A method and device for learning a parking path

Abstract

The application discloses a path learning method and device for memory parking, wherein the method comprises the following steps: recording the starting time and the ending time of path learning for memory parking, and acquiring the path learning duration; collecting the real-time speed of the vehicle multiple times within the path learning duration, and calculating the vehicle learning speed; calculating the learning path length according to the path learning duration and the vehicle learning speed; calculating the mapping data volume according to the path learning duration and the vehicle camera parameters; comparing the learning path length and the mapping data volume with the respective preset threshold values respectively, and if both of them do not exceed the preset threshold values, executing the delayed power-off mapping strategy, otherwise executing the delayed power-off offline storage strategy. In the case that the path learning time is relatively long and the data volume is relatively large, the application adopts the delayed power-off storage strategy, which not only ensures the mapping efficiency of path learning, but also ensures that the storage battery is not discharged and the power-off data is not lost, without the need to improve the endurance of the storage battery, thereby saving the cost.

Description

Technical Field

[0001] This invention belongs to the field of intelligent connected vehicle technology, specifically relating to a path learning method and device for memory parking. Background Technology

[0002] Currently, assisted parking systems with memory parking (HPA) functionality are gradually being adopted in the market. The usage process of memory parking is as follows: path learning → initial scene detection → trajectory tracking. The system first extracts and maps the surrounding environment features based on the forward-looking visual perception system, and records and saves the vehicle's trajectory, providing a reference feature map for subsequent initial positioning. Finally, it tracks the trajectory recorded in the map to park the vehicle in the parking space. This achieves the goal of freeing the driver at the initial cruise and ultimately realizing automatic parking in fixed spaces, enhancing the user's intelligent experience.

[0003] Path learning is the foundation for this memory parking function. The learning process generates a large amount of data, and the longer the learning path, the larger the data volume and the longer the mapping time. During the path learning phase, the system builds a map based on the environment and saves the driving trajectory. After parking, the user locks the car and turns off the power. However, the large amount of data stored in RAM will be lost if the power is turned off, and mapping cannot be completed directly, causing path learning to fail. But if the driver is reminded to stay in the car and wait for the data to be stored before turning off the power and locking the car, it will affect the user experience and diminish the product's intelligence.

[0004] In current solutions, when the stored route is short (around 100m), the parking controller can be automatically de-energized after the path learning is complete and the car is locked, until the mapping is finished. However, as the learned parking path becomes longer, the mapping time required after saving the route will also increase. In gasoline vehicles, relying solely on battery power makes offline mapping after the user disconnects the power insufficient, leading to battery depletion. Continuously increasing battery range presents technical challenges and increases costs—a highly sensitive issue in the automotive industry—making it neither economical nor practical. Summary of the Invention

[0005] The technical problem to be solved by the embodiments of the present invention is to provide a path learning method and apparatus for memory parking, so as to improve the efficiency of path learning and offline mapping and effectively prevent vehicle battery drain.

[0006] To solve the above-mentioned technical problems, the present invention provides a path learning method for memory parking, comprising:

[0007] In response to the activation of the memory parking function, the start and end times of the memory parking path learning are recorded, and the path learning duration is obtained.

[0008] The vehicle's real-time speed is collected multiple times within the path learning time, and the vehicle's learning speed is calculated based on the collected real-time vehicle speeds.

[0009] The learning path length is calculated based on the path learning time and the vehicle learning speed.

[0010] Based on the path learning time and the vehicle camera parameters, the vehicle surrounding environment data collected by the vehicle camera for mapping is calculated to obtain the mapping data volume.

[0011] The learning path length is compared with a first preset threshold, and the mapping data volume is compared with a second preset threshold. If the learning path length does not exceed the first preset threshold and the mapping data volume does not exceed the second preset threshold, then a delayed power-off mapping strategy is executed; otherwise, a delayed power-off offline storage strategy is executed.

[0012] Furthermore, the delayed power-off offline storage strategy specifically involves: offline saving the mapping data stored in the random access memory (RAM) to the embedded multimedia controller (EMMC); and then, after the vehicle is powered on again, retrieving the mapping data from the EMMC to the RAM to continue offline mapping until the offline mapping target is completed.

[0013] Furthermore, the path learning duration is obtained by subtracting the recorded start time from the end time of the path learning, and the difference is taken as the path learning duration.

[0014] The vehicle learning speed is calculated by taking a weighted average of the real-time speeds of multiple vehicles.

[0015] Furthermore, the learning path length is calculated by multiplying the path learning time by the vehicle learning speed, and the product is taken as the learning path length.

[0016] Furthermore, the method for calculating the amount of mapping data is as follows: multiply the frame rate and pixels of the vehicle camera by the path learning time, and use the product as the amount of mapping data.

[0017] The present invention also provides a path learning device for memory parking, comprising:

[0018] The recording module is used to respond to the activation of the memory parking function by recording the start and end times of the path learning of memory parking and obtaining the path learning duration.

[0019] The first calculation module is used to collect the real-time speed of the vehicle multiple times within the path learning time, and calculate the vehicle learning speed based on the collected real-time speeds.

[0020] The second calculation module is used to calculate the length of the learning path based on the path learning time and the vehicle learning speed.

[0021] The third calculation module is used to calculate the amount of mapping data by the vehicle surrounding environment data collected by the vehicle camera for mapping based on the path learning time and the vehicle camera parameters.

[0022] The strategy selection module is used to compare the learning path length with a first preset threshold and the mapping data volume with a second preset threshold. If the learning path length does not exceed the first preset threshold and the mapping data volume does not exceed the second preset threshold, then the delayed power-off mapping strategy is executed; otherwise, the delayed power-off offline storage strategy is executed.

[0023] Furthermore, the delayed power-off offline storage strategy specifically involves: offline saving the mapping data stored in the random access memory (RAM) to the embedded multimedia controller (EMMC); and then, after the vehicle is powered on again, retrieving the mapping data from the EMMC to the RAM to continue offline mapping until the offline mapping target is completed.

[0024] Furthermore, the recording module obtains the path learning duration by subtracting the start time and end time of the recorded path learning, and using the difference as the path learning duration.

[0025] The first calculation module calculates the vehicle learning speed by taking a weighted average of the real-time speeds of multiple vehicles.

[0026] Furthermore, the second calculation module calculates the learning path length by multiplying the path learning time by the vehicle learning speed, and the product is taken as the learning path length.

[0027] Furthermore, the third calculation module calculates the amount of mapping data by multiplying the frame rate and pixels of the vehicle camera by the path learning time, and the product is used as the amount of mapping data.

[0028] Implementing this invention has the following beneficial effects: Based on existing hardware, by setting a judgment mechanism for the mapping strategy, when the learning route time is short and the data volume is small, a delayed power-off mapping scheme is adopted after the user powers off; when the path learning time is long and the data volume is large, a delayed power-off storage of learning data and mapping is adopted after power-on is adopted. This not only ensures the mapping efficiency of path learning, but also ensures that the battery is not depleted and the data is not lost when conditions do not permit, without the need to improve the battery's endurance, thus saving costs. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a flowchart illustrating a path learning method for memory parking according to an embodiment of the present invention. Detailed Implementation

[0031] The following description of the embodiments is taken with reference to the accompanying drawings, which illustrate specific embodiments in which the invention can be implemented.

[0032] Please refer to Figure 1 As shown, Embodiment 1 of the present invention provides a path learning method for memory parking, including:

[0033] In response to the activation of the memory parking function, the start and end times of the memory parking path learning are recorded, and the path learning duration is obtained.

[0034] The vehicle's real-time speed is collected multiple times within the path learning time, and the vehicle's learning speed is calculated based on the collected real-time vehicle speeds.

[0035] The learning path length is calculated based on the path learning time and the vehicle learning speed.

[0036] Based on the path learning time and the vehicle camera parameters, the vehicle surrounding environment data collected by the vehicle camera for mapping is calculated to obtain the mapping data volume.

[0037] The learning path length is compared with a first preset threshold, and the mapping data volume is compared with a second preset threshold. If the learning path length does not exceed the first preset threshold and the mapping data volume does not exceed the second preset threshold, then a delayed power-off mapping strategy is executed; otherwise, a delayed power-off offline storage strategy is executed.

[0038] Specifically, when the driver selects to activate the memory parking function, the vehicle first performs path learning, recording the start time T1 (in seconds) and the end time T2 (in seconds). The path learning duration T is then calculated as T = T2 - T1 (T in seconds). Simultaneously, the vehicle's real-time speed V is collected multiple times within the path learning duration T (e.g., every 10 seconds). i (Unit: meters per second), then the real-time speeds V of the i vehicles were collected. iThe vehicle learning speed V (unit: meters per second) was calculated using a weighted average method.

[0039] Based on the obtained path learning time T and vehicle learning speed V, the learning path length S = V × T (S unit: meters) is calculated. The learning path length S will serve as a key parameter affecting the mapping time in this embodiment.

[0040] Another key parameter affecting the mapping time is the amount of mapping data D (unit: megabytes), which is calculated based on the path learning time T and the parameters of the vehicle camera. Specifically, D = f × T × P, where f (unit: frames / second) is the frame rate of the vehicle camera and P is the number of pixels of the vehicle camera.

[0041] This embodiment innovatively sets up a mapping strategy determination mechanism based on the two key parameters affecting mapping processing time mentioned above. Using a single-variable method, the mapping data volume D and the learning path length S are compared with their respective preset thresholds. If neither exceeds the preset threshold, a delayed power-off mapping strategy can be adopted to complete path learning in one go, ensuring efficient path learning. If either key parameter exceeds the preset threshold, a delayed power-off offline storage strategy is adopted. Therefore, this invention can ensure that long-distance path learning data is not lost without affecting the efficiency of short-distance route learning, and prevents battery depletion due to excessively long offline mapping time.

[0042] Specifically, the learning path length S is compared with the first preset threshold S. N To make a comparison, the amount of mapping data D is compared with the second preset threshold D. N Make a comparison, if S≤S N And D≤D N If S > S, then a delayed power grid construction strategy is adopted; N and / or D > D N In this case, a delayed power-off offline storage strategy is adopted. It is understandable that if S > S... N and / or D > D N This indicates that at least one key parameter in the mapping data volume D and the learning path length S exceeds the corresponding preset threshold. Delaying power-down is insufficient to complete offline mapping, and the battery is at risk of depletion. Therefore, a delayed power-down offline storage strategy is adopted, where the mapping data stored in RAM (Random Access Memory) is offline saved to the EMMC (Embedded Multimedia Controller). After the vehicle is powered on again, the mapping data in the EMMC is retrieved to RAM to continue offline mapping until the offline mapping target is completed. The first preset threshold S N Second preset threshold D N It can be determined based on comprehensive testing of different vehicle models and different parking environments.

[0043] Corresponding to the memory parking path learning method described in Embodiment 1 of the present invention, Embodiment 2 of the present invention also provides a memory parking path learning device, comprising:

[0044] The recording module is used to respond to the activation of the memory parking function by recording the start and end times of the path learning of memory parking and obtaining the path learning duration.

[0045] The first calculation module is used to collect the real-time speed of the vehicle multiple times within the path learning time, and calculate the vehicle learning speed based on the collected real-time speeds.

[0046] The second calculation module is used to calculate the length of the learning path based on the path learning time and the vehicle learning speed.

[0047] The third calculation module is used to calculate the amount of mapping data by the vehicle surrounding environment data collected by the vehicle camera for mapping based on the path learning time and the vehicle camera parameters.

[0048] The strategy selection module is used to compare the learning path length with a first preset threshold and the mapping data volume with a second preset threshold. If the learning path length does not exceed the first preset threshold and the mapping data volume does not exceed the second preset threshold, then the delayed power-off mapping strategy is executed; otherwise, the delayed power-off offline storage strategy is executed.

[0049] Furthermore, the delayed power-off offline storage strategy specifically involves: offline saving the mapping data stored in the random access memory (RAM) to the embedded multimedia controller (EMMC); and then, after the vehicle is powered on again, retrieving the mapping data from the EMMC to the RAM to continue offline mapping until the offline mapping target is completed.

[0050] Furthermore, the recording module obtains the path learning duration by subtracting the start time and end time of the recorded path learning, and using the difference as the path learning duration.

[0051] The first calculation module calculates the vehicle learning speed by taking a weighted average of the real-time speeds of multiple vehicles.

[0052] Furthermore, the second calculation module calculates the learning path length by multiplying the path learning time by the vehicle learning speed, and the product is taken as the learning path length.

[0053] Furthermore, the third calculation module calculates the amount of mapping data by multiplying the frame rate and pixels of the vehicle camera by the path learning time, and the product is used as the amount of mapping data.

[0054] For the working principle and process of this embodiment, please refer to the description of Embodiment 1 of the present invention, which will not be repeated here.

[0055] As can be seen from the above description, compared with the prior art, the beneficial effects of the present invention are as follows: Based on the existing hardware foundation, by setting a judgment mechanism for the mapping strategy, when the learning route time is short and the data volume is small, a delayed power-off mapping scheme is adopted after the user powers off; when the path learning time is long and the data volume is large, a delayed power-off storage of learning data and mapping is adopted, and then the map is rebuilt after power-on is adopted. This not only ensures the mapping efficiency of path learning, but also ensures that the battery is not depleted and the data is not lost when conditions do not permit, without the need to improve the battery's endurance, thus saving costs.

[0056] The above description is merely a preferred embodiment of the present invention and should not be construed as limiting the scope of the invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A path learning method for memorizing parking routes, characterized in that, include: In response to the activation of the memory parking function, the start and end times of the memory parking path learning are recorded, and the path learning duration is obtained. The vehicle's real-time speed is collected multiple times within the path learning time, and the vehicle's learning speed is calculated based on the collected real-time vehicle speeds. The learning path length is calculated based on the path learning time and the vehicle learning speed. Based on the path learning time and the vehicle-mounted camera parameters, the vehicle's surrounding environment data collected by the vehicle-mounted camera for mapping is calculated to obtain the mapping data volume; the calculation method for the mapping data volume is as follows: D = f × T × P ,in, D It refers to the amount of data used for mapping. f It's the frame rate of the car camera. T It is the path learning time. P It refers to the pixels of the vehicle's camera; The learning path length is compared with a first preset threshold, and the mapping data volume is compared with a second preset threshold. If the learning path length does not exceed the first preset threshold and the mapping data volume does not exceed the second preset threshold, then a delayed power-off mapping strategy is executed; otherwise, a delayed power-off offline storage strategy is executed.

2. The path learning method for memory parking according to claim 1, characterized in that, The delayed power-off offline storage strategy is as follows: the mapping data stored in the random access memory (RAM) is saved offline to the embedded multimedia controller (EMMC). When the vehicle is powered on again, the mapping data in the EMMC is retrieved back to the RAM to continue offline mapping until the offline mapping target is completed.

3. The path learning method for memory parking according to claim 2, characterized in that, The path learning duration is obtained by subtracting the start time and end time of the recorded path learning, and the difference is taken as the path learning duration. The vehicle learning speed is calculated by taking a weighted average of the real-time speeds of multiple vehicles.

4. The path learning method for memory parking according to claim 3, characterized in that, The learning path length is calculated by multiplying the learning time of the path by the learning speed of the vehicle, and the product is taken as the learning path length.

5. The path learning method for memory parking according to claim 4, characterized in that, The method for calculating the amount of mapping data is as follows: multiply the frame rate and pixel count of the vehicle camera by the path learning time, and use the product as the amount of mapping data.

6. A path learning device for memory parking, characterized in that, include: The recording module is used to respond to the activation of the memory parking function by recording the start and end times of the path learning of memory parking and obtaining the path learning duration. The first calculation module is used to collect the real-time speed of the vehicle multiple times within the path learning time, and calculate the vehicle learning speed based on the collected real-time speeds. The second calculation module is used to calculate the length of the learning path based on the path learning time and the vehicle learning speed. The third calculation module is used to calculate the amount of mapping data by analyzing the vehicle's surrounding environment data collected by the vehicle camera for mapping, based on the path learning time and the vehicle camera parameters. The calculation method for the amount of mapping data is as follows: D = f × T × P ,in, D It refers to the amount of data used for mapping. f It's the frame rate of the car camera. T It is the path learning time. P It refers to the pixels of the vehicle's camera; The strategy selection module is used to compare the learning path length with a first preset threshold and the mapping data volume with a second preset threshold. If the learning path length does not exceed the first preset threshold and the mapping data volume does not exceed the second preset threshold, then the delayed power-off mapping strategy is executed; otherwise, the delayed power-off offline storage strategy is executed.

7. The memory parking path learning device according to claim 6, characterized in that, The delayed power-off offline storage strategy is as follows: the mapping data stored in the random access memory (RAM) is saved offline to the embedded multimedia controller (EMMC). When the vehicle is powered on again, the mapping data in the EMMC is retrieved back to the RAM to continue offline mapping until the offline mapping target is completed.

8. The path learning device for memory parking according to claim 7, characterized in that, The recording module obtains the path learning duration by subtracting the start time and end time of the recorded path learning, and using the difference as the path learning duration. The first calculation module calculates the vehicle learning speed by taking a weighted average of the real-time speeds of multiple vehicles.

9. The memory parking path learning device according to claim 8, characterized in that, The second calculation module calculates the learning path length by multiplying the path learning time by the vehicle learning speed, and the product is taken as the learning path length.

10. The path learning device for memory parking according to claim 9, characterized in that, The third calculation module calculates the amount of mapping data by multiplying the frame rate and pixels of the vehicle camera by the path learning time, and the product is used as the amount of mapping data.

Images