban vehicle emergency navigation, and relates to road network speed data prediction, road network situation information hierarchical updating and a vehicle emergency path planning method based on a situation road network.
Background
With the continuous promotion of urbanization construction in China, urban traffic networks become more complex, the whole urban traffic network is heavily loaded, and the problems of urban traffic jam and unreasonable traffic resource distribution are prominent. When an emergency occurs in a city, the vehicle congestion situation will become more severe over time. Particularly, in a real traffic network, the state of the network is influenced by various factors, and has complicated and variable characteristics, for example, traffic flow, weather conditions, emergencies, traffic control and the like all affect the state of the traffic network, and the traffic network is changed drastically. The situation data which has the space-time characteristic and seriously influences the road network and the urban road network environment with various states bring opportunities and challenges to the development of the vehicle emergency navigation. When the road weight and the structural topology information in the situation traffic network change, the network cannot enter a stable state within a period of time, most of the current navigation methods cannot make effective decisions for drivers in a time-varying situation network environment, and when emergency path planning is performed on vehicles, the number of vehicles on a certain route rapidly increases within a short time, so that various secondary traffic jams and traffic accidents are induced. In addition, due to the fact that road weight changes continuously due to the dynamically changing road network environment and situation data of road network changes are not fully utilized, a conventional path planning method cannot converge to a global optimal solution, so that decision after navigation is influenced, and the performance of emergency navigation is seriously influenced. Therefore, vehicle emergency navigation based on a situation road network with space-time characteristics is one of necessary means for reducing huge economic loss and casualties caused by emergency situations.
Aiming at the problems, the invention provides a vehicle emergency navigation architecture and a method for a situation road network with space-time characteristics after an emergency occurs in a city. The method can effectively utilize time-varying situation information in the road network and the topological relation of the road network, integrate the situation information of the road network which is greatly changed due to the emergency into the analysis of the road network, improve the rationality of the situation analysis and enable the emergency navigation to meet the actual requirement. The method comprises the steps of firstly carrying out vector rasterization on a road network, and recommending an evacuation area to vehicles needing evacuation near an emergency by calculating the commuting capacity (evacuation capacity) of each grid. And then, hierarchical division and establishment of situation indexes are carried out on the data level of the road network, so that the situation information is updated rapidly and reasonably utilized. On the navigation method level, the situation information and the layered road network structure are combined to construct an emergency path planning model, so that the result of path search can be rapidly converged to an optimal solution, and a balance is sought between the potential situation change of path selection and the travel cost.
Based on the thought, the method comprises the steps of firstly carrying out rasterization processing on a Beijing city road network, predicting road speed by combining historical speed data of the road network, calculating the commuting capacity of the grids in the road network according to the number of nodes, the entrance and exit degrees and the road prediction speed contained in each grid in a grid map, sequencing the grids according to the size of the commuting capacity, and calculating the area which is most suitable for inducing vehicles to evacuate in the road network when an emergency occurs; secondly, carrying out hierarchical division and core sub-network extraction on the urban traffic network, constructing a network situation information tree index, reducing redundant information of path retrieval in the road network, and improving the updating speed of the network situation information so as to realize the rapid adjustment of the shortest path sequence between nodes; and finally, providing a heuristic vehicle emergency path planning algorithm by combining the recommended evacuation area and the actual navigation demand under the condition of road network situation information change, and providing powerful support and help for emergency navigation and evasive route planning of urban vehicles.
Disclosure of Invention
The invention provides an urban vehicle emergency path planning architecture and method based on a situation grid road network. The general architecture of the vehicle emergency navigation system consists of three parts, namely a vehicle evacuation area recommendation layer and a road network data index from bottom to topAnd a vehicle path planning layer. The framework firstly maps a road network into a two-dimensional plane coordinate system through a lowest vehicle evacuation area recommendation layer, and performs equidistant grid division on the basis of the two-dimensional plane coordinate system, so that a map subjected to grid division processing is divided into ng×ngThe grid predicts the speed of roads in the road network by a neural network method, calculates the evacuation capacity of each grid by combining the predicted speed of the road network and the communication relation between the grids, and recommends the area which is most suitable for vehicle evacuation under the current condition to the user; and secondly, the road network data index layer calculates the intermediate centrality of the nodes in the traffic network by carrying out subnet extraction layering operation on the traffic network, and contracts the nodes with lower intermediate centrality measurement in the sub-network. And then the situation data information collected by various sensors in the road network is fused into the situation analysis of the road network, and a road network situation information tree index is constructed. The method achieves the effect of simplification while ensuring the integrity of the road network structure, thereby improving the quick update of situation information and realizing a quick mapping mechanism of a path; and finally, on a vehicle path planning layer, combining the processing results of the two layers to the road network, providing a heuristic dynamic path planning algorithm aiming at urban emergency, and realizing the planning of the vehicle emergency route by the recommended evacuation area and the analysis of the situation information of the road.
(1) Vehicle evacuation area recommendation layer
The grid with higher commuting capability in a certain range of the emergency occurrence point is recommended to the user as an evacuable grid by calculating the commuting capability of each grid in the vector grid network. Firstly, the road network data is subjected to rasterization processing by using an equal interval rasterization method, so that the road network data can be divided into a plurality of regional sub-blocks through reasonable equal interval sizes. After an emergency occurs in a city, a large number of vehicles can rush in and pile up on traffic roads in a road network in an area near the emergency in a short time, so that congestion phenomena of different degrees occur. Due to the topology and communication characteristics of a traffic network, the road congestion can be diffused from a few intersections to a plurality of intersections connected with each other, and regional congestion is formed. And these congested intersections are necessarily geographically connected intersections of consecutive several adjacent roads. The area made up of these intersections can be defined as a congested area. The number of intersections and the number of entrances and exits are large, and the grids with high road network speed can obtain higher commuting capacity and sequencing values, which means that the areas have better junction conversion function and evacuation capacity. By analyzing intersection nodes contained in the grids, the entrance and exit degrees of the grids and the communication relation among the grids, the evacuation capacity of a certain grid in the whole grid network can be effectively judged.
Converting longitude and latitude coordinates of nodes in the road network data into geodetic coordinates, and obtaining plane coordinate data corresponding to the nodes in the road network by using Gaussian projection alignment, wherein a node plane coordinate set is expressed as Coordxy＝{(x1,y1),(x2,y2),…(xi,yi)}. Firstly, determining the position of the origin of coordinates, and finding out the node coordinate (X) with the minimum X-axis value through the node plane coordinate datao,yo) Coordinate (x) of node having minimum value with Y axisa,ya) Two points satisfy x respectivelyo＜xothersAnd ya＜yothers. Translating the minimum point of the X-axis coordinate value to make the abscissa zero to obtain (0, y)o) Translating the minimum point of the Y-axis coordinate value to make the ordinate zero (x)a0), and translating all points according to the translation amounts of the two points, the translated (0, y)o)、(xa0) and partial nodes are reproduced on the coordinate axes as shown in FIG. 1.
The grids are connected through the roads, the speed of the connecting roads among the grids is analyzed, and the communication capacity of the grids and other grids can be calculated. However, since the speed of the roads in the road network changes with time, it is necessary to estimate the road congestion in a future time period by predicting the speed of the road network, and to select a vehicle evacuation area by appropriately calculating the grid commuting capacity. Since the road network speed changes with time, the road network speed prediction problem is converted into a time series prediction problem. But only the time-series prediction of the network speed data is farFar from insufficient, the road network speed is also correlated at the spatial level, so that the road network speed prediction needs to be analyzed at two levels of time and space. Fig. 2 is a schematic diagram of prediction of a time-space sequence of road network speed, the model is input into road network historical speed data collected by a vehicle speed sensor in an urban road network, and prediction data of road network speed is obtained through a neural network prediction model. Since the network speed data changes with time in a continuous space, it is necessary to extract the spatial correlation characteristic of the network speed data efficiently. Since the sensor distribution in the road network is of a non-Euclidean structure, the graph convolutional neural network (GCN) has excellent performance in processing graph structure data. RNN is always a more mainstream processing method in the time series prediction problem, wherein LSTM shows better performance, so the invention provides a road network speed prediction hybrid model GLN combining GCN and LSTM. The model is based on three correlations of road network speed at a time level: proximity, periodicity and trend, and the LSTM network is adopted to capture three time dependencies in road network speed data respectively, so that more accurate prediction of road speed is realized. Original time series X of road network speed data0＝(x1,x2,…,xm) Normalization was performed to obtain the sequence X' ═ { X1’,x2’,…,xn' reconstructing three subdata sequences, time sequences X ', obtained by reconstructing the normalized link speed data 'Tw、X’TdAnd X'TmAs inputs to the proximity sequence, periodic sequence, and trend sequence components.
X’Tm＝(x’Tm1,x’Tm2,...,x’Tm(n-1))T
X’Td＝(x’Td1,x’Td2,...,x’Td(n-1))T
X’Tw＝(x’Tw1,x’Tw2,...,x’Tw(n-1))T
The input of the road network speed prediction model for training the time sequence is the three sequences. Corresponding to the training setThe theoretical output of (1) is Y ═ Y1,Y2,…,Yn-1]T＝[x2’,x3’,…,xn’]T. According to the construction of the three subsequences, the three subsequences are firstly input into the GCN module for learning the spatial features, then the output of the GCN is used as the input of the LSTM module for learning the time dimension, and the corresponding GLN model is shown in FIG. 3.
After rasterization and road network speed prediction are carried out on the road network, the commuting capability of the grid is calculated. The Grid commuting capability calculation and sorting recommendation algorithm is a Dynamic Grid Page sorting (DGPR) algorithm improved on the basis of a traditional Page sorting algorithm (PageRank algorithm). The web pages can be considered as a large directed graph, which is similar to the abstract directed graph of the rasterized traffic network in this document. Fig. 4 is a schematic diagram of a grid network, which divides the network into 2 × 2 grids. The blue dots in the figure represent road nodes in the road network and the red lines represent roads interconnected between different grids. One grid in the grid network can be abstractly regarded as a webpage, and the roads communicated with each other among the grids can be regarded as links among the webpages. Therefore, we can perform a link analysis algorithm on the rasterized road network. The rank value for each grid is expressed as:
DGPR(Gridid)＝∑Pri*(VGridid/V)+(1-d)/N
Pri＝∑Iej*d*(Rs/120)
wherein, DGPR (Grid)id) Is the rank value of the trellis after iteration of the DGPR algorithm. This value represents the commuting capacity of the grid. N represents the number of grids containing the node. Pr (Pr) ofiIs the DGPR value for the ith trellis, i represents the ID of the trellis. VGrididIs GrididV is the total number of vertices in the entire network. The damping coefficient d is 0.85. IejRepresenting the in-degree of the jth grid. (Rs/120) represents the normalized road speed. When the vehicles are needed to be evacuated due to serious congestion or emergency situations in the road network, recommending the most appropriate evacuation area to the user according to the road condition at the moment of emergency situation.
(2) Road network data index layer
The layer is the construction of road network data situation indexes. According to the layer, the road network is firstly subjected to layering processing, the situation information in the road network is fused into the layered road network, the situation information tree index based on the layered road network is constructed, and the structure of the road network is simplified on the premise of ensuring the integrity of the road network through reasonable contraction of nodes, so that the calculation cost of multiple times of route planning when an emergency path planning algorithm is used for processing a dynamic road network environment is reduced. The method comprises the steps of dividing sub-networks of a road network and extracting a sub-network core road network, wherein the node intermediation centrality of the road network refers to the ratio of the number of shortest paths passing through a certain node in a traffic network to the total number of the shortest paths in the whole traffic network, and the value is used for quantifying the importance of the node in the road network. The higher the center measurement of the medium is, the stronger the conversion capability of the traffic junction of the node in the whole road network is, and the better the evacuation capability and the trafficability of the vehicle are. The optimization formula of the node mediation centrality is expressed as follows:
ε is the total number of shortest paths in the statistical road network, Sp(S, t) represents the shortest path from node S to node t, Sp(s,t|vi) Representing a transit node viShortest path, Z' (v)i) Denoted as node viCentral intermediaries in the traffic network. And according to the fact that the nodes with high intermediacy in the road network are used as central nodes of the dichotomy clustering, the nodes with the adjacent intermediacy as the subordinates of the nodes are classified into the sub-road network. And (4) performing contraction operation on nodes with lower intermediacy in the sub-network, and reserving core nodes in the road network to form a road network core network. The method comprises the steps of carrying out layered processing on a road network to generate a road network layer with a plurality of sub-regions, searching next optimal adjacent nodes on different road network layers according to the sub-regions where the current nodes and target nodes are located to expand, adjusting in real time to search the optimal nodes on the road network layers in different grade ranges according to the conditions of the road network layers where the search nodes and the target nodes are located, and continuously reducing the number of nodes and the number of roadsThe purpose of finding the optimal path is achieved by the network scale.
In a traffic network with constantly changing situation, the optimal route under the current environment needs to be calculated and planned for many times in the emergency route planning process due to the surrounding changing situation environment during the driving process of the vehicle. However, such repeated routing will cause huge calculation cost and time delay of the algorithm. In order to improve the efficiency of emergency route planning, the core road network calculates and stores the connection relation Tuple between the nodes in the road network in advance. In the process of dividing each step of road network sub-network, the shortest path propagation relation between the nodes is calculated in advance according to the hierarchical relation between the nodes, and the nodes which are split from each other are divided into the sub-networks of the next layer. The traffic network forms a tree-shaped road network index structure as shown in fig. 5 after extraction, layering and situation information fusion of the core subnetworks, each layer in the tree-shaped road network represents a sub-road network at the same level, and each branch represents a sub-road network and a contracted core road network. The core road network of each branch stores a situation distance matrix between nodes contained in the core road network and a connection relation Tuple table used for recording the shortest path sequence of each node and nodes which are not directly connected with the node. When a path is searched, the idea of a 2-Hop Cover algorithm is adopted, and the shortest path between nodes is quickly addressed in a dynamically-changed road network situation environment through a layered road network structure and a constructed situation information index tree.
(3) Vehicle path planning layer
The uppermost layer is an emergency path planning layer of the vehicle, and the emergency path planning aims to enable the vehicle to leave an emergency area and reach a destination in the shortest time. Therefore, the invention provides a new situation-based road network dynamic path planning algorithm with a heuristic function for urban emergencies, the heuristic function has the advantages that the optimal solution can be quickly and effectively found, and vehicles can quickly and accurately judge whether the current road can be used as the evacuation choice in advance in a heuristic mode. The path planning layer of the invention is divided into two stages, wherein the first stage is the minimum sub-network exploration based on the situation tree of the layered road network, and the second stage is the emergency route planning based on the situation environment.
When in the first phase, the algorithm is essentially a tree search pattern, a probe algorithm developed through the hierarchy of the vehicle starting point with the smallest sub-network and the target location with the smallest sub-network. Firstly, a gradient that a vehicle starting point points to an evacuation target point is established, an algorithm explores a central node of the self-network and an upper-layer sub-network to which the self-network belongs along the direction of gradient diffusion, and a sub-network is selected according to a situation minimum distance function. The optimal path sd (t) obtained according to the situation distance is represented as:
wherein Sd (v)i,vj) Representing a distance, v, between two nodes incorporating situational informationoRepresenting slave subnet SglevelThe starting point in (1), te (t)1,t2,,…,tn) And (3) time sequence of updating the environmental situation information in the time-path network. Centralizing the subnet broker above voAnd the gradient directions are the sameMAdded to the discovery queue IQ, the owner does not explore to the destination node through neighboring nodes. When there is no node meeting the requirement in the sub-network, the upper sub-network Sg containing the sub-network is connected tolevel-1Searching until the nodes meeting the conditions are searched, and returning to the minimum-containing sub-network of the starting point and the destination node.
The second stage of the layer is route planning according to Situation information, a route planning algorithm queries the current optimal route according to an update time window of the Situation information, and therefore the invention provides a two-way Heuristic route planning algorithm BSHP (bidirectional configuration Road Network theoretical route planning) algorithm based on a Situation Road Network. The search function f (n) of the BSHP algorithm is expressed as:
F(n)＝G(n)+H(n)
g (n) of the BSHP algorithm search function represents the actual travel time required for the vehicle from the start point to the target point. lLRepresents a section of the vehicle formed by two adjacent nodes in the optimal path sd (t). Phi is a(t)The operator is a situation operator at the time t, and the value of the operator is changed along with the periodic updating of situation information.
For a vehicle arriving on a section lLThe road network speed of the road section at the starting point. G (n) may be expressed as a summation of the ratio of all the segment lengths in the path to the road speed at which the segments are reached. H (n) is a heuristic factor of the algorithm, and a node v of the position of the automobile at the moment t is usedoTo the target node vdBetween Ed (v) is the Euclidean distance Ed (v)o,vd) The ratio to the average running speed of the vehicle. The selection of the road by the BSHP algorithm is based on the sum of the actual time of travel plus the estimated time, since the presence of the g (n) function ensures that the time for the vehicle to traverse the selected route is minimal.
In conclusion, the invention provides a novel vehicle emergency navigation method under the condition that an emergency happens in a city. FIG. 5 depicts an operational flow diagram of the entire framework, first, the Beijing City road network is rasterized at equal intervals, the road speed is predicted through the historical speed data of the road network, and the commuting capability of the grid is calculated through the topological structure of the grid road network with space-time characteristics and the communication relation, so as to establish a vehicle emergency safe evacuation area recommendation model; establishing an index tree structure for updating road network situation information through the hierarchical processing of the centrality of the intermediaries of the road network; finally, an optimal path query method of a road network in a time-varying situation environment and an emergency path planning algorithm with a heuristic method are provided. The invention enables the vehicle to drive from the emergency area to the safe area in the shortest time, achieves good vehicle evacuation effect and enables the route planned by the emergency navigation of the vehicle to be more accurate and reliable.
Drawings
FIG. 1 is a node display diagram of a minimum coordinate point of an XY axis of a planar coordinate system of a road network;
FIG. 2 is a schematic diagram of prediction of a time-space sequence of speed of a road network;
FIG. 3 is a GLN model diagram of a hybrid neural network;
FIG. 4 is a schematic diagram of a grid network;
FIG. 5 is a schematic diagram of a tree index structure of the road network;
FIG. 6 is a flow chart of the operation of the framework;
FIG. 7 is a schematic diagram of a gridded display of a five-ring inner network of Beijing;
FIG. 8 is a histogram of travel times for an algorithmically planned route;
FIG. 9 is an algorithmic runtime line graph under different trellis divisions;
FIG. 10 is a line graph of algorithm access node number under different trellis divisions;
fig. 11 is a line graph of evacuation time under different grid divisions.
Detailed Description
The method has the advantages that the test is developed and realized through Java language, the Tencent map is crawled to obtain the historical speed data of the road network, the road network data come from OSM, effective road network rasterization and road network speed prediction are realized through experiments, the road network data when the city has an emergency are simulated, and the effectiveness of the framework is verified through comparing other path planning algorithms. Four groups of experiments are arranged, and the first experiment verifies the effectiveness of the road network prediction model GLN; calculating the grid evacuation capacity by the experimental two-pass through a DGPR algorithm; experiment three is to verify the high efficiency of the BSHP in a changed situation road network environment by comparing with other path planning methods; the effectiveness of the invention is verified by simulation and simulation of the experiment.
The method comprises the steps of firstly converting road network data into a coordinate system, obtaining plane coordinate data corresponding to nodes in the road network by using Gaussian projection forward calculation, and dividing the road network data into n equal parts of five-ring internal road networks in Beijing City according to coordinatesgAn equally spaced grid of 18, 36, 72, 144, 288, 576. FIG. 7 shows different grades of the five-ring intranet in BeijingAnd (5) a gridding effect graph. Firstly, a road network road speed prediction model is evaluated, and the efficiency of the GLN model provided by the invention is verified by comparing other traditional methods with the existing road network speed model. The effectiveness of the model is verified on a real road network historical speed data set, the road network historical speed is from road network speed data of a detector near Tokyo four rings from No. 9 and No. 1 to No. 11 and No. 1 in 2020, 5 minutes are averaged, the road network speed data of the previous month is taken as training data and the speed data of the last month is taken as test data in the experiment, the road network historical time window input by the model is 60 minutes, namely 12 road network speed data observed values are input (M is 12), and the road speeds (H is 3, 6 and 9) in the future 15 minutes, 30 minutes and 45 minutes are predicted. The baseline model of the experiment selects a fully connected neural network (FC-LSTM), a gated cyclic unit model (GRU), a support vector machine regression (SVR), an autoregressive moving average model (ARIMA), a historical average model (HA) and a space-time convolution network (STGCN). Table 1 shows a comparison of the GLN model and other baseline methods on the road network speed data set for 15 min, 30 min, 45 min predicted performance. As can be seen, the GLN model obtains the best prediction performance under all evaluation indexes of almost all prediction horizons, and the effectiveness of the time-space traffic prediction model GLN is proved.
TABLE 1 comparison of Performance of different predictive models on a historical speed dataset for a road network
Table 1 shows the overall performance evaluation of all prediction models in this experiment, in which the results of different prediction steps H were compared. From the results, the GLN model provided by the invention realizes the minimum prediction error on three different prediction step sizes and most of the evaluated indexes. In addition to the overall advantages of the GLN model, it was found that using the conventional predictive model performed well in short term predictions, but errors were generated due to its inability to efficiently capture and analyze nonlinear spatio-temporal correlations between road network speeds, and also due to the lack of modeling of sensor network spatial correlations in the conventional approach. And the accumulation of errors along with the increase of the number of steps makes the long-term prediction value of the errors more inaccurate. Unlike the traditional model, the STGCN model using graph convolution and the GLN model provided by the invention have obvious improvement on the short-term and long-term performance of the prediction of the road network speed by analyzing and modeling the sensor space topological relation. In addition, the GLN model provided by the invention considers three different time relations in time series prediction, so that the performance is improved greatly. However, with the increase of the prediction step number H, the precision of the GLN model is reduced to some extent, but the performance is still slightly superior to that of other prediction models, and the effectiveness of the GLN model is reflected.
The vehicle emergency evacuation area recommendation experiment is performed based on a five-ring internal grid diagram in Beijing city shown in FIG. 7. And calculating the evacuation capacity of each grid containing the nodes of the road network according to a formula. Value 1 is specified for the initial evacuation capacity Value of each grid by calculating the number of nodes and the number of ingress and egress edges included in each grid. For example, when a grid contains 5 edges pointing to the grid, then the initial weight of each edge is 1/5. In the iteration of the grid evacuation capacity calculation algorithm, the normalized road speed is reassigned to each edge until the rank Value converges. And updating the speed information of the grid network after the time interval of t, recalculating the grid grade value and sequencing the grid grade value. With n g144 and ngTable 2 shows the grid evacuation capability values and ranking results calculated by the algorithm, as an example of the grid map of 288.
TABLE 2 commuting ability values and ranking results of the grid
Table 2 shows the results of the sorting of the grid evacuation capacity in descending order. Wherein the Rank value represents the grid's energy per dispersionForce is at the ranking Value of the entire grid map, Value represents the evacuation capacity of the grid as calculated by the DGPR algorithm. For example, when n g144, the Grid with the strongest commuting capabilityid＝80-65，n g288, the strongest Commuting capability Gridid28-91. When an emergency occurs in a city, vehicles in the vicinity need to be emergently evacuated. The evacuation capacity of a grid reflects the importance of the grid in the road network. The larger the Value of the grid, the stronger its ability to unblock vehicles in the road network. Therefore, if the grid with a higher grade value is used as an evacuation area, the evacuation effect of the emergency navigation is better.
The road network situation information acceleration algorithm experiment carries out comparison tests on a bidirectional heuristic path planning algorithm BSHP based on the situation road network, a Bi-Directional Dijkstra algorithm, an A algorithm and a dynamic A algorithm. In order to compare performance of the algorithm in road network environment influence of different situations, road network data of a Beijing city road network at different moments are selected for experimental setting. As shown in table 3, path planning was performed for 8:30 am and 14:00 pm, respectively. The path planning in the dynamically changing situation environment road network can be regarded as the prediction of the future path node sequence, so the root mean square error RMSE and the Accuracy Accuracy can be used as the evaluation indexes of the comparison experiment. Two criteria are defined as follows:
w in the formulaLAnd the situation-based path weight between the starting point and the end point of the vehicle is calculated through the posterior situation of the final path. OmegaLRepresenting the actual road situation weight. And n is the number of node pairs counted in the situation road network.
TABLE 3 Experimental results of different path planning methods on situation road network
For the traditional path planning algorithm, the BSHP algorithm provided by the invention analyzes the selection of the path from the dimensions of time and space. The method is characterized in that the same vehicle starting point and the same vehicle finishing point are selected from a Beijing city road network, the situation environment of the road network is changed through different time points, and the road network data in the peak time period is selected to simulate the traffic jam caused by an emergency. As can be seen from Table 3, the BSHP proposed by the present invention is superior to other algorithms in both the RMSE index and the Accuracy index at different query times. The CH-BiDijkstra algorithm and the BSHP algorithm both contain nodes which are visited by the router node contraction process and are much smaller in number than other algorithms, the BSHP algorithm not only contracts the nodes and searches for the minimum contained subnet to further reduce the whole range, but also explores the number of nodes which is much smaller than the CH-BiDijkstra algorithm. BSHP also has the best performance in terms of algorithm runtime. The histogram of vehicle travel times for the routes planned by the above five algorithms during peak and off-peak hours is shown in fig. 8, from which it can be seen that the planned routes, while less dominant during off-peak hours, provide less time-consuming routes in the event that the road network is congested during peak hours.
And finally, performing a simulation experiment, and verifying the effectiveness of the invention by simulating the vehicle evacuation time in emergency areas in six different equidistant grid networks. The experiment is carried out by carrying out comparison tests under different evacuation distances and different evacuation vehicles. The comparative experiments were carried out at distances of 6.8km, 15.6km and 22.1km respectively at different evacuation distances. Experiments compare the efficiency of the BSHP algorithm in planning different raster network paths. The experimental results are shown in fig. 8 and fig. 9, and it can be seen that the running time of the algorithm increases with the number of grids, and the number of nodes visited when the algorithm searches for a path is continuously reducedThe reason is that as the number of grids increases, the number of nodes containing nodes in each grid decreases, and the algorithm runs for ngCompare n when 576g288 is increased because the algorithm increases the number of decisions between grids when the number of grids is excessive, resulting in increased run time. Evacuation was simulated by simulating 100, 200, 500 vehicles in experiments with different numbers of evacuation vehicles (vehicles evacuated from their location to the grid recommended by the algorithm). Evacuation time for evacuating different numbers of vehicles as shown in fig. 10, the time consumed for evacuating the same vehicle becomes shorter and shorter as the number of grids increases, but when n isgThe increase in evacuation at 576 was due to the algorithm exploring more grids as a result of the increased number of grids. The evacuation time increases slightly with increasing number of evacuation vehicles, because the road's inherent traffic capacity results in a certain maximum throughput of the road at the same time.
The four groups of experiments show that the model and the algorithm provided by the invention can be used for emergency navigation of urban road vehicles, and the problem of path planning in a time-varying situation road network environment is solved. The evacuation area recommendation is realized by predicting the speed of the road network and calculating the evacuation capacity of the grids; layering road network layers, establishing a situation information tree and an optimal path search index to realize rapid road network situation information updating and path search; finally, a heuristic bidirectional path planning algorithm based on the space-time characteristic situation road network is provided, vehicle emergency route planning is carried out under the time-varying situation road network environment, and the timeliness and the accuracy of the algorithm are higher than those of other path planning algorithms. The method and the system fill up the defect of emergency navigation of the vehicle under the emergency background by the path planning algorithm.
Claims (6)
1. A vehicle emergency navigation method with spatio-temporal characteristic situation information is characterized in that a vehicle emergency navigation general architecture for realizing the method consists of three parts, namely a vehicle evacuation area recommendation layer, a road network data index layer and a vehicle path planning layer from bottom to top; the method is characterized in that: firstly, mapping the road network to a two-dimensional plane through the lowest vehicle evacuation area recommendation layerIn the plane coordinate system, the equal-spacing grid division is carried out based on the two-dimensional plane coordinate system, so that the map after the grid division processing is divided into ng×ngThe grid predicts the speed of roads in the road network by a neural network method, calculates the evacuation capacity of each grid by combining the predicted speed of the road network and the communication relation between the grids, and recommends the area which is most suitable for vehicle evacuation under the current condition to the user; secondly, the road network data index layer calculates the intermediary centrality of the nodes in the traffic network by carrying out subnet extraction layering operation on the traffic network, and contracts the nodes with lower intermediary centrality measurement in the sub-network; then, situation data information acquired by various sensors in the road network is fused into situation analysis of the road network, and a road network situation information tree index is constructed; and finally, on a vehicle path planning layer, combining the processing results of the previous two layers on the road network, and realizing the planning of the vehicle emergency route by the recommended evacuation area and the analysis of the situation information of the road.
2. The vehicle emergency navigation method with spatiotemporal characteristic situational information according to claim 1, characterized in that: in the vehicle evacuation area recommendation layer, recommending grids with higher commuting capacity within a certain range of the emergency occurrence point to users as evacuable grids by calculating the commuting capacity of each grid in the vector grid road network; performing rasterization processing on road network data by using an equal interval rasterization method, so that the road network data can be divided into a plurality of region sub-blocks by the size of an equal interval; due to the topology and the communication characteristics of a traffic network, the road congestion can be diffused from a few intersections to a plurality of intersections which are mutually connected, and regional congestion is formed; the regional congested intersection is necessarily an intersection connected with a plurality of continuous adjacent roads in geographic position; the area formed by the intersections can be defined as a congestion area; the number of intersections and the number of entrances and exits are large, and the grids with high road network speed obtain higher commuting capacity and sequencing values; and judging the evacuation capacity of a certain grid in the whole grid network by analyzing intersection nodes contained in the grid, the entrance and exit degree of the grid and the communication relation between the grids.
3. The vehicle emergency navigation method with spatiotemporal characteristic situational information according to claim 2, characterized in that: converting longitude and latitude coordinates of nodes in the road network data into geodetic coordinates, obtaining plane coordinate data corresponding to the nodes in the road network by Gaussian projection, and expressing a node plane coordinate set as Coordxy＝{(x1,y1),(x2,y2),…(xi,yi) }; firstly, determining the position of the origin of coordinates, and finding out the node coordinate (X) with the minimum X-axis value through the node plane coordinate datao,yo) Coordinate (x) of node having minimum value with Y axisa,ya) Two points satisfy x respectivelyo＜xothersAnd ya＜yothers(ii) a Translating the minimum point of the X-axis coordinate value to make the abscissa zero to obtain (0, y)o) Translating the minimum point of the Y-axis coordinate value to make the ordinate zero (x)a0), and translating all the points according to the translation amounts of the two points;
the grids are connected with each other through the road, the speed of the road connected between the grids is analyzed, and the communication capacity of the grids and other grids is calculated; the input of the road network speed space-time sequence prediction model is road network historical speed data collected by a vehicle speed sensor in an urban road network, and the prediction data of the road network speed is obtained through a neural network prediction model; the speed data of the network changes along with the change of time in a continuous space, so the space correlation characteristics of the speed data of the network need to be effectively extracted; a road network speed prediction hybrid model GLN combining GCN and LSTM; three relevance of road network speed in a time plane are adopted: proximity, periodicity and trend, and the LSTM network is adopted to respectively capture three time dependencies in road network speed data, so that more accurate prediction of road speed is realized; original time series X of road network speed data0＝(x1,x2,…,xm) Normalization was performed to obtain the sequence X' ═ { X1’,x2’,…,xn' }, three subdata sequences and time sequences obtained by reconstructing the normalized road network speed dataX’Tw、X’TdAnd X'TmAs inputs for proximity sequence, periodic sequence, and trend sequence components;
X'Tm＝(x'Tm1,x'Tm2,...,x'Tm(n-1))T
X'Td＝(x'Td1,x'Td2,...,x'Td(n-1))T
X'Tw＝(x'Tw1,x'Tw2,...,x'Tw(n-1))T
the input of the road network speed prediction model for training the time sequence is the three sequences; the theoretical output corresponding to the training set is Y ═ Y1,Y2,…,Yn-1]T＝[x2’,x3’,…,xn’]T(ii) a According to the construction of the three subsequences, the three subsequences are firstly input into a GCN module for learning the spatial characteristics, and then the output of the GCN is used as the input of an LSTM module for learning the time dimension.
4. The vehicle emergency navigation method with spatiotemporal characteristic situational information according to claim 3, characterized in that: after rasterization and road network speed prediction are carried out on the road network, the commuting capability of the grid is calculated; the grid commuting capability calculation and sorting recommendation algorithm is a dynamic grid page sorting algorithm improved on the basis of a traditional page sorting algorithm; a directed graph is considered between the web pages, and the grid road network is divided into 2 multiplied by 2 grids; one grid in the grid network is abstractly regarded as a webpage, and the mutually communicated roads among the grids are regarded as links among the webpages; therefore, a link analysis algorithm is carried out on the rasterized road network; the rank value for each grid is expressed as:
Pri＝∑Iej*d*(Rs/120)
wherein, DGPR (Grid)id) Is the ordering value of the grid after iteration of the DGPR algorithm; this value represents the commuting capacity of the grid; n represents the number of grids containing nodes; pr (Pr) ofiIs the DGPR value of the ith trellis, i represents the ID of the trellis; vGrididIs GrididV is the total number of vertices in the entire network; the damping coefficient d is 0.85; iejRepresents the in-degree of the jth grid; (Rs/120) represents normalized road speed; when the vehicles are needed to be evacuated due to serious congestion or emergency situations in the road network, recommending the most appropriate evacuation area to the user according to the road condition at the moment of emergency situation.
5. The vehicle emergency navigation method with spatiotemporal characteristic situational information according to claim 1, characterized in that: in the road network data index layer, a road network data situation index is to be constructed; firstly, performing layering processing on a road network, then integrating situation information in the road network into the layered road network, constructing a situation information tree index based on the layered road network, and simplifying the structure of the road network on the premise of ensuring the integrity of the road network through reasonable contraction of nodes; dividing sub-networks of a road network and extracting a core road network of the sub-networks, wherein the node intermediation centrality of the road network refers to the ratio of the number of shortest paths passing through a certain node in a traffic network to the total number of the shortest paths in the whole traffic network, and is used for quantifying the importance of the node in the road network; the optimization formula of the node mediation centrality is expressed as follows:
ε is the total number of shortest paths in the statistical road network, Sp(S, t) represents the shortest path from node S to node t, Sp(s,t|vi) Representing a transit node viShortest path, Z' (v)i) Denoted as node viCentral intermediacy metrics in the traffic network; taking a node with high intermediacy in the road network as a central node of the dichotomy cluster, and classifying adjacent nodes as auxiliary nodes into a sub-road network; shrink operations are taken on less-intervening nodes in the sub-network, andreserving core nodes in a road network to form a road network core network; the method comprises the steps of carrying out layered processing on a road network to generate a road network layer with a plurality of sub-regions, searching next optimal adjacent nodes on different road network layers for expansion according to the sub-regions where the current nodes and target nodes are located, adjusting in real time to search optimal nodes on the road network layers in different grade ranges according to the conditions of the road network layer where the search nodes and the target nodes are located, and achieving the purpose of searching optimal paths by continuously reducing the number of nodes and the scale of the road network;
in a traffic network with constantly changing situation, the optimal route under the current environment needs to be calculated and planned for many times in the emergency route planning process due to the situation environment changing around in the driving process of the vehicle; the method comprises the steps that a core road network calculates and stores a connection relation Tuple between nodes in the road network in advance; in the process of dividing each step of road network sub-network, the shortest path propagation relation between the nodes is calculated in advance according to the hierarchical relation between the nodes, and the nodes which are split from each other are divided into the sub-networks of the next layer; extracting, layering and fusing situation information of a traffic network through a core sub-network to form a tree network index structure, wherein each layer in the tree structure represents a sub-network at the same level, and each branch represents a sub-network and a contracted core network; the core road network of each branch stores a situation distance matrix between nodes contained in the core road network and a connection relation Tuple table for recording the shortest path sequence of each node and nodes which are not directly connected with the node; when a path is searched, the idea of a 2-Hop Cover algorithm is adopted, and the shortest path between nodes is quickly addressed in a dynamically-changed road network situation environment through a layered road network structure and a constructed situation information index tree.
6. The vehicle emergency navigation method with spatiotemporal characteristic situational information according to claim 1, characterized in that: in the vehicle path planning layer, the uppermost layer is an emergency path planning layer of the vehicle, a new situation road network-based dynamic path planning algorithm with a heuristic function is provided for urban emergency, the heuristic function has the advantage of quickly and effectively finding the optimal solution, and the vehicle can quickly and accurately judge whether the current road is used as the evacuation selection in advance in a heuristic mode; the path planning layer is divided into two stages, the first stage is minimum sub-network exploration based on a layered road network situation tree, and the second stage is emergency route planning based on a situation environment;
when the vehicle is in the first stage, a detection algorithm is developed through the hierarchy of the minimum sub-networks of the vehicle starting point and the minimum sub-networks of the target location; firstly, establishing a gradient that a vehicle starting point points to an evacuation target point, searching a central node of the self-network and an upper-layer sub-network to which the central node belongs along the gradient diffusion direction by an algorithm, and selecting a sub-network according to a situation minimum distance function; the optimal path sd (t) obtained according to the situation distance is represented as:
wherein Sd (v)i,vj) Representing a distance, v, between two nodes incorporating situational informationoRepresenting slave subnet SglevelThe starting point in (1), te (t)1,t2,,…,tn) Time sequence of updating environmental situation information in the time-path network; centralizing the subnet broker above voAnd the gradient directions are the sameMAdding the data into a discovery queue IQ, and mainly not exploring to a destination node through adjacent nodes; when there is no node meeting the requirement in the sub-network, the upper sub-network Sg containing the sub-network is connected tolevel-1Searching until the nodes meeting the conditions are searched, and returning the minimum sub-network of the starting point and the destination node;
the second stage is route planning according to situation information and a bidirectional heuristic path planning algorithm BSHP algorithm based on a situation road network; the search function f (n) of the BSHP algorithm is expressed as:
F(n)＝G(n)+H(n)
g (n) of the BSHP algorithm search function represents the actual travel time required for the vehicle from the starting point to the target point; lLRepresents a section formed by two adjacent nodes of the vehicle in the optimal path SD (t);
the operator is a situation operator at the time t, and the value of the operator is changed along with the periodic update of situation information;
for a vehicle arriving on a section lLThe road network speed of the road section at the starting point; g (n) is expressed as the sum of the ratio of all the road segment lengths in the path to the road network speed when the road segment is reached; h (n) is a heuristic factor of the algorithm, and a node v of the position of the automobile at the moment t is usedoTo the target node vdBetween Ed (v) is the Euclidean distance Ed (v)o,vd) Expressed as a ratio to the average travel speed of the vehicle; the selection of the road by the BSHP algorithm is based on the sum of the actual time of travel plus the estimated time, since the presence of the g (n) function ensures that the time for the vehicle to traverse the selected route is minimal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011495841.6A CN112797995B (en) | 2020-12-17 | 2020-12-17 | Vehicle emergency navigation method with space-time characteristic situation information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011495841.6A CN112797995B (en) | 2020-12-17 | 2020-12-17 | Vehicle emergency navigation method with space-time characteristic situation information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112797995A true CN112797995A (en) | 2021-05-14 |
| CN112797995B CN112797995B (en) | 2023-02-28 |
Family
ID=75806864
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011495841.6A Active CN112797995B (en) | 2020-12-17 | 2020-12-17 | Vehicle emergency navigation method with space-time characteristic situation information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112797995B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113282093A (en) * | 2021-07-21 | 2021-08-20 | 中国科学院自动化研究所 | Robot navigation method, device, electronic equipment and storage medium |
| CN113903171A (en) * | 2021-09-27 | 2022-01-07 | 北京航空航天大学 | Vehicle crowd sensing node optimization method based on spatial-temporal characteristics of highway network |
| CN114023093A (en) * | 2021-08-29 | 2022-02-08 | 北京工业大学 | Dynamic evacuation method for multiple vehicles smoothly driven in short distance |
| CN114419917A (en) * | 2021-12-15 | 2022-04-29 | 电子科技大学 | Traffic jam grooming method and system based on single-direction graph |
| CN114997297A (en) * | 2022-05-26 | 2022-09-02 | 哈尔滨工业大学 | Target movement intention reasoning method and system based on multistage regional division |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005321323A (en) * | 2004-05-10 | 2005-11-17 | Alpine Electronics Inc | Navigation system and route guiding method |
| CN103337162A (en) * | 2013-07-16 | 2013-10-02 | 四川大学 | Real-time planning and dynamic scheduling system for urban emergency rescue channel |
| CN109900276A (en) * | 2019-04-01 | 2019-06-18 | 河北工业大学 | The real-time emergency route planing method in station based on point-line-surface barrier model construction |
| CN110887495A (en) * | 2019-11-28 | 2020-03-17 | 深圳市海力特科技有限责任公司 | Method for applying real-time road conditions of cloud platform to urban emergency GIS platform |
| CN110907966A (en) * | 2019-11-22 | 2020-03-24 | 东华理工大学 | Emergency vehicle navigation system and method based on real-time traffic flow in time of Internet of things |
| CN111524353A (en) * | 2020-04-28 | 2020-08-11 | 中国计量大学 | Method for traffic text data for speed prediction and travel planning |
-
2020
- 2020-12-17 CN CN202011495841.6A patent/CN112797995B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005321323A (en) * | 2004-05-10 | 2005-11-17 | Alpine Electronics Inc | Navigation system and route guiding method |
| CN103337162A (en) * | 2013-07-16 | 2013-10-02 | 四川大学 | Real-time planning and dynamic scheduling system for urban emergency rescue channel |
| CN109900276A (en) * | 2019-04-01 | 2019-06-18 | 河北工业大学 | The real-time emergency route planing method in station based on point-line-surface barrier model construction |
| CN110907966A (en) * | 2019-11-22 | 2020-03-24 | 东华理工大学 | Emergency vehicle navigation system and method based on real-time traffic flow in time of Internet of things |
| CN110887495A (en) * | 2019-11-28 | 2020-03-17 | 深圳市海力特科技有限责任公司 | Method for applying real-time road conditions of cloud platform to urban emergency GIS platform |
| CN111524353A (en) * | 2020-04-28 | 2020-08-11 | 中国计量大学 | Method for traffic text data for speed prediction and travel planning |
Non-Patent Citations (2)
| Title |
|---|
| BOWEN YANG等: "A Novel Urban Emergency Path Planning Method Based on Vector Grid Map", 《IEEE ACCESS》 * |
| 王杰: "基于交通态势认知的动态路径推荐方法研究与仿真", 《中国优秀硕士学位论文全文数据库（电子期刊）》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113282093A (en) * | 2021-07-21 | 2021-08-20 | 中国科学院自动化研究所 | Robot navigation method, device, electronic equipment and storage medium |
| CN114023093A (en) * | 2021-08-29 | 2022-02-08 | 北京工业大学 | Dynamic evacuation method for multiple vehicles smoothly driven in short distance |
| CN114023093B (en) * | 2021-08-29 | 2024-03-15 | 北京工业大学 | Multi-vehicle dynamic evacuation method driven smoothly in short distance |
| CN113903171A (en) * | 2021-09-27 | 2022-01-07 | 北京航空航天大学 | Vehicle crowd sensing node optimization method based on spatial-temporal characteristics of highway network |
| CN113903171B (en) * | 2021-09-27 | 2022-07-15 | 北京航空航天大学 | Vehicle crowd sensing node optimization method based on spatial-temporal characteristics of highway network |
| CN114419917A (en) * | 2021-12-15 | 2022-04-29 | 电子科技大学 | Traffic jam grooming method and system based on single-direction graph |
| CN114997297A (en) * | 2022-05-26 | 2022-09-02 | 哈尔滨工业大学 | Target movement intention reasoning method and system based on multistage regional division |
| CN114997297B (en) * | 2022-05-26 | 2024-05-03 | 哈尔滨工业大学 | Target movement intention reasoning method and system based on multi-level region division |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112797995B (en) | 2023-02-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112797995B (en) | Vehicle emergency navigation method with space-time characteristic situation information | |
| Jagadeesh et al. | Heuristic techniques for accelerating hierarchical routing on road networks | |
| JP4705176B2 (en) | Route finding system | |
| CN109959388B (en) | Intelligent traffic refined path planning method based on grid expansion model | |
| CN102538806B (en) | A kind of paths planning method and relevant device | |
| CN112629533A (en) | Refined path planning method based on road network rasterization road traffic flow prediction | |
| CN110516702B (en) | Discrete path planning method based on streaming data | |
| CN114463977A (en) | Path planning method based on vehicle-road collaborative multi-source data fusion traffic flow prediction | |
| Ai et al. | Generation of constrained network Voronoi diagram using linear tessellation and expansion method | |
| CN115409256B (en) | Route recommendation method for avoiding congestion area based on travel time prediction | |
| CN110836675A (en) | Decision tree-based automatic driving search decision method | |
| Liu et al. | Polestar: An intelligent, efficient and national-wide public transportation routing engine | |
| Yang et al. | A novel urban emergency path planning method based on vector grid map | |
| CN112991743A (en) | Real-time traffic risk AI prediction method based on driving path and system thereof | |
| CN113865589A (en) | Long-distance rapid path planning method based on terrain slope | |
| Terekhov et al. | Forecasting a global air passenger demand network using weighted similarity-based algorithms | |
| Yin et al. | A totally Astar-based multi-path algorithm for the recognition of reasonable route sets in vehicle navigation systems | |
| CN114578848A (en) | Unmanned aerial vehicle routing inspection path planning method based on discrete point density and global planning | |
| CN112797997B (en) | Emergency path planning architecture and method based on grid road network | |
| Menguc et al. | A data driven approach to forecasting traffic speed classes using extreme gradient boosting algorithm and graph theory | |
| CN111797331B (en) | Multi-target multi-constraint route recommendation method based on crowd sensing | |
| CN116542709A (en) | Electric vehicle charging station planning analysis method based on traffic situation awareness | |
| Hu et al. | SOUP: A fleet management system for passenger demand prediction and competitive taxi supply | |
| CN116187851A (en) | Bayesian-based automatic driving road and vehicle road matching risk index assessment method | |
| CN115293427A (en) | Planning method and system for optimal travel route of city early peak |
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 |