A test route generation method, test method and device based on semantic map

Abstract

The present invention discloses a test route generation method, a test method and a device based on a semantic map. The method comprises: loading a semantic map according to a test area to obtain a test map; for each intersection on the test map, pairing the entry navigation point of the intersection with the exit navigation point of the intersection that it can reach one by one to generate a plurality of navigation point pairs of each intersection; traversing a plurality of navigation point pairs of each intersection, matching each navigation point pair of the current intersection with each navigation point pair of the next intersection that it can reach, and obtaining a plurality of first matching pairs; taking each navigation point pair of each intersection as a node, and adding connecting edges between two navigation point pairs in each first matching pair to construct a directed graph; connecting all connecting edges according to the directed graph to connect all nodes in series to generate a test route. The present invention can automatically generate a test route by fully covering the intersections in the semantic map test area, which is conducive to improving the test efficiency of the semantic map.

Description

Technical Field

[0001] The present invention relates to the technical field of road testing, and in particular to a test route generation method based on a semantic map, a test method and a test device. Background Art

[0002] At present, engineers mainly mark the high-precision maps collected by vehicles to generate semantic maps, design and create test routes based on the semantic maps, arrange vehicles for autonomous driving tests, and find out potential erroneous markings in the semantic maps, such as lane line omissions and marking errors. This method of manually generating test routes not only consumes a lot of manpower, but also the generated test routes are difficult to fully cover the semantic maps, resulting in low efficiency in semantic map testing. Summary of the invention

[0003] In order to overcome the defects of the prior art, the present invention provides a test route generation method, a test method and a device based on a semantic map, which can automatically generate a test route for intersections in a semantic map test area, thereby improving the test efficiency of the semantic map.

[0004] In order to solve the above technical problems, in a first aspect, an embodiment of the present invention provides a test route generation method based on a semantic map, comprising:

[0005] Load the semantic map according to the test area to obtain the test map;

[0006] For each intersection on the test map, the entry navigation point of the intersection is matched with the exit navigation point of the accessible intersection one by one to generate several navigation point pairs for each intersection;

[0007] Traversing a plurality of navigation point pairs at each intersection, matching each navigation point pair at the current intersection with each navigation point pair at the next intersection that can be reached, and obtaining a plurality of first matching pairs;

[0008] Each navigation point pair at each intersection is taken as a node, and a connecting edge is added between two navigation point pairs in each first matching pair to construct a directed graph;

[0009] Connect all the connected edges according to the directed graph to connect all the nodes in series and generate a test route.

[0010] Furthermore, the test route generation method based on the semantic map further includes:

[0011] When the navigation point pair of the current intersection reaches the navigation point pair of the next intersection through the boundary of the test area, the navigation point pair of the current intersection is matched with each navigation point pair of the target intersection that it can reach, to obtain a plurality of second matching pairs; wherein the distance between the current intersection and the target intersection is less than the preset out-of-domain distance;

[0012] Each navigation point pair at each intersection is taken as a node, and a connecting edge is added between the two navigation point pairs in each first matching pair, and a connecting edge is added between the two navigation point pairs in each second matching pair to construct a directed graph.

[0013] Furthermore, for each intersection on the test map, the entry navigation point of the intersection is paired with the exit navigation point of the reachable intersection one by one to generate several navigation point pairs for each intersection, specifically:

[0014] For each intersection on the test map, the entry navigation point of the intersection is paired with the exit navigation point of the reachable intersection one by one to generate several initial navigation point pairs, and one initial navigation point pair is selected from all the initial navigation point pairs on each edge of the intersection as the navigation point pair to obtain several navigation point pairs.

[0015] Furthermore, the method of connecting all the connection edges according to the directed graph to connect all the nodes in series to generate a test route is specifically as follows:

[0016] Based on the directed graph, several navigation point pairs of each intersection are traversed to determine whether the navigation point pair of the current intersection does not reach the navigation point pair of another intersection after passing through the boundary of the test area. If so, the navigation point pair of the current intersection is removed to obtain an intermediate graph;

[0017] Based on the intermediate graph, traverse a number of navigation point pairs at each intersection, match each navigation point pair at the current intersection with each navigation point pair at another intersection it reaches, obtain multiple third matching pairs, and add connecting edges between two navigation point pairs in each third matching pair to generate a complete graph;

[0018] According to the complete graph, all the connected edges are connected to connect all the nodes in series to generate a test route.

[0019] Further, after connecting all the connection edges according to the directed graph to connect all the nodes in series to generate a test route, the method further includes:

[0020] According to the speed limit information of the test map and the length of the test route, the test route is split into a number of test sub-routes whose test duration is less than a preset test duration, to obtain a test route set.

[0021] Further, after the semantic map is loaded according to the test area to obtain the test map, before the entry navigation point of the intersection is matched with the exit navigation point of the reachable intersection for each intersection on the test map one by one to generate a plurality of navigation point pairs for each intersection, the method further includes:

[0022] Each intersection in the semantic map is traversed, and the intersections in the test area are collected into the intersection set of the test map.

[0023] In a second aspect, an embodiment of the present invention provides a test route generation device based on a semantic map, comprising:

[0024] A test map acquisition module is used to load a semantic map according to a test area to obtain a test map;

[0025] A navigation point pair generation module is used to match the entry navigation point of the intersection with the exit navigation point of the reachable intersection for each intersection on the test map, and generate a number of navigation point pairs for each intersection;

[0026] A navigation point pair matching module, used to traverse a plurality of navigation point pairs at each intersection, match each navigation point pair at the current intersection with each navigation point pair at the next intersection that can be reached, and obtain a plurality of first matching pairs;

[0027] A directed graph construction module, used to take each navigation point pair at each intersection as a node and add a connecting edge between two navigation point pairs in each first matching pair to construct a directed graph;

[0028] The test route generation module is used to connect all the connection edges according to the directed graph to connect all the nodes in series and generate a test route.

[0029] In a third aspect, an embodiment of the present invention provides a testing method based on a semantic map, comprising:

[0030] Load the semantic map according to the test area to obtain the test map;

[0031] For each intersection on the test map, the entry navigation point of the intersection is matched with the exit navigation point of the accessible intersection one by one to generate several navigation point pairs for each intersection;

[0032] Traversing a plurality of navigation point pairs at each intersection, matching each navigation point pair at the current intersection with each navigation point pair at the next intersection that can be reached, and obtaining a plurality of first matching pairs;

[0033] Each navigation point pair at each intersection is taken as a node, and a connecting edge is added between two navigation point pairs in each first matching pair to construct a directed graph;

[0034] Connect all the connected edges according to the directed graph to connect all the nodes in series and generate a test route;

[0035] A test route is sent to the vehicle so that the vehicle can perform autonomous driving tests according to the test route, and the test map is corrected based on the acquired test data.

[0036] Furthermore, the semantic map-based testing method further includes:

[0037] When the navigation point pair of the current intersection reaches the navigation point pair of the next intersection through the boundary of the test area, the navigation point pair of the current intersection is matched with each navigation point pair of the target intersection that it can reach, to obtain a plurality of second matching pairs; wherein the distance between the current intersection and the target intersection is less than the preset out-of-domain distance;

[0038] Each navigation point pair at each intersection is taken as a node, and a connecting edge is added between the two navigation point pairs in each first matching pair, and a connecting edge is added between the two navigation point pairs in each second matching pair to construct a directed graph.

[0039] In a fourth aspect, an embodiment of the present invention provides a testing device based on a semantic map, comprising:

[0040] A test map acquisition module is used to load a semantic map according to a test area to obtain a test map;

[0041] A navigation point pair generation module is used to match the entry navigation point of the intersection with the exit navigation point of the reachable intersection for each intersection on the test map, and generate a number of navigation point pairs for each intersection;

[0042] A navigation point pair matching module, used to traverse a plurality of navigation point pairs at each intersection, match each navigation point pair at the current intersection with each navigation point pair at the next intersection that can be reached, and obtain a plurality of first matching pairs;

[0043] A directed graph construction module, used to take each navigation point pair at each intersection as a node and add a connecting edge between two navigation point pairs in each first matching pair to construct a directed graph;

[0044] A test route generation module is used to connect all the connection edges according to the directed graph to connect all the nodes in series and generate a test route;

[0045] The autonomous driving test module is used to send a test route to the vehicle, enable the vehicle to perform autonomous driving tests according to the test route, and correct the test map based on the acquired test data.

[0046] The embodiments of the present invention have the following beneficial effects:

[0047] By loading the semantic map according to the test area, the test map is obtained. For each intersection on the test map, the entry navigation point of the intersection is matched with the exit navigation point of the intersection that can be reached, and a number of navigation point pairs of each intersection are generated. The navigation point pairs of each intersection are traversed, and each navigation point pair of the current intersection is matched with each navigation point pair of the next intersection that can be reached, and multiple first matching pairs are obtained. Each navigation point pair of each intersection is used as a node, and connecting edges are added between two navigation point pairs in each first matching pair to construct a directed graph. All nodes are connected in series by connecting all connecting edges according to the directed graph to generate a test route, and the test route is automatically generated. Compared with the prior art, the embodiment of the present invention uses a marked semantic map to traverse each intersection in the test area on the semantic map, and automatically generates a test route covering the left turn, right turn, straight, and U-turn directions of each intersection. It can fully cover the intersections in the test area of ​​the semantic map to automatically generate a test route, which is conducive to improving the test efficiency of the semantic map. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A flowchart of a method for generating a test route based on a semantic map in a first embodiment of the present invention;

[0049] Figure 2 A schematic diagram of an intersection according to an example of the first embodiment of the present invention;

[0050] Figure 3 A schematic diagram of an intersection according to an example of the first embodiment of the present invention;

[0051] Figure 4 A schematic diagram of another example of an intersection in the first embodiment of the present invention;

[0052] Figure 5 A schematic diagram of another example of an intersection in the first embodiment of the present invention;

[0053] Figure 6 Schematic diagram of the structure of a test route generation device based on a semantic map in the second embodiment of the present invention;

[0054] Figure 7 is a flowchart of a testing method based on a semantic map in the third embodiment of the present invention;

[0055] Figure 8 4 is a schematic diagram of the structure of a testing device based on a semantic map in the fourth embodiment of the present invention. DETAILED DESCRIPTION

[0056] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] It should be noted that the step numbers in the text are only for the convenience of explaining the specific embodiment and do not limit the order of execution of the steps. The method provided in this embodiment can be executed by a related terminal device, and the following description is taken as an example of a server as the execution subject.

[0058] like Figure 1 As shown, the first embodiment of the present invention provides a test route generation method based on a semantic map, comprising steps S11 to S15:

[0059] S11, loading the semantic map according to the test area to obtain a test map;

[0060] S12. For each intersection on the test map, pair the entry navigation point of the intersection with the exit navigation point of the reachable intersection one by one to generate a plurality of navigation point pairs for each intersection;

[0061] S13, traversing a plurality of navigation point pairs at each intersection, matching each navigation point pair at the current intersection with each navigation point pair at the next intersection that can be reached, and obtaining a plurality of first matching pairs;

[0062] S14, taking each navigation point pair at each intersection as a node, and adding a connecting edge between two navigation point pairs in each first matching pair to construct a directed graph;

[0063] S15. Connect all the connection edges according to the directed graph to connect all the nodes in series and generate a test route.

[0064] It should be noted that the semantic map is a marked semantic map, which is marked with travel direction, intersection labels, intersection coordinates, scale information, speed limit information, etc.

[0065] As an example, in step S11, the user selects a test area on the semantic map through the user terminal according to the actual test requirements, and initiates a test route generation request to the server. The server responds to the test route generation request initiated by the user terminal, loads the semantic map corresponding to the test area according to the marked semantic map, and obtains the test map.

[0066] Among them, according to actual testing needs, the user can directly select a test area on the semantic map through the user terminal, or specify multiple test points on the semantic map through the user terminal, and the server will determine the test area based on the specified multiple test points, such as determining a polygonal area formed by connecting multiple test points as the test area, or determining the minimum circular area containing multiple test points as the test area.

[0067] In step S12, after the test map is obtained, the intersections located in the test area are screened out according to the coordinates of each intersection on the semantic map, and all the intersections on the test map are obtained. For each intersection on the test map, all the entry navigation points and all the exit navigation points of the same intersection are first found, and then the arrival relationship between all the entry navigation points and all the exit navigation points of the intersection is determined according to the travel direction on the semantic map, and finally the entry navigation points of the intersection are matched one by one with the exit navigation points of the intersection that can be reached by them, and several navigation point pairs of the intersection are generated, thereby generating several navigation point pairs of each intersection.

[0068] For example, Figure 2 As shown, for intersection A on the test map, find all entry navigation points A of intersection A 01 , A 02 , A 03 , and all exits from intersection A 11 , A 12 , A 13 , according to the direction of travel on the semantic map, determine all entry navigation points A 01 , A 02 , A 03 and all departures from navigation point A 11 , A 12 , A 13 The arrival relationship between them is due to A 01 Availability 11 , A 12 , A 13 , A 02 Availability 11 , A 12 , A 13 , A 03 Availability 11 , A 12 , A 13 , so A 01 Respectively with A 11 , A 12 , A 13 Pair and get the navigation point pair (A 01 , A 11 )、(A 01 , A 12 )、(A 01, A 13 ), A 02 Respectively with A 11 , A 12 , A 13 Pair and get the navigation point pair (A 02 , A 11 )、(A 02 , A 12 )、(A 02 , A 13 ), A 03 Respectively with A 11 , A 12 , A 13 Pair and get the navigation point pair (A 03 , A 11 )、(A 03 , A 12 )、(A 03 , A 13 ), thereby generating several navigation point pairs of intersection A. When all the intersections on the test map are traversed, several navigation point pairs of each intersection are generated.

[0069] In step S13, several navigation point pairs at various intersections are traversed. For a navigation point pair of the current intersection being traversed, a departure navigation point is first extracted from the navigation point pair, and then the arrival relationship between the departure navigation point and all the entry navigation points of the next intersection that can be reached by the departure navigation point is determined according to the travel direction on the semantic map. Finally, the navigation point pair to which all the entry navigation points of the next intersection that can be reached by the departure navigation point belong is found, and the navigation point pair is matched with each navigation point pair of the next intersection that can be reached by the departure navigation point, to obtain multiple first matching pairs.

[0070] For example, Figure 3 As shown, when traversing to the navigation point pair (A 01 , A 11 ), from the navigation point to (A 01 , A 11 ) to extract the departure navigation point A 11 , according to the direction of travel on the semantic map, determine the departure point A 11 All entry navigation points B with the next intersection B that it can reach 01 , B 02 , B 03 The arrival relationship between them is due to A 11 Can reach B 01 Unreachable B 02 and B 03 , so find B 01 The navigation point pair (B 01 , B 11 )、(B 01, B 12 )、(B 01 , B 13 ), and the navigation point is aligned to (A 01 , A 11 ) are respectively related to the navigation point pair (B 01 , B 11 )、(B 01 , B 12 )、(B 01 , B 13 ) matches, and obtains three first matching pairs [(A 01 , A 11 ), (B 01 , B 11 )]、[(A 01 , A 11 ), (B 01 , B 12 )]、[(A 01 , A 11 ), (B 01 , B 13 )]. When all navigation point pairs at each intersection on the test map are traversed, multiple first matching pairs are obtained.

[0071] In step S14, each navigation point pair at each intersection is taken as a node, and connecting edges are added between two navigation point pairs in each first matching pair to construct a directed graph.

[0072] The length of the connecting edge between the two navigation point pairs in the first matching pair is the distance between the two navigation point pairs. The server can calculate the distance between the two navigation point pairs according to the scale information on the test map and the length of the travel path between the two navigation point pairs on the test map.

[0073] In step S15, all nodes are connected in series by connecting all connecting edges according to the directed graph to generate a test route, thereby automatically generating a test route covering left turn, right turn, straight ahead, and U-turn directions at each intersection.

[0074] This embodiment uses a marked semantic map to traverse each intersection in the test area on the semantic map, automatically generates a test route covering the left turn, right turn, straight go, and U-turn directions of each intersection. It can fully cover the intersections in the semantic map test area to automatically generate a test route, which is beneficial to improving the testing efficiency of the semantic map.

[0075] In a preferred embodiment, the test route generation method based on the semantic map also includes: when the navigation point pair of the current intersection reaches the navigation point pair of the next intersection through the boundary of the test area, the navigation point pair of the current intersection is matched with each navigation point pair of the target intersection that it can reach, to obtain multiple second matching pairs; wherein the distance between the current intersection and the target intersection is less than the preset out-of-domain distance; then each navigation point pair of each intersection is taken as a node, and connecting edges are added between the two navigation point pairs in each first matching pair, and connecting edges are added between the two navigation point pairs in each second matching pair, to construct a directed graph.

[0076] As an example, in step S13, if a navigation point pair of the current intersection traversed is a navigation point pair that passes through the boundary of the test area to reach the next intersection, then the departure navigation point is first extracted from the navigation point pair, and then the arrival relationship between the departure navigation point and all the entry navigation points of the target intersection that it can reach is determined based on the travel direction on the semantic map and the preset out-of-domain distance. Finally, the navigation point pair to which all the entry navigation points of the target intersection that can be reached by the departure navigation point belongs is found, and the navigation point pair is matched with each navigation point pair of the target intersection that can be reached by the departure navigation point to obtain multiple second matching pairs.

[0077] For example, Figure 4 As shown, when traversing to the navigation point pair (A 01 , A 12 ), due to the navigation point (A 01 , A 12 ) is the navigation point pair (B 03 , B 13 ), so you need to confirm the navigation point again (A 01 , A 12 ) After passing the test area boundary, whether there is a return to the test area, and confirm the navigation point (A 01 , A 12 ) and the length of the path between each intersection in the test area exceeds the preset out-of-domain distance, from the navigation point to (A 01 , A 12 ) to extract the departure navigation point A 12 , according to the direction of travel on the semantic map and the preset distance outside the domain, determine the departure point A 12 All entry navigation points B with its reachable target intersection B 01 , B 02 , B 03 The arrival relationship between them is due to A 12 Can reach B 03 Unreachable B 01 and B 02 , so find B03 The navigation point pair (B 03 , B 11 )、(B 03 , B 12 )、(B 03 , B 13 ), and the navigation point is aligned to (A 01 , A 12 ) are respectively related to the navigation point pair (B 03 , B 11 )、(B 03 , B 12 )、(B 03 , B 13 ) matches, and obtains three first matching pairs [(A 01 , A 12 ), (B 03 , B 11 )]、[(A 01 , A 12 ), (B 03 , B 12 )]、[(A 01 , A 12 ), (B 03 , B 13 )]. When all navigation point pairs at each intersection on the test map are traversed, multiple second matching pairs are obtained.

[0078] Then in step S14, each navigation point pair at each intersection is taken as a node, and a connecting edge is added between the two navigation point pairs in each first matching pair, and a connecting edge is added between the two navigation point pairs in each second matching pair, so as to construct a directed graph.

[0079] The length of the connecting edge between the two navigation point pairs in the second matching pair is the distance between the two navigation point pairs. The server can calculate the distance between the two navigation point pairs according to the scale information on the test map and the length of the travel path between the two navigation point pairs on the test map.

[0080] This embodiment automatically generates a test route by adding target intersections reachable from the current intersection when the navigation point pair at the current intersection reaches the navigation point pair at the next intersection through the boundary of the test area. This can automatically generate a test route that fully covers the intersections in the test area on the semantic map, thereby improving the testing efficiency of the semantic map.

[0081] In a preferred embodiment, for each intersection on the test map, the entry navigation point of the intersection is paired one-to-one with the exit navigation point of the intersection that it can reach, generating several navigation point pairs for each intersection. Specifically, for each intersection on the test map, the entry navigation point of the intersection is paired one-to-one with the exit navigation point of the intersection that it can reach, generating several initial navigation point pairs, and selecting an initial navigation point pair from all the initial navigation point pairs on each edge of the intersection as a navigation point pair, to obtain several navigation point pairs.

[0082] As an example, after obtaining the test map, the intersections located in the test area are screened out according to the coordinates of each intersection on the semantic map to obtain all the intersections on the test map. For each intersection on the test map, first find all the entry navigation points and all the exit navigation points of the same intersection, then determine the arrival relationship between all the entry navigation points and all the exit navigation points of the intersection according to the travel direction on the semantic map, then pair the entry navigation points of the intersection with the exit navigation points of the intersection that can be reached by it one by one, generate several initial navigation point pairs of the intersection, and finally arbitrarily select an initial navigation point pair from all the initial navigation point pairs on each edge of the intersection as a navigation point pair, obtain several navigation point pairs, and thus generate several navigation point pairs of each intersection.

[0083] For example, Figure 2 , Figure 5 As shown, for intersection A on the test map, find all entry navigation points A of intersection A 01 , A 02 , A 03 , and all exits from intersection A 11 , A 12 , A 13 , according to the direction of travel on the semantic map, determine all entry navigation points A 01 , A 02 , A 03 and all departures from navigation point A 11 , A 12 , A 13 The arrival relationship between them is due to A 01 Availability 11 , A 12 , A 13 , A 02 Availability 11 , A 12 , A 13 , A 03 Availability 11 , A 12 , A 13 , so A 01 Respectively with A 11 , A12 , A 13 Pair and get the initial navigation point pair (A 01 , A 11 )、(A 01 , A 12 )、(A 01 , A 13 ), A 02 Respectively with A 11 , A 12 , A 13 Pair and get the initial navigation point pair (A 02 , A 11 )、(A 02 , A 12 )、(A 02 , A 13 ), A 03 Respectively with A 11 , A 12 , A 13 Pair and get the initial navigation point pair (A 03 , A 11 )、(A 03 , A 12 )、(A 03 , A 13 ), thereby generating several initial navigation point pairs of intersection A. Since there are only initial navigation point pairs (A 01 , A 11 ), there is an initial navigation point pair (A 01 , A 12 )、(A 01 , A 13 ), there is an initial navigation point pair (A 02 , A 11 )、(A 03 , A 11 ), there is an initial navigation point pair (A 02 , A 12 )、(A 02 , A 13 )、(A 03 , A 12 )、(A 03 , A 13 ), select any initial navigation point pair from all the initial navigation point pairs on each side of intersection A as the navigation point pair, for example, select the initial navigation point pair (A 01 , A 11 ) as the navigation point pair, select the initial navigation point pair (A 01 , A 12) as the navigation point pair, select the initial navigation point pair (A 02 , A 11 ) as the navigation point pair, select the initial navigation point pair (A) on the “—” side of intersection A. 02 , A 12 ) as navigation point pairs, and obtain several navigation point pairs, thereby generating several navigation point pairs of each intersection. When all the intersections on the test map are traversed, several navigation point pairs of each intersection are generated.

[0084] This embodiment deduplicates several initial navigation point pairs at each intersection to obtain several navigation point pairs, which can reduce the amount of data for subsequent traversal pairing, reduce the processing pressure of the server, and help improve the testing efficiency of the semantic map.

[0085] In a preferred embodiment, the method of connecting all the connecting edges according to a directed graph to connect all the nodes in series to generate a test route is specifically as follows: based on the directed graph, traverse a number of navigation point pairs at each intersection, determine whether the navigation point pair at the current intersection does not reach the navigation point pair at another intersection after passing the boundary of the test area, and if so, eliminate the navigation point pair at the current intersection to obtain an intermediate graph; based on the intermediate graph, traverse a number of navigation point pairs at each intersection, match each navigation point pair at the current intersection with each navigation point pair at another intersection it reaches, obtain multiple third matching pairs, and add connecting edges between two navigation point pairs in each third matching pair to generate a complete graph; connect all the connecting edges according to the complete graph to connect all the nodes in series to generate a test route.

[0086] As an example, the navigation point pairs in the directed graph that cannot return to the test area after crossing the boundary of the test area are eliminated to obtain an intermediate graph. By eliminating the navigation point pairs that cannot return, the interference of useless navigation point pairs can be eliminated, which is conducive to generating accurate test routes. Considering that there may be only some connecting edges between the navigation point pairs in the intermediate graph, and the other navigation point pairs are disconnected, it is necessary to complete the connecting edges of these navigation point pairs to generate a complete graph, so as to connect all the connecting edges according to the complete graph to connect all the nodes in series and generate a test route.

[0087] For example, in a directed graph G, the classic algorithm Tarjan algorithm is used to find strongly connected components (a graph in which every pair of nodes can reach each other is called a strongly connected graph. The maximal strongly connected subgraph of a directed graph is called a strongly connected component). Only the nodes in the strongly connected component with the largest number of nodes are left to eliminate some points that will reach the map boundary and cannot return to the map, and obtain the intermediate graph. For each node in the intermediate graph, the classic algorithm Dijkstra is used to find the single-source shortest path, obtain the distance between each pair of nodes, and fill the intermediate graph into a complete graph. In order to cover the left turn, right turn, straight, and U-turn directions of each intersection, a test route is automatically generated. The test route is required to cover all nodes in the complete graph, so it is converted to a traveling salesman problem (TSP). Any intelligent algorithm (such as genetic algorithm, ant colony algorithm, etc.) can be used to find a shorter circular route (also called Hamiltonian circuit) that connects all nodes in series.

[0088] This embodiment can further comprehensively cover the intersections in the semantic map test area by first eliminating the navigation point pairs that cannot be returned and then completing the connecting edges between some disconnected navigation point pairs, automatically generating accurate test routes, which is beneficial to improving the testing efficiency of the semantic map.

[0089] In a preferred embodiment, after connecting all the connecting edges according to the directed graph to connect all the nodes in series to generate a test route, it also includes: according to the speed limit information of the test map and the length of the test route, splitting the test route into a number of test sub-routes with a test duration less than a preset test duration to obtain a test route set.

[0090] As an example, considering that the generated test route may be relatively long, the vehicle cannot complete the test by running the entire test route in one go, and the driver also needs time to rest, after the test route is generated, the test route is split into several test sub-routes with a test duration less than a preset test duration according to the speed limit information on the test map and the length of the test route, to obtain a test route set.

[0091] This embodiment can ensure the safety of subsequent semantic map testing by splitting the test route into several test sub-routes.

[0092] In a preferred embodiment, after the semantic map is loaded according to the test area to obtain the test map, and before the entry navigation point of the intersection and the exit navigation point of the reachable intersection are matched one-to-one for each intersection on the test map to generate a number of navigation point pairs for each intersection, it also includes: traversing each intersection in the semantic map and collecting the intersections in the test area into the intersection set of the test map.

[0093] As an example, based on the intersection labels and intersection coordinates on the semantic map, it is determined whether each intersection on the semantic map is located in the test area. If so, the intersection is considered to be an intersection within the test area and is collected in the intersection set of the test map.

[0094] Based on the same inventive concept as the first embodiment, the second embodiment of the present invention provides Figure 6 A test route generation device based on a semantic map is shown, comprising: a test map acquisition module 21, used to load a semantic map according to a test area to obtain a test map; a navigation point pair generation module 22, used to match the entry navigation point of the intersection with the exit navigation point of the reachable intersection for each intersection on the test map, and generate a number of navigation point pairs for each intersection; a navigation point pair matching module 23, used to traverse a number of navigation point pairs at each intersection, match each navigation point pair of the current intersection with each navigation point pair of the next reachable intersection, and obtain a plurality of first matching pairs; a directed graph construction module 24, used to take each navigation point pair of each intersection as a node, and add connecting edges between two navigation point pairs in each first matching pair to construct a directed graph; a test route generation module 25, used to connect all the connecting edges according to the directed graph to connect all the nodes in series, and generate a test route.

[0095] In a preferred embodiment, the navigation point pair matching module 23 is also used to match the navigation point pair of the current intersection with each navigation point pair of the reachable target intersection when the navigation point pair of the current intersection reaches the navigation point pair of the next intersection through the boundary of the test area, so as to obtain multiple second matching pairs; wherein the distance between the current intersection and the target intersection is less than the preset out-of-domain distance; and the directed graph construction module 24 is used to use each navigation point pair of each intersection as a node, and add connecting edges between the two navigation point pairs in each first matching pair, and add connecting edges between the two navigation point pairs in each second matching pair, so as to construct a directed graph.

[0096] In a preferred embodiment, for each intersection on the test map, the entry navigation point of the intersection is paired one-to-one with the exit navigation point of the intersection that it can reach, generating several navigation point pairs for each intersection. Specifically, for each intersection on the test map, the entry navigation point of the intersection is paired one-to-one with the exit navigation point of the intersection that it can reach, generating several initial navigation point pairs, and selecting an initial navigation point pair from all the initial navigation point pairs on each edge of the intersection as a navigation point pair, to obtain several navigation point pairs.

[0097] In a preferred embodiment, the method of connecting all the connecting edges according to a directed graph to connect all the nodes in series to generate a test route is specifically as follows: based on the directed graph, traverse a number of navigation point pairs at each intersection, determine whether the navigation point pair at the current intersection does not reach the navigation point pair at another intersection after passing the boundary of the test area, and if so, eliminate the navigation point pair at the current intersection to obtain an intermediate graph; based on the intermediate graph, traverse a number of navigation point pairs at each intersection, match each navigation point pair at the current intersection with each navigation point pair at another intersection it reaches, obtain multiple third matching pairs, and add connecting edges between two navigation point pairs in each third matching pair to generate a complete graph; connect all the connecting edges according to the complete graph to connect all the nodes in series to generate a test route.

[0098] In a preferred embodiment, after connecting all the connecting edges according to the directed graph to connect all the nodes in series to generate a test route, it also includes: according to the speed limit information of the test map and the length of the test route, splitting the test route into a number of test sub-routes with a test duration less than a preset test duration to obtain a test route set.

[0099] In a preferred embodiment, after the semantic map is loaded according to the test area to obtain the test map, and before the entry navigation point of the intersection and the exit navigation point of the reachable intersection are matched one-to-one for each intersection on the test map to generate a number of navigation point pairs for each intersection, it also includes: traversing each intersection in the semantic map and collecting the intersections in the test area into the intersection set of the test map.

[0100] like Figure 7 As shown, the third embodiment of the present invention provides a testing method based on a semantic map, comprising steps S31 to S36:

[0101] S31, loading the semantic map according to the test area to obtain a test map;

[0102] S32, for each intersection on the test map, pairing the entry navigation point of the intersection with the exit navigation point of the reachable intersection one by one, and generating a plurality of navigation point pairs for each intersection;

[0103] S33, traversing a plurality of navigation point pairs at each intersection, matching each navigation point pair at the current intersection with each navigation point pair at the next intersection that can be reached, to obtain a plurality of first matching pairs;

[0104] S34, taking each navigation point pair at each intersection as a node, and adding a connecting edge between two navigation point pairs in each first matching pair to construct a directed graph;

[0105] S35, connecting all the connection edges according to the directed graph to connect all the nodes in series, generating a test route;

[0106] S36. Send a test route to the vehicle, so that the vehicle performs an autonomous driving test according to the test route, and correct the test map according to the acquired test data.

[0107] It should be noted that the semantic map is a marked semantic map, which is marked with travel direction, intersection labels, intersection coordinates, scale information, speed limit information, etc.

[0108] As an example, in step S31, the user selects a test area on the semantic map through the user terminal according to the actual test requirements, and initiates a test route generation request to the server. The server responds to the test route generation request initiated by the user terminal, loads the semantic map corresponding to the test area according to the marked semantic map, and obtains the test map.

[0109] Among them, according to actual testing needs, the user can directly select a test area on the semantic map through the user terminal, or specify multiple test points on the semantic map through the user terminal, and the server will determine the test area based on the specified multiple test points, such as determining a polygonal area formed by connecting multiple test points as the test area, or determining the minimum circular area containing multiple test points as the test area.

[0110] In step S32, after the test map is obtained, the intersections located in the test area are screened out according to the coordinates of each intersection on the semantic map, and all the intersections on the test map are obtained. For each intersection on the test map, all the entry navigation points and all the exit navigation points of the same intersection are first found, and then the arrival relationship between all the entry navigation points and all the exit navigation points of the intersection is determined according to the travel direction on the semantic map, and finally the entry navigation points of the intersection are matched one by one with the exit navigation points of the intersection that can be reached by the intersection, and several navigation point pairs of the intersection are generated, thereby generating several navigation point pairs of each intersection.

[0111] In step S33, several navigation point pairs at various intersections are traversed. For a navigation point pair of the current intersection being traversed, a departure navigation point is first extracted from the navigation point pair. Then, based on the travel direction on the semantic map, the arrival relationship between the departure navigation point and all the entry navigation points of the next intersection that can be reached by the departure navigation point is determined. Finally, the navigation point pair to which all the entry navigation points of the next intersection that can be reached by the departure navigation point belong is found, and the navigation point pair is matched with each navigation point pair of the next intersection that can be reached by the departure navigation point, to obtain multiple first matching pairs.

[0112] In step S34, each navigation point pair at each intersection is taken as a node, and connecting edges are added between two navigation point pairs in each first matching pair to construct a directed graph.

[0113] The length of the connecting edge between the two navigation point pairs in the first matching pair is the distance between the two navigation point pairs. The server can calculate the distance between the two navigation point pairs according to the scale information on the test map and the length of the travel path between the two navigation point pairs on the test map.

[0114] In step S35, all nodes are connected in series by connecting all connection edges according to the directed graph to generate a test route, thereby automatically generating a test route covering left turn, right turn, straight ahead, and U-turn directions at each intersection.

[0115] In step S36, the user can mark the generated test route as a coverage test, name it accordingly, and upload it to the server. When the vehicle is testing, it will automatically capture the test route from the server, and the server will send the captured test route to the vehicle. The safety officer can select a test route for coverage testing so that the vehicle can automatically traverse the test route. If the vehicle interrupts the test, the vehicle will record the current location, and the next test will start from the last test location. If the safety officer finds that there is a problem with the vehicle during the autonomous driving process, the safety officer will report the problem to the vehicle through the user terminal. The vehicle will automatically collect test data and upload the test data to the server so that engineers can view the uploaded test data and correct any problematic areas. After correcting the semantic map, the vehicle will re-perform the autonomous driving test until all problems in the test area are resolved.

[0116] This embodiment uses a marked semantic map to traverse each intersection in the test area on the semantic map, automatically generates a test route covering the left turn, right turn, straight go, and U-turn directions of each intersection, allows the vehicle to perform autonomous driving tests according to the test route, and corrects the test map based on the acquired test data. It can fully cover the intersections in the semantic map test area and automatically generate a test route, which is beneficial to improving the testing efficiency of the semantic map.

[0117] In a preferred embodiment, the test route generation method based on the semantic map also includes: when the navigation point pair of the current intersection reaches the navigation point pair of the next intersection through the boundary of the test area, the navigation point pair of the current intersection is matched with each navigation point pair of the target intersection that it can reach, to obtain multiple second matching pairs; wherein the distance between the current intersection and the target intersection is less than the preset out-of-domain distance; then each navigation point pair of each intersection is taken as a node, and connecting edges are added between the two navigation point pairs in each first matching pair, and connecting edges are added between the two navigation point pairs in each second matching pair, to construct a directed graph.

[0118] As an example, in step S33, if a navigation point pair of the current intersection traversed is a navigation point pair that passes through the boundary of the test area to reach the next intersection, then the departure navigation point is first extracted from the navigation point pair, and then the arrival relationship between the departure navigation point and all the entry navigation points of the target intersection that it can reach is determined based on the travel direction on the semantic map and the preset out-of-domain distance. Finally, the navigation point pair to which all the entry navigation points of the target intersection that can be reached by the departure navigation point belongs is found, and the navigation point pair is matched with each navigation point pair of the target intersection that can be reached by the departure navigation point, to obtain multiple second matching pairs.

[0119] Then in step S34, each navigation point pair at each intersection is taken as a node, and a connecting edge is added between the two navigation point pairs in each first matching pair, and a connecting edge is added between the two navigation point pairs in each second matching pair, so as to construct a directed graph.

[0120] The length of the connecting edge between the two navigation point pairs in the second matching pair is the distance between the two navigation point pairs. The server can calculate the distance between the two navigation point pairs according to the scale information on the test map and the length of the travel path between the two navigation point pairs on the test map.

[0121] This embodiment automatically generates a test route by adding target intersections reachable from the current intersection when the navigation point pair at the current intersection reaches the navigation point pair at the next intersection through the boundary of the test area. This can automatically generate a test route that fully covers the intersections in the test area on the semantic map, thereby improving the testing efficiency of the semantic map.

[0122] In a preferred embodiment, for each intersection on the test map, the entry navigation point of the intersection is paired one-to-one with the exit navigation point of the intersection that it can reach, generating several navigation point pairs for each intersection. Specifically, for each intersection on the test map, the entry navigation point of the intersection is paired one-to-one with the exit navigation point of the intersection that it can reach, generating several initial navigation point pairs, and selecting an initial navigation point pair from all the initial navigation point pairs on each edge of the intersection as a navigation point pair, to obtain several navigation point pairs.

[0123] As an example, after obtaining the test map, the intersections located in the test area are screened out according to the coordinates of each intersection on the semantic map to obtain all the intersections on the test map. For each intersection on the test map, first find all the entry navigation points and all the exit navigation points of the same intersection, then determine the arrival relationship between all the entry navigation points and all the exit navigation points of the intersection according to the travel direction on the semantic map, then pair the entry navigation points of the intersection with the exit navigation points of the intersection that can be reached by it one by one, generate several initial navigation point pairs of the intersection, and finally arbitrarily select an initial navigation point pair from all the initial navigation point pairs on each edge of the intersection as a navigation point pair, obtain several navigation point pairs, and thus generate several navigation point pairs of each intersection.

[0124] This embodiment deduplicates several initial navigation point pairs at each intersection to obtain several navigation point pairs, which can reduce the amount of data for subsequent traversal pairing, reduce the processing pressure of the server, and help improve the testing efficiency of the semantic map.

[0125] In a preferred embodiment, the method of connecting all the connecting edges according to a directed graph to connect all the nodes in series to generate a test route is specifically as follows: based on the directed graph, traverse a number of navigation point pairs at each intersection, determine whether the navigation point pair at the current intersection does not reach the navigation point pair at another intersection after passing the boundary of the test area, and if so, eliminate the navigation point pair at the current intersection to obtain an intermediate graph; based on the intermediate graph, traverse a number of navigation point pairs at each intersection, match each navigation point pair at the current intersection with each navigation point pair at another intersection it reaches, obtain multiple third matching pairs, and add connecting edges between two navigation point pairs in each third matching pair to generate a complete graph; connect all the connecting edges according to the complete graph to connect all the nodes in series to generate a test route.

[0126] As an example, the navigation point pairs in the directed graph that cannot return to the test area after crossing the boundary of the test area are eliminated to obtain an intermediate graph. By eliminating the navigation point pairs that cannot return, the interference of useless navigation point pairs can be eliminated, which is conducive to generating accurate test routes. Considering that there may be only some connecting edges between the navigation point pairs in the intermediate graph, and the other navigation point pairs are disconnected, it is necessary to complete the connecting edges of these navigation point pairs to generate a complete graph, so as to connect all the connecting edges according to the complete graph to connect all the nodes in series and generate a test route.

[0127] For example, in a directed graph G, the classic algorithm Tarjan algorithm is used to find strongly connected components (a graph in which every pair of nodes can reach each other is called a strongly connected graph. The maximal strongly connected subgraph of a directed graph is called a strongly connected component). Only the nodes in the strongly connected component with the largest number of nodes are left to eliminate some points that will reach the map boundary and cannot return to the map, and obtain the intermediate graph. For each node in the intermediate graph, the classic algorithm Dijkstra is used to find the single-source shortest path, obtain the distance between each pair of nodes, and fill the intermediate graph into a complete graph. In order to cover the left turn, right turn, straight, and U-turn directions of each intersection, a test route is automatically generated. The test route is required to cover all nodes in the complete graph, so it is converted to a traveling salesman problem (TSP). Any intelligent algorithm (such as genetic algorithm, ant colony algorithm, etc.) can be used to find a shorter circular route (also called Hamiltonian circuit) that connects all nodes in series.

[0128] This embodiment can further comprehensively cover the intersections in the semantic map test area by first eliminating the navigation point pairs that cannot be returned and then completing the connecting edges between some disconnected navigation point pairs, automatically generating accurate test routes, which is beneficial to improving the testing efficiency of the semantic map.

[0129] In a preferred embodiment, after connecting all the connecting edges according to the directed graph to connect all the nodes in series to generate a test route, it also includes: according to the speed limit information of the test map and the length of the test route, splitting the test route into a number of test sub-routes with a test duration less than a preset test duration to obtain a test route set.

[0130] As an example, considering that the generated test route may be relatively long, the vehicle cannot complete the test by running the entire test route in one go, and the driver also needs time to rest, after the test route is generated, the test route is split into several test sub-routes with a test duration less than a preset test duration according to the speed limit information on the test map and the length of the test route, to obtain a test route set.

[0131] This embodiment can ensure the safety of subsequent semantic map testing by splitting the test route into several test sub-routes.

[0132] In a preferred embodiment, after the semantic map is loaded according to the test area to obtain the test map, and before the entry navigation point of the intersection and the exit navigation point of the reachable intersection are matched one-to-one for each intersection on the test map to generate a number of navigation point pairs for each intersection, it also includes: traversing each intersection in the semantic map and collecting the intersections in the test area into the intersection set of the test map.

[0133] As an example, based on the intersection labels and intersection coordinates on the semantic map, it is determined whether each intersection on the semantic map is located in the test area. If so, the intersection is considered to be an intersection within the test area and is collected in the intersection set of the test map.

[0134] Based on the same inventive concept as the third embodiment, the fourth embodiment of the present invention provides Figure 8 A test device based on a semantic map is shown, comprising: a test map acquisition module 41, used to load a semantic map according to a test area to obtain a test map; a navigation point pair generation module 42, used to match the entry navigation point of the intersection with the exit navigation point of the reachable intersection for each intersection on the test map, and generate a number of navigation point pairs for each intersection; a navigation point pair matching module 43, used to traverse a number of navigation point pairs at each intersection, match each navigation point pair of the current intersection with each navigation point pair of the next reachable intersection, and obtain a plurality of first matching pairs; a directed graph construction module 44, used to take each navigation point pair of each intersection as a node, and add connecting edges between two navigation point pairs in each first matching pair to construct a directed graph; a test route generation module 45, used to connect all connecting edges according to the directed graph to connect all nodes in series to generate a test route; an automatic driving test module 46, used to send the test route to the vehicle, so that the vehicle performs an automatic driving test according to the test route, and correct the test map according to the acquired test data.

[0135] In a preferred embodiment, the navigation point pair matching module 43 is also used to match the navigation point pair of the current intersection with each navigation point pair of the reachable target intersection when the navigation point pair of the current intersection reaches the navigation point pair of the next intersection through the boundary of the test area, so as to obtain multiple second matching pairs; wherein the distance between the current intersection and the target intersection is less than the preset out-of-domain distance; and the directed graph construction module 44 is used to use each navigation point pair of each intersection as a node, and add connecting edges between the two navigation point pairs in each first matching pair, and add connecting edges between the two navigation point pairs in each second matching pair, so as to construct a directed graph.

[0136] In a preferred embodiment, for each intersection on the test map, the entry navigation point of the intersection is paired one-to-one with the exit navigation point of the intersection that it can reach, generating several navigation point pairs for each intersection. Specifically, for each intersection on the test map, the entry navigation point of the intersection is paired one-to-one with the exit navigation point of the intersection that it can reach, generating several initial navigation point pairs, and selecting an initial navigation point pair from all the initial navigation point pairs on each edge of the intersection as a navigation point pair, to obtain several navigation point pairs.

[0137] In a preferred embodiment, the method of connecting all the connecting edges according to a directed graph to connect all the nodes in series to generate a test route is specifically as follows: based on the directed graph, traverse a number of navigation point pairs at each intersection, determine whether the navigation point pair at the current intersection does not reach the navigation point pair at another intersection after passing the boundary of the test area, and if so, eliminate the navigation point pair at the current intersection to obtain an intermediate graph; based on the intermediate graph, traverse a number of navigation point pairs at each intersection, match each navigation point pair at the current intersection with each navigation point pair at another intersection it reaches, obtain multiple third matching pairs, and add connecting edges between two navigation point pairs in each third matching pair to generate a complete graph; connect all the connecting edges according to the complete graph to connect all the nodes in series to generate a test route.

[0138] In a preferred embodiment, after connecting all the connecting edges according to the directed graph to connect all the nodes in series to generate a test route, it also includes: according to the speed limit information of the test map and the length of the test route, splitting the test route into a number of test sub-routes with a test duration less than a preset test duration to obtain a test route set.

[0139] In a preferred embodiment, after the semantic map is loaded according to the test area to obtain the test map, and before the entry navigation point of the intersection and the exit navigation point of the reachable intersection are matched one-to-one for each intersection on the test map to generate a number of navigation point pairs for each intersection, it also includes: traversing each intersection in the semantic map and collecting the intersections in the test area into the intersection set of the test map.

[0140] In summary, the implementation of the embodiments of the present invention has the following beneficial effects:

[0141] By loading the semantic map according to the test area, the test map is obtained. For each intersection on the test map, the entry navigation point of the intersection is matched with the exit navigation point of the intersection that can be reached, and a number of navigation point pairs of each intersection are generated. The navigation point pairs of each intersection are traversed, and each navigation point pair of the current intersection is matched with each navigation point pair of the next intersection that can be reached, and multiple first matching pairs are obtained. Each navigation point pair of each intersection is used as a node, and connecting edges are added between two navigation point pairs in each first matching pair to construct a directed graph. All nodes are connected in series by connecting all connecting edges according to the directed graph to generate a test route, and the test route is automatically generated. The embodiment of the present invention uses a marked semantic map to traverse each intersection in the test area on the semantic map, and automatically generates a test route covering the left turn, right turn, straight, and U-turn directions of each intersection. It can fully cover the intersections in the semantic map test area to automatically generate a test route, which is conducive to improving the test efficiency of the semantic map.

[0142] The above is a preferred embodiment of the present invention. It should be pointed out that a person skilled in the art can make several improvements and modifications without departing from the principle of the present invention. These improvements and modifications are also considered to be within the scope of protection of the present invention.

[0143] Those skilled in the art can understand that all or part of the processes in the above embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes in the above embodiments. The storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc.

Claims

1. A test route generation method based on semantic map, characterized in that: include: Load the semantic map according to the test area to obtain the test map; For each intersection on the test map, the entry navigation point of the intersection is matched with the exit navigation point of the accessible intersection one by one to generate several navigation point pairs for each intersection; Traversing a plurality of navigation point pairs at each intersection, matching each navigation point pair at the current intersection with each navigation point pair at the next intersection that can be reached, and obtaining a plurality of first matching pairs; Each navigation point pair at each intersection is taken as a node, and a connecting edge is added between two navigation point pairs in each first matching pair to construct a directed graph; When the navigation point pair of the current intersection reaches the navigation point pair of the next intersection through the boundary of the test area, the navigation point pair of the current intersection is matched with each navigation point pair of the target intersection that it can reach, and multiple second matching pairs are obtained; wherein the distance between the current intersection and the target intersection is less than the preset out-of-domain distance; then each navigation point pair of each intersection is taken as a node, and a connecting edge is added between the two navigation point pairs in each first matching pair, and a connecting edge is added between the two navigation point pairs in each second matching pair, to construct a directed graph; Connect all the connected edges according to the directed graph to connect all the nodes in series and generate a test route.

2. The method for generating a test route based on a semantic map according to claim 1, characterized in that: For each intersection on the test map, the entry navigation point of the intersection is matched with the exit navigation point of the reachable intersection one by one to generate several navigation point pairs for each intersection, specifically: For each intersection on the test map, the entry navigation point of the intersection is paired with the exit navigation point of the reachable intersection one by one to generate several initial navigation point pairs, and one initial navigation point pair is selected from all the initial navigation point pairs on each edge of the intersection as the navigation point pair to obtain several navigation point pairs.

3. The test route generation method based on semantic map according to claim 1, characterized in that: The method of connecting all the connection edges according to the directed graph to connect all the nodes in series and generate a test route is specifically as follows: Based on the directed graph, several navigation point pairs of each intersection are traversed to determine whether the navigation point pair of the current intersection does not reach the navigation point pair of another intersection after passing through the boundary of the test area. If so, the navigation point pair of the current intersection is removed to obtain an intermediate graph; Based on the intermediate graph, traverse a number of navigation point pairs at each intersection, match each navigation point pair at the current intersection with each navigation point pair at another intersection it reaches, obtain multiple third matching pairs, and add connecting edges between two navigation point pairs in each third matching pair to generate a complete graph; According to the complete graph, all the connected edges are connected to connect all the nodes in series to generate a test route.

4. The method for generating a test route based on a semantic map according to claim 1, characterized in that: After connecting all the connection edges according to the directed graph to connect all the nodes in series to generate a test route, the method further includes: According to the speed limit information of the test map and the length of the test route, the test route is split into a number of test sub-routes whose test duration is less than a preset test duration, to obtain a test route set.

5. The method for generating a test route based on a semantic map according to claim 1, characterized in that: After the semantic map is loaded according to the test area to obtain the test map, before the entry navigation point of each intersection on the test map is matched with the exit navigation point of the reachable intersection one by one to generate a plurality of navigation point pairs for each intersection, the method further includes: Each intersection in the semantic map is traversed, and the intersections in the test area are collected into the intersection set of the test map.

6. A test route generation device based on semantic map, characterized in that: include: A test map acquisition module is used to load a semantic map according to a test area to obtain a test map; A navigation point pair generation module is used to match the entry navigation point of the intersection with the exit navigation point of the reachable intersection for each intersection on the test map, and generate a number of navigation point pairs for each intersection; A navigation point pair matching module, used to traverse a plurality of navigation point pairs at each intersection, match each navigation point pair at the current intersection with each navigation point pair at the next intersection that can be reached, and obtain a plurality of first matching pairs; A directed graph construction module, used to take each navigation point pair at each intersection as a node and add a connecting edge between two navigation point pairs in each first matching pair to construct a directed graph; When the navigation point pair of the current intersection reaches the navigation point pair of the next intersection through the boundary of the test area, the navigation point pair matching module matches the navigation point pair of the current intersection with each navigation point pair of the target intersection that it can reach, and obtains multiple second matching pairs; wherein the distance between the current intersection and the target intersection is less than the preset out-of-domain distance; then the directed graph construction module uses each navigation point pair of each intersection as a node, and adds a connecting edge between the two navigation point pairs in each first matching pair, and adds a connecting edge between the two navigation point pairs in each second matching pair, to construct a directed graph; The test route generation module is used to connect all the connection edges according to the directed graph to connect all the nodes in series and generate a test route.

7. A testing method based on semantic map, characterized in that: include: Load the semantic map according to the test area to obtain the test map; For each intersection on the test map, the entry navigation point of the intersection is matched with the exit navigation point of the accessible intersection one by one to generate several navigation point pairs for each intersection; Traversing a plurality of navigation point pairs at each intersection, matching each navigation point pair at the current intersection with each navigation point pair at the next intersection that can be reached, and obtaining a plurality of first matching pairs; Each navigation point pair at each intersection is taken as a node, and a connecting edge is added between two navigation point pairs in each first matching pair to construct a directed graph; When the navigation point pair of the current intersection reaches the navigation point pair of the next intersection through the boundary of the test area, the navigation point pair of the current intersection is matched with each navigation point pair of the target intersection that it can reach, and multiple second matching pairs are obtained; wherein the distance between the current intersection and the target intersection is less than the preset out-of-domain distance; each navigation point pair of each intersection is taken as a node, and a connecting edge is added between the two navigation point pairs in each first matching pair, and a connecting edge is added between the two navigation point pairs in each second matching pair, to construct a directed graph; all nodes are connected in series by connecting all the connecting edges according to the directed graph to generate a test route; A test route is sent to the vehicle so that the vehicle can perform autonomous driving tests according to the test route, and the test map is corrected based on the acquired test data.

8. A testing device based on semantic map, characterized in that: include: A test map acquisition module is used to load a semantic map according to a test area to obtain a test map; A navigation point pair generation module is used to match the entry navigation point of the intersection with the exit navigation point of the reachable intersection for each intersection on the test map, and generate a number of navigation point pairs for each intersection; A navigation point pair matching module, used to traverse a plurality of navigation point pairs at each intersection, match each navigation point pair at the current intersection with each navigation point pair at the next intersection that can be reached, and obtain a plurality of first matching pairs; A directed graph construction module, used to take each navigation point pair at each intersection as a node and add a connecting edge between two navigation point pairs in each first matching pair to construct a directed graph; When the navigation point pair of the current intersection reaches the navigation point pair of the next intersection through the boundary of the test area, the navigation point pair matching module matches the navigation point pair of the current intersection with each navigation point pair of the target intersection that it can reach, and obtains multiple second matching pairs; wherein the distance between the current intersection and the target intersection is less than the preset out-of-domain distance; then the directed graph construction module uses each navigation point pair of each intersection as a node, and adds a connecting edge between the two navigation point pairs in each first matching pair, and adds a connecting edge between the two navigation point pairs in each second matching pair, to construct a directed graph; A test route generation module is used to connect all the connection edges according to the directed graph to connect all the nodes in series and generate a test route; The autonomous driving test module is used to send a test route to the vehicle, enable the vehicle to perform autonomous driving tests according to the test route, and correct the test map based on the acquired test data.

Images