TWI807771B - Adaptive Navigation Method and Cloud Navigation Path and Map Publishing Platform Utilizing it - Google Patents

Abstract

An adaptive navigation method is implemented by a cloud navigation path and map publishing platform, which includes: Executing a path planning operation based on a starting point and an end point specified by a self-driving platform. The path planning operation includes calculating a global navigation path with the shortest distance, the shortest time, or the lowest energy consumption based on the road network and weights in the map information stored in a high-precision electronic map storage unit. The factors of the weight include lane width, number of lanes, road edge width, road slope and road regulations. The formed two-dimensional set, the one-dimensional set formed by the height information corresponding to the global navigation path, the one-dimensional set formed by the distance from the stop line ahead, and the one-dimensional set formed by the distance to the end point; read the SLAM feature map within the range of the global path from the high-precision electronic map storage unit, and perform a grid processing on the SLAM feature map to obtain a grid SLAM feature map; and transmit the sets of the global navigation path and the grid SLAM feature map to the self-driving platform.

Description

Adaptive Navigation Method and Cloud Navigation Path and Map Publishing Platform Utilizing it

The present invention relates to a self-driving platform or a road smart vehicle, in particular to an information processing platform that can generate adaptive navigation paths and maps on the cloud to assist the self-driving platform to operate accurately.

Today's self-driving platforms (such as self-driving cars, robots, etc.) or road smart vehicles (such as smart cleaning vehicles, unmanned vehicles) mainly use some sensing elements (such as cameras, lidars, and radars) to scan the surrounding environment of the path, use a positioning element (such as GNSS/INS) to capture the positioning information of the self-driving platform, and use a computing platform (Computing Platform) to perform a global path calculation (Glob al planning) program and an area path calculation (Local planning) program.

It is worth mentioning that the GNSS/INS of a traditional self-driving car will reduce its positioning accuracy due to the refraction or shading of satellite signals after entering the urban area, which will affect the driving safety of the self-driving car; and the GNSS/INS of the robot vehicle will also reduce its positioning accuracy due to the lack of satellite signals after entering the room, which will affect the driving safety of the robot.

In order to solve the above positioning uncertainty problem, the computing platform of a self-driving platform must use the GNSS/INS fusion lidar and camera as the relative positioning basis to execute a simultaneous positioning and mapping (Simultaneous Localization and Mapping; SLAM) program to improve positioning reliability. That is, in order to achieve good perception ability (Perception ability), the computing platform must fuse the sensing data of these sensing elements so that the shortcomings of each sensing element can be complemented and the advantages can be multiplied. That is to say, when a self-driving platform is running, it must calculate a large amount of surrounding environment perception data at any time.

From the above, it can be seen that the computing platform of a self-driving platform not only needs to calculate the positioning fusion results at any time, but the preprocessing high-precision electronic map, feature map and vector map required by the SLAM technology occupy a large amount of storage space on the computing platform.

In order to solve the above problems, a novel adaptive navigation method is urgently needed in the field.

The main purpose of the present invention is to disclose an adaptive navigation method, which can calculate the navigation path between two fixed points required by a self-driving platform on the cloud, construct a SLAM feature map corresponding to the navigation path, and send the navigation path and the SLAM feature map back to the self-driving platform, thereby eliminating the computing and storage burden of the self-driving platform.

Another object of the present invention is to disclose a cloud navigation path and map publishing platform, which can effectively reduce the computing power and data storage space requirements of multiple self-driving platforms wirelessly connected to it through the above-mentioned adaptive navigation method, thereby greatly reducing the follow-up maintenance and operation costs of each self-driving platform.

In order to achieve the aforementioned purpose, an adaptive navigation method is proposed, which is implemented by a cloud navigation path and map publishing platform, and it includes:

Execute a route planning operation based on the starting point and end point specified by a self-driving platform. The route planning operation includes calculating the shortest distance, the shortest time, or the lowest energy-consuming global navigation route based on the road network and weights in the map information stored in a high-precision electronic map storage unit. The factors of the weight include lane width, number of lanes, road edge width, road slope and road regulations. form a one-dimensional set and a one-dimensional set formed by the distance to the destination;

Reading the SLAM feature map within the scope of the global path from the high-precision electronic map storage unit, and performing a gridding process on the SLAM feature map to obtain a grid SLAM feature map; and

The sets of the global navigation paths and the grid SLAM feature map are transmitted to the self-driving platform.

In one embodiment, the adaptive navigation method further includes performing a detour path planning operation to provide at least one regional detour path when an obstacle appears in front of the self-driving platform and needs to change lanes.

In an embodiment, the detour path planning operation includes:

Receive the current location, heading, and speed information of the self-driving platform and combine the lane width, number of lanes, road edge width, road slope, and road regulations in the current driving direction in the map information of the high-precision electronic map storage unit to plan at least one regional detour path, and then guide to meet the path before driving in the global navigation path. Each of the regional detour paths includes a two-dimensional set of longitude and latitude. form a one-dimensional set and a one-dimensional set formed by the distance to the destination;

Read the SLAM feature map within the range of the detour path of each area from the high-precision electronic map storage unit;

obtaining a grid SLAM feature map according to the SLAM feature map; and

Send each detour route of the area and the grid SLAM feature map to the self-driving platform.

In order to achieve the aforementioned purpose, the present invention further proposes a cloud navigation path and map publishing platform, which implements an adaptive navigation method, the method comprising:

Execute a route planning operation based on the starting point and end point specified by a self-driving platform. The route planning operation includes calculating the shortest distance, the shortest time, or the lowest energy-consuming global navigation route based on the road network and weights in the map information stored in a high-precision electronic map storage unit. The factors of the weight include lane width, number of lanes, road edge width, road slope and road regulations. form a one-dimensional set and a one-dimensional set formed by the distance to the destination;

Reading the SLAM feature map within the scope of the global path from the high-precision electronic map storage unit, and performing a gridding process on the SLAM feature map to obtain a grid SLAM feature map; and

The sets of the global navigation paths and the grid SLAM feature map are transmitted to the self-driving platform.

In one embodiment, the adaptive navigation method further includes performing a detour path planning operation to provide at least one regional detour path when an obstacle appears in front of the self-driving platform and needs to change lanes.

In an embodiment, the detour path planning operation includes:

Receive the current location, heading, and speed information of the self-driving platform and combine the lane width, number of lanes, road edge width, road slope, and road regulations in the current driving direction in the map information of the high-precision electronic map storage unit to plan at least one regional detour path, and then guide to meet the path before driving in the global navigation path. Each of the regional detour paths includes a two-dimensional set of longitude and latitude. form a one-dimensional set and a one-dimensional set formed by the distance to the destination;

Read the SLAM feature map within the range of the detour path of each area from the high-precision electronic map storage unit;

obtaining a grid SLAM feature map according to the SLAM feature map; and

Send each detour route of the area and the grid SLAM feature map to the self-driving platform.

In order to enable your review committee members to further understand the structure, features and purpose of the present invention, drawings and detailed descriptions of preferred specific embodiments are hereby attached.

Please refer to FIG. 1 , which shows a block diagram of an embodiment of the cloud navigation path and map publishing platform of the present invention. As shown in FIG. 1 , a navigation route and map distribution platform 100 transmits navigation information to respective driving platforms 110 via a wireless network 10, wherein the navigation route and map distribution platform 100 has a network interface 101, a central processing unit 102, a global route calculation module 103, an area route calculation module 104, a high-precision electronic map storage unit 105 and a grid SLAM feature map storage unit 106.

The network interface 101 is used for communicating with respective driving platforms 110 via the wireless network 10 .

The central processing unit 102 is used to activate the global path calculation module 103 or the regional path calculation module 104 according to the navigation request information received by the network interface 101, to generate a grid SLAM feature map according to the map information stored in the high-precision electronic map storage unit 105, and store the grid SLAM feature map in the grid SLAM feature map storage unit 106, and send the packaged data of the grid SLAM feature map to a self-driving platform 110. That is to say, when a self-driving platform 110 needs navigation information from A to B, it only needs to transmit the locations of A and B to the navigation route and map distribution platform 100 via the wireless network 10, and the navigation route and map distribution platform 100 will send the route information connecting A and B to the self-driving platform 110.

In detail, the global route calculation module 103 executes a route planning algorithm based on the starting point, midway point (option), and end point information specified by a self-driving platform 110, such as but not limited to A* or Dijkstra’s common route planning algorithm, to calculate a global navigation route with the shortest distance, the shortest time, the lowest energy consumption, and the lane level of a suggested route according to the road network and weights in the map information stored in the high-precision electronic map storage unit 105. The global navigation route includes: longitude, A two-dimensional set of latitude, height information corresponding to the global navigation route (one-dimensional set), distance to the stop line ahead (one-dimensional set), and distance to the end point (one-dimensional set). Afterwards, the global path calculation module 103 will form these sets into a vector and send it back to the self-driving platform 110 . Please refer to FIG. 2 , which is a schematic diagram of the operation of the smart cloud navigation path and map publishing platform in FIG. 1 to obtain the global navigation path and grid SLAM feature map of the present invention. As shown in Figure 2, in the process of executing the path planning algorithm, the global path calculation module 103 will be a fixed point according to the starting point, midway point (option), and end point information specified by a self-driving platform 110; read road network information and height information from the high-precision electronic map storage unit 105 and generate a global path accordingly; read the SLAM feature map 106a within the scope of the global path from the high-precision electronic map storage unit 105; obtain a grid SLAM feature map according to the SLAM feature map 106a; The global path and the grid SLAM feature map are sent to the self-driving platform 110 .

In addition, the regional route calculation module 104 is used to execute a detour route planning algorithm when an obstacle appears in front of the self-driving platform 110 and needs to change lanes. The detour route planning algorithm includes: receiving the current position, heading, and speed information of the self-driving platform 110 and combining the lane width, number of lanes, road edge width, road slope and road regulations in the high-precision electronic map to plan multiple regional detour routes. The final guide will meet with the previous route in the global navigation route to continue driving. , wherein, the road edge is a path for overtaking, and each detour path in this area includes: a two-dimensional set composed of longitude and latitude, height information corresponding to each detour path in this area (one-dimensional set), distance from the stop line ahead (one-dimensional set) and distance to the end point (one-dimensional set). Afterwards, the area route calculation module 104 will form these sets into a vector and send it back to the self-driving platform 110 . Please refer to FIG. 3 , which is a schematic diagram of the operation of the smart cloud navigation path and map publishing platform in FIG. 1 to obtain the regional detour path and grid SLAM feature map of the present invention. As shown in Figure 3, in the process of executing the detour route planning algorithm, the regional route calculation module 104 will use the current position of a self-driving platform 110 as a fixed point; read road network information and height information from the high-precision electronic map storage unit 105 and generate multiple detour routes accordingly; read the SLAM feature maps within the range of these detour routes from the high-precision electronic map storage unit 105; obtain a grid SLAM feature map according to the SLAM feature map; Driving platform 110.

In addition, the grid SLAM feature map is based on the road SLAM features of the global navigation path or regional detour path of a self-driving platform 110, and the data is packaged in the unit of M meter x M meter grid and sent back to the self-driving platform 110 accordingly, wherein M is a positive real number.

In addition, the high-precision electronic map is mainly stored in the database of the navigation path and map publishing platform 100 in a structured form, wherein the database stores and describes the geometric information of points, lines, and surfaces and their corresponding attribute relationships in different forms for reference by the global route calculation module 103 or the regional route calculation module 104. In addition, SLAM feature map data that is unstructured or cannot be stored in a form can be stored in the form of an additional identification code (ID).

According to the above description, the present invention discloses an adaptive navigation method. Please refer to FIG. 4 , which shows a flow chart of an embodiment of an adaptive navigation method of the present invention, wherein the method is implemented by a cloud navigation route and map publishing platform. As shown in Figure 4, the adaptive navigation method includes: performing a route planning operation based on a starting point and an end point specified by a self-driving platform. The route planning operation includes calculating a global navigation route with the shortest distance, the shortest time, or the lowest energy consumption according to the road network and weights in the map information stored in a high-precision electronic map storage unit. The factors of the weight include lane width, number of lanes, road edge width, road slope, and road regulations. A dimension set, a one-dimensional set formed by the distance from the stop line ahead, and a one-dimensional set formed by the distance to the end point (step a); read the SLAM feature map within the scope of the global path from the high-precision electronic map storage unit, and carry out a gridding process on the SLAM feature map to obtain a grid SLAM feature map (step b); and transmit the sets of the global navigation path and the grid SLAM feature map to the self-driving platform (step c).

In addition, the adaptive navigation method may further include a detour path planning operation to provide at least one regional detour route when an obstacle appears in front of the self-driving platform and must change lanes. The detour route planning calculation includes: receiving the current position, heading, and speed information of the self-driving platform and combining the lane width, number of lanes, road edge width, road slope and road regulations in the current driving direction in the map information of the high-precision electronic map storage unit to plan at least one regional detour route, and then redirect to the front of the global navigation route. Path rendezvous, wherein each detour path in the area includes a two-dimensional set composed of longitude and latitude, a one-dimensional set composed of height information corresponding to the global navigation path, a one-dimensional set formed by distance from the stop line in front, and a one-dimensional set formed by distance to the end point; read the SLAM feature map within the range of each detour path in the area from the high-precision electronic map storage unit; obtain a grid SLAM feature map according to the SLAM feature map; and transmit each detour path and the grid SLAM feature map to the self-driving platform.

In addition, in order to allow readers to understand the technical solutions of the present invention, Figures 5 to 7 show various schematic maps generated during the execution of the adaptive navigation method of the present invention, wherein Figure 5 is a schematic diagram of a grid map used in the present invention; Figure 6a is a structurable vector map in the high-precision electronic map used in the present invention; Figure 6b is an illustration of expanding the attributes of points, lines, and planes in the vector map of Figure 6a and displaying lane widths and traffic lights; and Figure 7 is a cloud navigation of the present invention The route and map release platform sends back the global route of a self-driving platform (it is a short-distance bus connection case, the outbound journey is a green line, and the return journey is a red line) and the grid slam feature map (bright blue background) passed by.

With the design disclosed above, the present invention has the following advantages:

1. The adaptive navigation method of the present invention can calculate the navigation path between two fixed points required by a self-driving platform on the cloud, construct a SLAM feature map corresponding to the navigation path, and send the navigation path and the SLAM feature map back to the self-driving platform, thereby exempting the computing and storage burden of the self-driving platform.

2. The cloud navigation path and map publishing platform of the present invention can effectively reduce the computing power and data storage space requirements of multiple self-driving platforms wirelessly connected to it through the above-mentioned adaptive navigation method, thereby greatly reducing the follow-up maintenance and operation costs of each self-driving platform.

What is disclosed in this case is a preferred embodiment. For example, any partial changes or modifications derived from the technical ideas of this case and easily deduced by those who are familiar with the technology are within the scope of the patent right of this case.

To sum up, regardless of the purpose, means and efficacy of this case, it shows that it is very different from the conventional technology, and it is the first invention to be practical, which indeed meets the patent requirements of the invention. I sincerely hope that the review committee will be aware of it and grant a patent as soon as possible to benefit the society. This is my best prayer.

10: Wireless network 100: Navigation path and map publishing platform 101: Network interface 102: Central processing unit 103:Global path calculation module 104: Area path calculation module 105: High-precision electronic map storage unit 106:Grid SLAM feature map storage unit 106a: SLAM Feature Map 110: Self-driving platform Step a: Execute a path planning operation according to the starting point and end point specified by a self-driving platform. The path planning operation includes calculating the shortest distance, the shortest time or a global navigation path with the lowest energy consumption based on the road network and weights in the map information stored in a high-precision electronic map storage unit. Wherein, the factors of the weight include lane width, number of lanes, road edge width, road slope and road regulations. A one-dimensional set of distances from lines and a one-dimensional set of distances to endpoints Step b: read the SLAM feature map within the scope of the global path from the high-precision electronic map storage unit, and perform a grid processing on the SLAM feature map to obtain a grid SLAM feature map Step c: Send the sets of global navigation paths and the grid SLAM feature map to the self-driving platform

FIG. 1 shows a block diagram of an embodiment of the cloud navigation path and map publishing platform of the present invention. FIG. 2 is a schematic diagram of the operation of the cloud navigation path and map publishing platform in FIG. 1 to obtain the global navigation path and grid SLAM feature map of the present invention. FIG. 3 is a schematic diagram of the operation of the cloud navigation route and map publishing platform in FIG. 1 to obtain the regional detour route and the grid SLAM feature map of the present invention. FIG. 4 shows a flowchart of an embodiment of the adaptive navigation method of the present invention. FIG. 5 is a schematic diagram of a grid map used in the present invention. Fig. 6a is a structurable vector map in the high-precision electronic map adopted in the present invention. Fig. 6b is an illustration of expanding the attributes of points, lines, and planes in the vector map of Fig. 6a and displaying lane widths and signal lights. Fig. 7 is the global route and the grid slam characteristic map that the cloud navigation route and map publishing platform of the present invention sends back to a self-driving platform.

Step a: Execute a path planning operation according to the starting point and end point specified by a self-driving platform. The path planning operation includes calculating the shortest distance, the shortest time or a global navigation path with the lowest energy consumption based on the road network and weights in the map information stored in a high-precision electronic map storage unit. Wherein, the factors of the weight include lane width, number of lanes, road edge width, road slope and road regulations. A one-dimensional set of distances from lines and a one-dimensional set of distances to endpoints

Step b: read the SLAM feature map within the scope of the global path from the high-precision electronic map storage unit, and perform a grid processing on the SLAM feature map to obtain a grid SLAM feature map

Step c: Send the sets of global navigation paths and the grid SLAM feature map to the self-driving platform

Claims (2)

An adaptive navigation method is realized by a cloud navigation path and map publishing platform, which includes: performing a path planning operation according to a starting point and an end point specified by a self-driving platform, the path planning operation includes calculating a global navigation path with the shortest distance, the shortest time or the lowest energy consumption according to the road network and weights in the map information stored in a high-precision electronic map storage unit, wherein the factors of the weight include lane width, number of lanes, road edge width, road slope, and road regulations. The height information corresponding to the global navigation path constitutes a one-dimensional set, a one-dimensional set formed by the distance from the stop line ahead, and a one-dimensional set formed by the distance to the end point; read the SLAM feature map within the range of the global path from the high-precision electronic map storage unit, and perform a grid processing on the SLAM feature map to obtain a grid SLAM feature map; transmit the sets of the global navigation path and the grid SLAM feature map to the self-driving platform; Provide at least one regional detour route; wherein, the detour route planning calculation includes: receiving the current position, heading, and speed information of the self-driving platform and combining the lane width, number of lanes, road edge width, road slope and road regulations in the current driving direction in the map information of the high-precision electronic map storage unit to perform at least one detour The planning of the travel path is then directed to meet with the previous path in the global navigation path, wherein each detour path in this area includes a two-dimensional set composed of longitude and latitude, a one-dimensional set formed by height information corresponding to the global navigation path, a one-dimensional set formed by a distance from the stop line ahead, and a one-dimensional set formed by a distance to the end point; the SLAM feature map within the range of the detour path in each area is read from the high-precision electronic map storage unit; a grid SLAM feature map is obtained according to the SLAM feature map; Each detour path in the area and the grid SLAM feature map are sent to the self-driving platform. A cloud navigation path and map publishing platform, which implements an adaptive navigation method, the method includes: performing a path planning operation according to a starting point and an end point specified by a self-driving platform, the path planning operation includes calculating a global navigation path with the shortest distance, the shortest time, or the lowest energy consumption according to the road network and weights in the map information stored in a high-precision electronic map storage unit, wherein the factors of the weight include lane width, number of lanes, road edge width, road slope, and road regulations. The global navigation path includes a two-dimensional set composed of longitude and latitude. The height information corresponding to the global navigation path constitutes a one-dimensional set, a one-dimensional set formed by the distance from the stop line ahead, and a one-dimensional set formed by the distance to the end point; the SLAM feature map within the range of the global path is read from the high-precision electronic map storage unit, and a grid processing is performed on the SLAM feature map to obtain a grid SLAM feature map; Sending the sets of global navigation paths and the grid SLAM feature map to the self-driving platform; and performing a detour path planning operation to provide at least one area detour path when an obstacle appears in front of the self-driving platform and must change lanes; wherein, the detour path planning operation includes: receiving the current position, heading, and speed information of the self-driving platform and combining the lane width, number of lanes, road edge width, road slope, and road regulations in the current driving direction in the map information of the high-precision electronic map storage unit to perform at least one area detour. The planning of the path is then directed to meet with the path ahead of travel in the global navigation path, wherein each detour path in this area includes a two-dimensional set composed of longitude and latitude, a one-dimensional set formed by height information corresponding to the global navigation path, a one-dimensional set formed by the distance from the stop line ahead, and a one-dimensional set formed by the distance to the end point; read the SLAM feature map within the range of the detour path in each area from the high-precision electronic map storage unit; obtain a grid SLAM feature map according to the SLAM feature map; The detour path and the grid SLAM feature map are sent to the self-driving platform.