CN115143977A - A fast high-precision map construction method, device and vehicle thereof - Google Patents

Abstract

The invention discloses a rapid high-precision map construction method, which comprises the following steps: the method comprises the following steps that S1, in an outdoor scene, whether GNSS signals of the outdoor scene are good or not is judged according to original measurement data of GNSS, and if yes, preset useful data in original measurement data of a plurality of sensors preset on an unmanned vehicle are continuously processed in a preset format; s2, generating a running track of the unmanned vehicle and marking a special function point or a special task function area according to the preset useful data of the sensor processed in the step S1; step S3, processing the driving track of the unmanned vehicle obtained in the step S2 and the marked special function points or special task function areas to generate a high-precision map; and S4, carrying out integrity detection on the high-precision map to obtain the complete high-precision map. The method can effectively improve the construction efficiency of the high-precision map and simultaneously improve the utilization rate and the adaptability of the unmanned vehicle to the GNSS signals.

Description

A fast high-precision map construction method, device and vehicle thereof

technical field

The invention relates to the technical field of automatic driving and high-precision map construction, in particular to a fast high-precision map construction method, device and vehicle.

Background technique

Autonomous driving technology is a hot topic in recent years. In the fields of alleviating traffic congestion, improving road safety, and reducing air pollution, autonomous driving will bring disruptive changes.

In the process of commercialization of autonomous driving, unmanned sweepers, unmanned express delivery vehicles and unmanned taxis in limited area scenarios provide substantial application scenarios for the implementation of autonomous driving technology. The high-precision map can provide a priori map information for the vehicle. Therefore, the automatic driving technology based on the high-precision map has become the best solution for L4 and L5 autonomous driving in the industry.

With the increasing demand for autonomous driving in related industries, the scene of autonomous driving has become more and more clear, especially in outdoor scenes with good GNSS (Global Navigation Satellite System, global navigation satellite system) signals, unmanned sanitation, unmanned distribution vehicles and other vehicles The increasing demand for autonomous driving has put forward a test for the construction efficiency, quality and accuracy of autonomous driving maps.

At present, in the autonomous driving industry, the most widely used map construction methods are the methods of constructing point clouds or grid maps based on laser or visual data. The high-precision map construction method is mainly divided into two strategies: offline and online.

Among them, the strategy of constructing high-precision maps offline is to first collect relevant sensor data through vehicles and other acquisition equipment, then transmit the offline data to the local or cloud, and finally complete the construction of high-precision maps locally or in the cloud. The strategy of constructing high-precision maps offline involves tedious processes such as uploading and downloading of original data, backup and distribution of result data, which requires more human intervention, and has a low degree of automation, resulting in reduced efficiency and serious waste of human resources and local computing resources. .

For the strategy of building high-precision maps online, firstly, the relevant sensor data of the vehicle is obtained in real time, and the local map is obtained by calculating the sensor data. Secondly, the map is optimized in real time according to the set strategy, and finally, the high-precision global consistency is obtained. map. For the strategy of online construction of high-precision maps, since the point cloud or grid map construction tasks are completed in the on-board processor, the performance requirements are very high, the large map is difficult to process, and the map quality and error cannot be guaranteed.

For the existing methods of constructing point clouds or grid maps based on laser or visual data, it is not friendly to the construction of high-precision maps in scenes with good GNSS signals such as squares. In the scenario where the GNSS signal is good and needs to be simple, the construction process of the point cloud or raster map is complex, time-consuming and resource-intensive.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a fast high-precision map construction method, device, and vehicle for the technical defects existing in the prior art.

To this end, the present invention provides a fast high-precision map construction method, which includes the following steps:

Step S1, in the outdoor scene, obtain the original measurement data of multiple preset sensors including GNSS installed on the unmanned vehicle, and then judge whether the GNSS signal of the outdoor scene is good according to the original measurement data of GNSS, and if so, Then continue to perform preset format processing operations on the preset useful data in the original measurement data of multiple sensors preset on the unmanned vehicle;

Step S2, according to the preset useful data of the preset multiple sensors processed in step S1, generate the driving track of the unmanned vehicle and mark the special function point or special task function area;

Step S3, processing the driving track of the unmanned vehicle and the marked special function point or special task function area obtained in step S2 to generate a high-precision map;

In step S4, integrity detection is performed on the high-precision map generated in step S3, and a complete high-precision map is finally obtained.

Preferably, step S1 specifically includes the following sub-steps:

Step S11, in the outdoor scene, obtain the original measurement data of a plurality of preset sensors including GNSS installed on the unmanned vehicle, and then judge whether the GNSS signal of the outdoor scene is good according to the original measurement data of the GNSS, and if so, Then continue to execute step S12;

Among them, multiple sensors are preset, including GNSS, IMU and wheel speedometer;

Step S12, according to the format requirements of the extended Kalman filter EKF fusion algorithm, parse the preset useful data in the original measurement data in step S11 into a corresponding specific format, and continue to perform step S2.

Preferably, in step S11, the GNSS is used to provide the original measurement data of the GNSS signal of the unmanned vehicle, which specifically includes: the latitude and longitude of the GNSS signal, the heading, the speed in the northeast sky direction, the number of stars received, the state, the heading mark position and horizontal precision factor;

IMU, which is used to provide the original measurement data of the IMU of the unmanned vehicle, including: acceleration and angular velocity in the x, y and z directions in the IMU coordinate system;

Wheel speedometer, which is used to provide the original measurement data of the wheel speed of the unmanned vehicle, including the speed of the left wheel and the speed of the right wheel of the unmanned vehicle;

In step S12, useful data is preset in the original measurement data of step S11, specifically including the latitude and longitude, heading, and speed in the northeast direction in the GNSS signal measured by GNSS, and the speed in the x, y and z directions measured by the inertial measurement unit IMU. Acceleration and angular velocity, as well as the left and right wheel speeds of the unmanned vehicle as measured by the wheel tachometer;

Wherein, in step S11, when the state of the GNSS signal, the number of stars received, the horizontal precision factor and the heading flag meet the requirements, and reach or exceed the set threshold, it is judged that the GNSS signal of the outdoor scene is good.

Preferably, step S2 specifically includes the following sub-steps:

Step S21, according to the preset useful data of the inertial measurement unit IMU and the preset useful data of the wheel speedometer processed in step S1, calculate the change amount of the unmanned vehicle in the preset period, and combine with the global navigation satellite system. The fusion pose of the unmanned vehicle calculated by GNSS in the previous preset period, and the latest predicted pose of the unmanned vehicle after the preset period is calculated;

In step S22, the preset useful data of the GNSS processed in step S1, the latest predicted pose of the unmanned vehicle obtained in step S21, the attitude information calculated and provided by the inertial measurement unit IMU, and the unmanned vehicle measured by the wheel speedometer are used. The speed of the left wheel and the speed of the right wheel of the vehicle are processed by the extended Kalman filter EKF fusion algorithm to obtain the fusion pose information of the unmanned vehicle and the driving trajectory of the unmanned vehicle with high precision and fusion processing;

Wherein, the fusion pose information includes fusion position information and fusion attitude information;

In step S23, it is judged whether the adaptation of the current outdoor scene is completed. If it is not completed, return to step S1. If the scene adaptation is completed, continue to execute step S3, that is, input the stored and marked data into the following steps. Step S3.

Preferably, in step S22, it also includes the step of: marking the special function point or the special task function area according to the needs of the operating user, which specifically includes the following processing steps:

Step S220, for the fusion location information of the unmanned vehicle that has undergone fusion processing, detect in real time whether the operating user has a marking requirement for inputting special function points or special task functional areas to the fusion location, and if so, corresponding to the fusion location. Mark the special function point or special task function area, and modify the function attribute information of the special function point or special task function area, if not, store the pose information of the fusion-processed unmanned vehicle in real time.

Preferably, step S3 specifically includes the following sub-steps:

Step S31, by extending the travel trajectory of the unmanned vehicle obtained in step S2 by a preset distance value K outward, to obtain a passable area where the unmanned vehicle travels in the current outdoor scene, and store it in a specific format;

In step S31, for different outdoor scenes, there are corresponding preset distance values K respectively;

Step S32, according to the functional attribute information of the special function point or the special task functional area marked in step S2, generate the corresponding special function point pose information or the functional area boundary and the passable area boundary of the special task functional area;

Step S33, according to the existing zigzag strategy or the global path planning strategy, according to the corresponding relationship between the special function point and the special task functional area, the boundary of the passable area and the boundary information of the special task functional area, automatically generate the unmanned vehicle in the normal state. The reference path during operation, and based on the corresponding relationship between the special function points and the special task function area, generate all the associated operation tasks in the current outdoor scene, and then store them to obtain a high-precision map.

Preferably, step S4 specifically includes the following sub-steps:

Step S41, check whether the file of the high-precision map generated in step S3 is complete, if yes, continue to execute step S42, if not, return to execute step S1;

Step S42: Check whether the preset high-precision necessary elements in the high-precision map generated in step S3 are complete, if so, use the high-precision map generated in step S3 as the final high-precision map to be released, if not, return Step S1 is performed.

Preferably, in step S42, the preset high-precision necessary elements include special function points, special task function areas, reference paths and tasks;

Among them, in the high-precision map generated in the detection step S3, whether the preset high-precision necessary elements are complete, specifically: for the three elements of the special function point, the special task function area, and the reference path, detect whether they exist, and if so, The three elements are considered complete; for tasks, it is detected whether the reference path corresponding to the operation task is connected. If it is connected, the task is considered complete; otherwise, the task is considered incomplete.

In addition, the present invention also provides a fast high-precision map construction device, which includes the following modules:

The sensor data preprocessing module is used to obtain the raw measurement data of multiple preset sensors including GNSS installed on the unmanned vehicle in the outdoor scene, and then according to the raw measurement data of GNSS (ie Global Navigation Satellite System), Determine whether the GNSS signal of the outdoor scene is good, and if so, continue to perform the preset format processing operation on the preset useful data in the original measurement data of the preset multiple sensors on the unmanned vehicle;

The vehicle trajectory and function information processing module is connected with the sensor data preprocessing module, and according to the preset useful data of the preset multiple sensors processed by the sensor data preprocessing module, generates the driving trajectory of the unmanned vehicle and marks the special function points or Special task functional area;

The high-precision map processing module is connected to the vehicle trajectory and function information processing module, and is used to process the driving trajectory of the unmanned vehicle and the marked special function points or special task function areas obtained from the vehicle trajectory and function information processing module. , generate a high-precision map, and then send it to;

The map integrity detection module is connected with the high-precision map processing module, and is used to perform integrity detection on the high-precision map generated by the high-precision map processing module, and finally obtain a complete high-precision map.

In addition, the present invention also provides a vehicle, including the aforementioned rapid high-precision map construction device.

It can be seen from the above technical solutions provided by the present invention that, compared with the prior art, the present invention provides a fast high-precision map construction method, device and vehicle, which are scientifically designed and can effectively improve the construction efficiency of high-precision maps At the same time, it improves the utilization and adaptability of unmanned vehicles to GNSS signals, especially the adaptability to environments such as large squares, which has great practical significance.

For the present invention, it is an efficient, low-consumption and high-precision map construction method, the design principle is scientific, easy to implement, the logic is clear, the adaptability to the scene with good GNSS signal is good, and it satisfies the high-precision scene map of the unmanned vehicle. The technical requirements of high construction efficiency, computing resource consumption and accuracy are conducive to ensuring the safety of autonomous vehicles, accelerating the commercialization of unmanned vehicles, and ensuring reliable commercialization.

Description of drawings

Fig. 1 is the main flow chart of a kind of fast high-precision map construction method provided by the present invention;

Fig. 2 is the overall work flow chart of a kind of fast high-precision map construction method provided by the present invention;

3 is a schematic diagram of a passable area of an unmanned vehicle in an outdoor scene obtained by applying the present invention.

Detailed ways

In order to make those skilled in the art better understand the solution of the present invention, the present invention is further described in detail below with reference to the accompanying drawings and embodiments.

Referring to Fig. 1 and Fig. 2, the present invention provides a fast high-precision map construction method, including the following steps:

Step S1, in the outdoor scene, obtain the original measurement data of a plurality of preset sensors including GNSS installed on the unmanned vehicle, and then judge the GNSS of the outdoor scene according to the original measurement data of GNSS (ie global navigation satellite system). Whether the signal is good, if so, continue to perform preset format processing operations on the preset useful data in the original measurement data of multiple preset sensors on the unmanned vehicle;

It should be noted that if the GNSS signal is not good, the next step will not be performed. At this time, the existing high-precision map construction method can be selected.

In the present invention, step S1 specifically includes the following sub-steps:

Step S11, in the outdoor scene, obtain the original measurement data of a plurality of preset sensors including GNSS installed on the unmanned vehicle, and then judge the GNSS of the outdoor scene according to the original measurement data of GNSS (that is, global navigation satellite system). Whether the signal is good, if yes, continue to execute step S12, otherwise, do not continue to execute;

Among them, a plurality of sensors are preset, including GNSS (ie global navigation satellite system), IMU (Inertial measurement unit, inertial measurement unit) and wheel speedometer;

It should be noted that, in the case of a square with good GNSS signal, the GNSS signal is good, that is, the status of the GNSS signal, the number of satellites received, the horizontal precision factor and the heading flag meet the requirements and meet or exceed the set threshold. .

In step S11, when the state of the GNSS signal, the number of stars received, the horizontal precision factor and the heading flag position meet the requirements and reach or exceed the set threshold, it is determined that the GNSS signal of the outdoor scene is good.

It should be noted that GNSS (Global Navigation Satellite System) is a space-based radio navigation and positioning system that can provide users with all-weather three-dimensional coordinates, speed and time information anywhere on the earth's surface or near-Earth space.

In the present invention, in step S11, the global navigation satellite system GNSS is used to provide the original measurement data of the GNSS signal of the unmanned vehicle, which specifically includes: the latitude and longitude of the GNSS signal, the heading, the speed in the northeast sky direction, the number of stars received, Status, heading flags and horizontal precision factor.

It should be noted that the number of stars received, the state, the heading flag and the horizontal precision factor of the GNSS signal are evaluation parameters used to judge whether the GNSS signal is good.

The inertial measurement unit IMU is used to provide the original measurement data of the IMU of the unmanned vehicle, including: acceleration and angular velocity in the x, y and z directions in the IMU coordinate system;

The wheel speed meter is used to provide the original measurement data of the wheel speed of the unmanned vehicle, including the speed of the left wheel and the speed of the right wheel of the unmanned vehicle.

Step S12, according to the format requirements of the EKF (Extended Kalman Filter, Extended Kalman Filter) fusion algorithm, the preset useful data in the original measurement data of Step S11 is parsed into a corresponding specific format (for example, png, xml and other formats), And continue to step S2.

In step S12, useful data is preset in the original measurement data of step S11, specifically including the latitude and longitude, heading, and the speed of the northeast direction in the GNSS signal measured by GNSS, and the three directions (x, y) measured by the inertial measurement unit IMU. , z) acceleration and angular velocity, and the left and right wheel speeds of the unmanned vehicle measured by the wheel speedometer.

It should be noted that, for the present invention, S1 does not involve data storage. After extracting the required useful data from the original measurement data of each sensor, the amount of data obtained is very small, and the single-frame sensor data is processed in real time. Therefore, it can be stored directly in the cache. The parsing method is also very simple. The preset useful data required by each sensor is directly extracted and stored in a defined data structure (such as png, xml and other formats), so that it can be input to the next step for operation.

Step S2, according to the preset useful data of the preset multiple sensors processed in step S1, generate the driving track of the unmanned vehicle and mark the special function point or special task function area;

In the present invention, step S2 specifically includes the following sub-steps:

Step S21, according to the preset useful data of the inertial measurement unit IMU processed in step S1 and the preset useful data of the wheel speedometer, calculate the pose ( Including the position and attitude) changes, combined with the unmanned vehicle fusion pose and attitude calculated by the global navigation satellite system GNSS in the previous preset cycle (specifically at a certain moment in the cycle), calculate the unmanned vehicle in the preset period. The latest predicted pose (including position and pose) after setting the period;

It should be noted that, in step S21, the amount of change in the pose of the unmanned vehicle within a preset period (that is, a period of time with a preset length) specifically includes: the change in the speed and position in the east direction, the speed and position in the north direction The amount of change, the amount of change in the sky speed and position, and the amount of change in the roll angle (roll), pitch angle (pitch) and yaw angle (yaw).

In step S21, the preset period is related to the vehicle speed, the preset period is inversely proportional to the vehicle speed, the higher the vehicle speed, the shorter the period, according to the average speed of the passenger car, the recommended preset period is T (T is less than or equal to 0.01ms )

In step S21, according to the preset useful data of the inertial measurement unit IMU and the preset useful data of the wheel speedometer processed in step S1, the position of the unmanned vehicle within a preset period (ie, a period of preset length) is calculated. The amount of change in posture, this calculation processing operation, is a well-known technical common sense, and a conventional calculation method is used, which will not be repeated here. For example, if the wheel speedometer measures the speed, then the speed can be integrated within the cycle to obtain the position change; the angular velocity of the IMU can be integrated to obtain the angle change, the acceleration can be integrated once to obtain the speed change, and the second integration can be used to obtain the position change. quantity.

It should be noted that, in step S21, the algorithm of the present invention operates in a strict periodic system, which is also the preset period T mentioned above, which can ensure that the frequency of the fusion pose output is fixed. In terms of specific implementation, the fusion pose of the unmanned vehicle calculated by GNSS in the last preset period is obtained at a time point (moment) in the last preset period T, and the fusion pose is output at the time point.

It should be noted that the pose provided by GNSS at each moment is an absolute pose. The pose change amount of the unmanned vehicle in a preset period (that is, a period of preset length), plus the fusion pose of the unmanned vehicle calculated by the global navigation satellite system GNSS in the previous preset period, can be obtained The latest predicted pose after a preset period.

Step S22, the preset useful data of the global navigation satellite system GNSS processed in step S1, the latest predicted pose of the unmanned vehicle obtained in step S21, the attitude information calculated by the inertial measurement unit IMU and provided by the wheel speedometer. The measured left wheel speed and right wheel speed of the unmanned vehicle are processed by the EKF (Extended Kalman Filter, Extended Kalman Filter) fusion algorithm to obtain high-precision, fusion-processed unmanned vehicle fusion pose information (including position and attitude) and the driving trajectory of the unmanned vehicle; wherein, the fusion pose information includes fusion position information and fusion attitude information;

It should be noted that the EKF (Extended Kalman Filter, Extended Kalman Filter) fusion algorithm is an existing well-known algorithm and a very mature technology. , to get the final high-precision pose of the unmanned vehicle at this moment.

In the present invention, in step S22, by storing the fused poses of the unmanned vehicle at each moment together, in the order of time, a driving trajectory of the unmanned vehicle can be formed.

In the present invention, in step S22, it also includes the step of: marking special function points or special task function areas according to the needs of operating users. Specifically, the following processing steps are included:

Step S220, for the fusion location information of the unmanned vehicle that has undergone fusion processing, detect in real time whether the operating user has a marking requirement for inputting special function points or special task functional areas to the fusion location, and if so, corresponding to the fusion location. Mark the special function point or special task function area, and modify the function attribute information (ie specific function information) of the special function point or special task function area, if not, store the pose information of the fusion-processed unmanned vehicle in real time (including position and attitude).

It should be noted that the special function point is a pose point (similar to a bus stop) with special functions, such as the departure point, the stop point and the end point of the passenger car. The special function point is that the customer selects the function point attribute through the button on the interactive page and issues it. The algorithm will detect whether the command exists at the end of each cycle. If it exists, record the fusion pose and function point attribute at this moment (ie specific functions or functions).

Special task functional area: It is an area with special functions, such as the speed limit area for passenger cars, the cleaning area for sanitation vehicles, as well as pedestrian lanes, intersections, uphill and downhill areas on the road, and special marking areas. The implementation method and function point are the same. Only the agreement is different.

In step S23, it is judged whether the adaptation of the current outdoor scene is completed. If it is not completed, return to step S1. If the scene adaptation is completed, continue to execute step S3, that is, input the stored and marked data into the following steps. Step S3.

In the present invention, in step S23, adaptation: refers to the entire process of constructing a high-precision map of the operating environment of the unmanned vehicle.

In step S23, it is judged that the adaptation is completed: the corresponding protocol is specified, and the client sends the end of the adaptation process through the relevant button on the interactive page.

That is, in step S23, it is determined whether the adaptation of the current outdoor scene is completed, specifically: when an adaptation end instruction input by the client (user) is received, it is determined that the adaptation of the outdoor scene is completed. If the adaptation end instruction input by the client (user) is not received, it is determined that the adaptation of the outdoor scene is not completed.

Step S3, processing the driving track of the unmanned vehicle and the marked special function point or special task function area obtained in step S2 to generate a high-precision map;

In the present invention, step S3 specifically includes the following sub-steps:

Step S31, by extending the travel trajectory of the unmanned vehicle obtained in step S2 to the outside (for example, in all directions of the three-dimensional space) by extending the preset distance value (that is, the K value, the K value is no value when the current outdoor scene is adapted. The average distance between the running route of the human vehicle and the scene boundary, this value is a fixed value of the adaptation constraint), and the passable area of the unmanned vehicle in the current outdoor scene is obtained. )storage;

In the present invention, in step S31, the preset distance value K is an empirical value, for example, if the adaptation scene is a standard urban road, the preset value is several lane width values, and if it is a square scene, it is A number of vehicle width values, if it is a park scene, it is half of the road width, etc. This value will choose different fixed values according to the scene (one-to-one correspondence is a constraint). For example, as shown in Figure 3. 3 is a schematic diagram of a passable area of an unmanned vehicle in an outdoor scene obtained by applying the present invention.

In step S31, for different outdoor scenes, there are corresponding preset distance values K respectively.

In step S31, the travel trajectory of the unmanned vehicle obtained in step S2 is extended to the outside (for example, in all directions of the three-dimensional space) by the preset distance value. The specific operation is as follows: Referring to FIG. The fusion position is the center, and the square area with a distance of K is expanded outward in a "cross" shape (that is, four directions in front, behind, left and right).

In the present invention, in step S31, the passable area: the area where the vehicle can travel, and the impassable area is the area where the vehicle cannot travel. The boundary between the two is a high-precision map boundary, which can be understood as an electronic fence. As mentioned above, the area is expanded outward in a "cross" shape, and all the fusion positions are fully expanded. The area within the area is a passable area, and the area not within the expanded area is an impassable area.

Step S32, according to the function attribute information of the special function point or the special task functional area marked in step S2 (specifically marked on the fusion pose of the unmanned vehicle), generate the corresponding special function point pose information or the function of the special task functional area. area boundaries and traversable area boundaries.

It should be noted that the boundary of the passable area includes the boundary of the functional area.

It should be noted that the pose information of the function point includes: acceleration and angular velocity in three directions (x, y, z), as well as roll angle (roll), pitch angle (pitch) and yaw angle (yaw). The function of the point pose information is to ensure that the unmanned vehicle meets the requirements of the vehicle pose here.

Functional area boundary: It is a set of curves consisting of several position points that only contain three directions (x, y, z). The function is to regulate the vehicle to perform different operations or tasks in different areas, such as deceleration and speed limit. , turn, etc.

In the present invention, the acquisition method of the functional area boundary of the special task functional area is the same as the acquisition method of the functional point and the functional area, but the protocol is different, including a starting point and an end point, and the starting point and the end point of the functional area appear in pairs. .

In the present invention, as shown in FIG. 3 , according to the “cross” shape (that is, the four directions of the front, the rear, the left and the right), it expands outward, and the expanded road section is the starting point and the end point of the functional area. and the generated functional area boundaries are simply and smoothly connected to obtain boundary curves.

In the present invention, in step S32, it should be noted that only some types of function points are in the functional area of special tasks, and such function points need to automatically generate corresponding tasks according to the established task mode rules (Provide the global relationship between task start and end point for downstream decision planning).

In the present invention, in step S32, it should be noted that the special function point must be the start or end point of the special task function area, which is the corresponding relationship between the special function point and the special task function area.

Step S33, according to the existing zigzag strategy or the global path planning strategy, according to the corresponding relationship between the special function point and the special task functional area, the boundary of the passable area and the boundary information of the special task functional area, automatically generate the unmanned vehicle in the normal state. The reference path during operation, and based on the corresponding relationship between the special function point and the special task functional area (that is, the special function point is the start or end point of the special task functional area), generate all associations in the current outdoor scene (there is a corresponding relationship) ), and then store them to obtain high-precision maps.

In step S33, it should be noted that the "back" shape is to achieve full coverage cleaning for the cleaning area of the sweeper, and a method of automatically generating a full coverage reference path is an existing conventional method of automatically generating a full coverage reference path. Methods. Global path planning is for passenger cars, which only need to pass through this road without performing various special tasks.

It should be noted that, limited by the vehicle hardware, the vehicle has a minimum turning radius. Therefore, in step S33, the curve radius at the turning point of the reference path of the unmanned vehicle during normal operation needs to be greater than the minimum turning radius of the unmanned vehicle. Turning radius.

It should be noted that, in step S33, regarding the reference path of the unmanned vehicle during normal operation, during normal operation, the unmanned vehicle needs to travel along this route

In step S33, it should be noted that, based on the existing back-shaped strategy or the global path planning strategy (such as the A* algorithm), according to the corresponding relationship between the special function point and the special task function area, the boundary of the passable area and the special The boundary information of the task functional area can automatically generate the reference path of the unmanned vehicle during normal operation. This is a known technology in the prior art, and details are not repeated here.

In step S33, with regard to all possible operation tasks in the current outdoor scene, it refers to the reference route generated through the corresponding relationship between the special function point and the special task function area, and the customer can select the reference route through the interactive interface. The designated tasks are operated.

For the present invention, according to the corresponding relationship between the special function point and the special task functional area, several related reference paths are generated, and each related reference path is an operation task, which is combined into all related operation tasks .

Step S4, perform integrity detection on the high-precision map generated in step S3, and finally obtain a complete high-precision map;

In the present invention, step S4 specifically includes the following sub-steps:

Step S41, check whether the file of the high-precision map generated in step S3 is complete, if yes, continue to execute step S42, if not, return to execute step S1;

In step S41, it is detected whether the files of the high-precision map generated in step S3 are complete, specifically: detecting whether all folders are generated in step S3 (that is, whether there is a problem that some folders are not generated), and whether all result files are generated ( That is, whether there is a problem that some result files are not generated). Without this detection step, high-precision integrity cannot be guaranteed, and autonomous driving cannot be achieved.

Step S42: Check whether the preset high-precision necessary elements in the high-precision map generated in step S3 are complete, if so, use the high-precision map generated in step S3 as the final high-precision map to be released, if not, return Step S1 is performed.

In step S42, the preset high-precision necessary elements include special function points, special task function areas, reference paths and tasks.

Among them, in the high-precision map generated in the detection step S3, whether the preset high-precision necessary elements are complete, specifically: for the three elements of the special function point, the special task function area, and the reference path, detect whether they exist, and if so, Then these three elements are considered complete; for tasks, it is to detect whether the reference path corresponding to the operation task (such as the main operation task selected and input by the customer) is connected (that is, not disconnected). If it is connected, the task is considered complete. , otherwise, the task is considered incomplete.

It should be noted that, for the present invention, if one of the two detections in step S41 and step S42 is not satisfied, delete the high-precision map data generated in step S3, re-adapt the map of the scene, and then re-step S1 ~ S4; if the detection is satisfied, the construction of the map is ended, and the high-precision map is released.

It should be noted that, for the fast high-precision map construction method provided by the present invention, an outdoor scene (such as a square) with a good GNSS signal is adapted according to a specific route, and according to the high-precision trajectory of the vehicle during the adaptation process and marked special function points or special task function area data, automatically generate a high-precision map including scene passable area, special function point or special task area, reference path and operational tasks, etc., and finally achieve scene map construction efficiency and computing resources. Low consumption and accuracy requirements improve the construction efficiency of high-precision maps, and at the same time improve the utilization and adaptability of unmanned vehicles to GNSS signals, especially the adaptability of environments such as large squares.

In addition, based on a fast high-precision map construction method provided by the present invention, in order to execute the above-mentioned fast high-precision map construction method, the present invention also provides a fast high-precision map construction device, the device includes the following modules:

The sensor data preprocessing module is used to obtain the raw measurement data of multiple preset sensors including GNSS installed on the unmanned vehicle in the outdoor scene, and then according to the raw measurement data of GNSS (ie Global Navigation Satellite System), Determine whether the GNSS signal of the outdoor scene is good, and if so, continue to perform the preset format processing operation on the preset useful data in the original measurement data of the preset multiple sensors on the unmanned vehicle;

The vehicle trajectory and function information processing module is connected with the sensor data preprocessing module, and according to the preset useful data of the preset multiple sensors processed by the sensor data preprocessing module, generates the driving trajectory of the unmanned vehicle and marks the special function points or Special task functional area;

The high-precision map processing module is connected to the vehicle trajectory and function information processing module, and is used to process the driving trajectory of the unmanned vehicle and the marked special function points or special task function areas obtained from the vehicle trajectory and function information processing module. , generate a high-precision map, and then send it to;

The map integrity detection module is connected with the high-precision map processing module, and is used to perform integrity detection on the high-precision map generated by the high-precision map processing module, and finally obtain a complete high-precision map.

In addition, the present invention also provides a vehicle including the aforementioned rapid high-precision map construction device.

To sum up, compared with the prior art, the present invention provides a fast high-precision map construction method, device, and vehicle, which are scientifically designed, can effectively improve the construction efficiency of high-precision maps, and improve the The utilization and adaptability of people and vehicles to GNSS signals, especially the adaptability of environments such as large squares, has great practical significance.

For the present invention, it is an efficient, low-consumption and high-precision map construction method, the design principle is scientific, easy to implement, the logic is clear, the adaptability to the scene with good GNSS signal is good, and it satisfies the high-precision scene map of the unmanned vehicle. The technical requirements of high construction efficiency, computing resource consumption and accuracy are conducive to ensuring the safety of autonomous vehicles, accelerating the commercialization of unmanned vehicles, and ensuring reliable commercialization.

The above are only the preferred embodiments of the present invention. It should be pointed out that for those skilled in the art, without departing from the principles of the present invention, several improvements and modifications can be made. It should be regarded as the protection scope of the present invention.

Claims (10)

1. a fast high-precision map construction method, is characterized in that, comprises the following steps: Step S1, in the outdoor scene, obtain the original measurement data of multiple preset sensors including GNSS installed on the unmanned vehicle, and then judge whether the GNSS signal of the outdoor scene is good according to the original measurement data of GNSS, and if so, Then continue to perform preset format processing operations on the preset useful data in the original measurement data of multiple sensors preset on the unmanned vehicle; Step S2, according to the preset useful data of the preset multiple sensors processed in step S1, generate the driving track of the unmanned vehicle and mark the special function point or special task function area; Step S3, processing the driving track of the unmanned vehicle and the marked special function point or special task function area obtained in step S2 to generate a high-precision map; In step S4, integrity detection is performed on the high-precision map generated in step S3, and a complete high-precision map is finally obtained. 2. The high-precision map construction method as claimed in claim 1, wherein step S1 specifically comprises the following sub-steps: Step S11, in the outdoor scene, obtain the original measurement data of a plurality of preset sensors including GNSS installed on the unmanned vehicle, and then judge whether the GNSS signal of the outdoor scene is good according to the original measurement data of GNSS, and if so, Then continue to execute step S12; Among them, multiple sensors are preset, including GNSS, IMU and wheel speedometer; Step S12, according to the format requirements of the extended Kalman filter EKF fusion algorithm, parse the preset useful data in the original measurement data in step S11 into a corresponding specific format, and continue to perform step S2. 3. high-precision map construction method as claimed in claim 1 is characterized in that, in step S11, GNSS, for providing the original measurement data of GNSS signal of unmanned vehicle, specifically comprises: latitude, longitude, heading, north Speed in the east direction, number of stars received, status, heading mark position and horizontal precision factor; IMU, which is used to provide the original measurement data of the IMU of the unmanned vehicle, including: acceleration and angular velocity in the x, y and z directions in the IMU coordinate system; Wheel speedometer, which is used to provide the original measurement data of the wheel speed of the unmanned vehicle, including the speed of the left wheel and the speed of the right wheel of the unmanned vehicle; In step S12, useful data is preset in the original measurement data of step S11, specifically including the latitude and longitude, heading, and speed in the northeast direction in the GNSS signal measured by GNSS, and the speed in the x, y and z directions measured by the inertial measurement unit IMU. Acceleration and angular velocity, as well as the left and right wheel speeds of the unmanned vehicle as measured by the wheel tachometer; Wherein, in step S11, when the state of the GNSS signal, the number of stars received, the horizontal precision factor and the heading flag meet the requirements, and reach or exceed the set threshold, it is judged that the GNSS signal of the outdoor scene is good. 4. The high-precision map construction method as claimed in claim 1, wherein step S2 specifically comprises the following sub-steps: Step S21, according to the preset useful data of the inertial measurement unit IMU and the preset useful data of the wheel speedometer processed in step S1, calculate the change amount of the unmanned vehicle in the preset period, and combine with the global navigation satellite system. The fusion pose of the unmanned vehicle calculated by GNSS in the previous preset period, and the latest predicted pose of the unmanned vehicle after the preset period is calculated; In step S22, the preset useful data of the GNSS processed in step S1, the latest predicted pose of the unmanned vehicle obtained in step S21, the attitude information calculated and provided by the inertial measurement unit IMU, and the unmanned vehicle measured by the wheel speedometer are used. The speed of the left wheel and the speed of the right wheel of the vehicle are processed by the extended Kalman filter EKF fusion algorithm to obtain the fusion pose information of the unmanned vehicle and the driving trajectory of the unmanned vehicle with high precision and fusion processing; Wherein, the fusion pose information includes fusion position information and fusion attitude information; In step S23, it is judged whether the adaptation of the current outdoor scene is completed. If it is not completed, return to step S1. If the scene adaptation is completed, continue to execute step S3, that is, input the stored and marked data into the following steps. Step S3. 5. The method for constructing a high-precision map according to claim 4, wherein in step S22, it also includes the step of: marking special function points or special task function areas according to the needs of operating users, specifically including the following processing steps : Step S220, for the fusion location information of the unmanned vehicle that has undergone fusion processing, detect in real time whether the operating user has a marking requirement for inputting special function points or special task functional areas to the fusion location, and if so, corresponding to the fusion location. Mark the special function point or special task function area, and modify the function attribute information of the special function point or special task function area, if not, store the pose information of the fusion-processed unmanned vehicle in real time. 6. The high-precision map construction method as claimed in claim 1, wherein step S3 specifically comprises the following sub-steps: Step S31, by extending the travel trajectory of the unmanned vehicle obtained in step S2 by a preset distance value K outward, to obtain a passable area where the unmanned vehicle travels in the current outdoor scene, and store it in a specific format; In step S31, for different outdoor scenes, there are corresponding preset distance values K respectively; Step S32, according to the functional attribute information of the special function point or the special task functional area marked in step S2, generate the corresponding special function point pose information or the functional area boundary and the passable area boundary of the special task functional area; Step S33, according to the existing zigzag strategy or the global path planning strategy, according to the corresponding relationship between the special function point and the special task functional area, the boundary of the passable area and the boundary information of the special task functional area, automatically generate the unmanned vehicle in the normal state. The reference path during operation, and based on the corresponding relationship between the special function points and the special task function area, generate all the associated operation tasks in the current outdoor scene, and then store them to obtain a high-precision map. 7. The high-precision map construction method as claimed in claim 1, wherein step S4 specifically comprises the following sub-steps: Step S41, check whether the file of the high-precision map generated in step S3 is complete, if yes, continue to execute step S42, if not, return to execute step S1; Step S42: Check whether the preset high-precision necessary elements in the high-precision map generated in step S3 are complete, if so, use the high-precision map generated in step S3 as the final high-precision map to be released, if not, return Step S1 is performed. 8. The method for constructing a high-precision map according to claim 7, wherein in step S42, the preset high-precision necessary elements include special function points, special task function areas, reference paths and tasks; Among them, in the high-precision map generated in the detection step S3, whether the preset high-precision necessary elements are complete, specifically: for the three elements of the special function point, the special task function area, and the reference path, detect whether they exist, and if so, The three elements are considered complete; for tasks, it is detected whether the reference path corresponding to the operation task is connected. If it is connected, the task is considered complete; otherwise, the task is considered incomplete. 9. A rapid high-precision map construction device, comprising the following modules: The sensor data preprocessing module is used to obtain the raw measurement data of multiple preset sensors including GNSS installed on the unmanned vehicle in the outdoor scene, and then according to the raw measurement data of GNSS (ie Global Navigation Satellite System), Determine whether the GNSS signal of the outdoor scene is good, and if so, continue to perform the preset format processing operation on the preset useful data in the original measurement data of the preset multiple sensors on the unmanned vehicle; The vehicle trajectory and function information processing module is connected with the sensor data preprocessing module, and according to the preset useful data of the preset multiple sensors processed by the sensor data preprocessing module, generates the driving trajectory of the unmanned vehicle and marks the special function points or Special task functional area; The high-precision map processing module is connected to the vehicle trajectory and function information processing module, and is used to process the driving trajectory of the unmanned vehicle and the marked special function points or special task function areas obtained from the vehicle trajectory and function information processing module. , generate a high-precision map, and then send it to; The map integrity detection module is connected with the high-precision map processing module, and is used to perform integrity detection on the high-precision map generated by the high-precision map processing module, and finally obtain a complete high-precision map. 10. A vehicle, characterized by comprising the rapid high-precision map construction device according to claim 9.

Images