A vehicle lighting control method based on multi-sensor fusion

Abstract

This invention discloses a vehicle lighting control method based on multi-sensor fusion, comprising the following steps: S1, performing vehicle matching and positioning through an inertial navigation system to obtain vehicle self-positioning data; S2, collecting environmental information of the road ahead of the vehicle through a camera; and S3, modeling based on the environmental information of the road ahead of the vehicle, and then matching it with the vehicle self-positioning data to complete real-time vehicle positioning based on a high-precision map. This invention provides a vehicle lighting control method based on multi-sensor fusion, which uses a visual detection algorithm to perceive road information and combines it with positioning system data fusion to achieve real-time vehicle positioning based on a high-precision map. It also provides a self-learning algorithm model to establish a database of frequently used historical routes, effectively reducing the repetitive computation of the AI ​​chip and thus reducing energy consumption. Furthermore, it accurately controls vehicle lights on frequently used road sections, effectively reducing the overall energy consumption of the vehicle.

Description

Technical Field

[0001] This invention relates to a vehicle lighting control method based on multi-sensor fusion. Background Technology

[0002] Currently, as the automotive industry accelerates its integration towards electrification, intelligence, connectivity, and sharing, high-precision maps, which are closely related to these trends, are becoming increasingly important. The advanced self-positioning assessment and surrounding environment perception capabilities of high-precision maps are fundamental to the safe operation of autonomous vehicles, playing an indispensable role in perceiving the road traffic environment, just like sensors. This is why more and more companies are investing in this area, leading to a continuous increase in industry interest.

[0003] Meanwhile, current national standards clearly state that new electric vehicles must reduce energy consumption by more than 14% in the second phase. Although most L3+ intelligent vehicles currently achieve point-to-point autonomous driving in specific scenarios through high-precision maps, they haven't established a commonly used road database to reduce algorithmic computation and thus lower vehicle energy consumption. Furthermore, proper headlight control can effectively reduce overall vehicle energy consumption, and the aforementioned perception of the road environment is primarily used for vehicle control rather than for headlight control, thus failing to achieve data reuse to some extent. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art. The present invention provides a vehicle lighting control method based on multi-sensor fusion. By perceiving road information through a visual detection algorithm and combining it with data fusion from a positioning system, the method achieves real-time vehicle positioning based on a high-precision map. At the same time, it provides a self-learning algorithm model to establish a database of commonly used historical routes, so as to effectively reduce the amount of repetitive calculations of the AI ​​chip and reduce energy consumption. Meanwhile, it accurately controls vehicle lights on commonly used road sections, effectively reducing the energy consumption of the entire vehicle.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] A vehicle lighting control method based on multi-sensor fusion, comprising:

[0007] Step S1: Perform vehicle matching and positioning using an inertial navigation system to obtain vehicle self-positioning data;

[0008] Step S2: Collect environmental information of the road in front of the vehicle using a camera;

[0009] Step S3: Model the environment information of the road in front of the vehicle, and then match it with the vehicle self-localization data to complete the real-time vehicle localization based on the high-precision map.

[0010] Step S4: Obtain historical driving data from the user's vehicle's historical trips and real-time driving data from the user's current trip.

[0011] Step S5: Remove outlier data from the acquired historical driving data and real-time driving data to form a new training dataset;

[0012] Step S6: Perform deep learning based on the data to be processed in the new training dataset, and establish a database of commonly used historical routes based on the learning results;

[0013] Step S7: When the vehicle starts, the real-time location of the vehicle based on the high-precision map is used to determine whether the current real-time path is a path in the commonly used historical route database.

[0014] If the determination is yes, then directly call the vehicle lighting control strategy in the commonly used historical route database to perform the relevant operations;

[0015] If the judgment is negative, then based on the driver's needs and combined with environmental perception and decision control results, a new route is added and saved to the frequently used historical route database.

[0016] Furthermore, step S1 specifically includes the following steps:

[0017] The GNSS+IMU combined positioning method is used, with the GNSS sensor providing absolute positioning. Let the position of the GNSS sensor be (x... i y i , z i The satellite position is (x p y p , z p The positioning model based on the pseudo-distance Si is as follows:

[0018]

[0019] Among them, t p t represents the time of the GNSS sensor clock module. i The time is for the satellite clock module;

[0020] By receiving data from multiple positioning satellites, multiple sets of pseudo-distance observation equations are established. Then, least-squares linearized equations are constructed to solve for the positioning point coordinates, thereby obtaining vehicle self-positioning data.

[0021] Furthermore, the driving data includes vehicle speed, engine speed, distance to the vehicle in front, accelerator and brake pedals, steering wheel data, headlight status, seat belt status, airbag status, and driving environment information.

[0022] Furthermore, the driving environment information is detected by a forward-facing camera and a surround-view camera, and the driving environment information includes traffic light information, zebra crossing information, traffic congestion information, and information about vehicles ahead.

[0023] Furthermore, step S5 specifically includes the following steps:

[0024] Abnormal data is removed from the historical and real-time driving data, including driving data outside the designed operating domain, driving data in abnormal driving scenarios, and driving data in accident scenarios. The remaining normal driving data is then used to form a new training dataset.

[0025] Furthermore, step S6 specifically includes the following steps:

[0026] The system selects driving parameters to be learned from the data to be processed in the new training dataset, and calculates the data to be processed according to the calculation rules corresponding to the driving parameters to obtain the learning results of the driving parameters. Based on the learning results, the system instructs the autonomous driving system to set the driving parameters according to the learning results during the autonomous driving trip, and finally forms a database of commonly used historical routes.

[0027] Furthermore, the vehicle lighting control strategy includes a current time period headlight control strategy, which is as follows:

[0028] When there is sufficient natural light during the day, there is no need to turn on the headlights;

[0029] As the light intensity gradually changes between dawn and dusk, the light intensity changes inversely with the light intensity.

[0030] When driving at night, turn the headlights up to 100% intensity.

[0031] Furthermore, the vehicle lighting control strategy includes a current road condition headlight control strategy, which is as follows:

[0032] When the current road section is a well-lit urban road section, turn on the lights at 50% intensity;

[0033] When the current road section is a rural road with poor lighting conditions, the light intensity is set to 100%.

[0034] Furthermore, the vehicle lighting control strategy includes a current weather condition headlight control strategy, which is as follows:

[0035] When the weather is clear and visibility is high, the light intensity should be turned up to 70%.

[0036] In rain, fog, or extreme weather, the light intensity is turned up to 100%.

[0037] Furthermore, the vehicle lighting control strategy includes a current driving condition headlight control strategy, which is as follows:

[0038] Based on the current vehicle speed, determine whether there are vehicles traveling at a corresponding distance ahead. If there are vehicles traveling ahead, the current driving situation is to follow the vehicle in front. The headlight illumination distance is adjusted according to the actual distance to the vehicle.

[0039] By adopting the above technical solution, the present invention has the following beneficial effects:

[0040] 1. By establishing a database of frequently used historical routes, the amount of road perception calculations for frequently used road sections is greatly reduced, thereby reducing energy consumption caused by repeated calculations and playing a role in energy conservation and emission reduction.

[0041] 2. By establishing a database of frequently used historical routes, the lighting on commonly used road sections can be adjusted and controlled, minimizing energy waste caused by the misuse of lighting and thus playing a role in energy conservation and emission reduction.

[0042] 3. The multiple sensors described in this invention are standard configurations for intelligent vehicles, are easy to install, and are readily adopted.

[0043] 4. This invention employs a self-learning algorithm, which can optimize the algorithm based on the user's personalized driving data during driving to achieve performance parameters that conform to the user's driving habits. Attached Figure Description

[0044] Figure 1 This is a flowchart of the vehicle lighting control method based on multi-sensor fusion according to the present invention;

[0045] Figure 2 A flowchart illustrating the creation of the commonly used historical route database for this invention;

[0046] Figure 3 This is a flowchart of the data processing of the present invention;

[0047] Figure 4 This is a flowchart of the multi-sensor fusion process of the present invention;

[0048] Figure 5 This is an example diagram of the commonly used historical route database of the present invention. Detailed Implementation

[0049] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0050] like Figure 1 As shown, this embodiment provides a vehicle lighting control method based on multi-sensor fusion, which includes:

[0051] Step S1: Perform vehicle matching and positioning using an inertial navigation system to obtain vehicle self-positioning data. This includes the following steps:

[0052] Regarding vehicle self-localization, this embodiment uses a GNSS+IMU combined positioning method. The GNSS sensor provides absolute positioning. Let the position of the GNSS sensor be (x... i y i , z i The satellite position is (x p y p , z p The positioning model based on the pseudo-distance Si is as follows:

[0053]

[0054] Among them, t p t represents the time of the GNSS sensor clock module. i The time is for the satellite clock module;

[0055] By receiving data from multiple positioning satellites, multiple sets of pseudo-distance observation equations are established. Then, least-squares linearized equations are constructed to solve for the positioning point coordinates, thereby obtaining vehicle self-positioning data.

[0056] Step S2: Collect environmental information of the road ahead of the vehicle using a camera.

[0057] Step S3: Model the road environment information ahead of the vehicle, and then match it with the vehicle's self-localization data to complete real-time vehicle positioning based on a high-precision map. Due to the accumulation error problem in positioning technology, visual SLAM technology is also used in step S2 of this embodiment to achieve high-precision positioning. Visual SLAM technology involves using a camera to collect data about the road environment ahead, creating a model, and then matching it with the vehicle's self-localization data to complete high-precision vehicle positioning.

[0058] Step S4: Obtain historical driving data from the user's vehicle's historical trips and real-time driving data for the current trip. Driving data includes vehicle speed, engine speed, distance to the vehicle in front, accelerator, brake, and steering wheel data, headlight status, seatbelt status, airbag status, and driving environment information. Driving environment information is detected by the forward-facing camera and surround-view cameras, and includes traffic light information, pedestrian crossing information, traffic congestion information, and information about the vehicle in front.

[0059] Step S5: Remove outliers from the acquired historical driving data and real-time driving data to form a new training dataset. This includes the following steps:

[0060] like Figure 3 As shown, outlier data is removed from the acquired historical and real-time driving data, including driving data outside the designed operating domain, driving data from abnormal driving scenarios, and driving data from accident scenarios. The remaining normal driving data forms a new training dataset. Outlier data primarily refers to driving data outside the designed operating domain, driving data from abnormal driving scenarios, and driving data from accident scenarios. Because this outlier data does not meet safe driving requirements or traffic rules, it is necessary to further remove outlier data from the driving data to ensure that the system parameters after self-learning by the autonomous driving system can guarantee driving safety. This embodiment uses a real-time outlier removal method to reduce data storage volume and the occupation of system storage space.

[0061] Step S6: Perform deep learning based on the data to be processed in the new training dataset, and establish a database of commonly used historical routes based on the learning results. This includes the following steps:

[0062] The system selects driving parameters to be learned from the data to be processed in the new training dataset, and calculates the data to be processed according to the calculation rules corresponding to the driving parameters to obtain the learning results of the driving parameters. Based on the learning results, the system instructs the autonomous driving system to set the driving parameters according to the learning results during the autonomous driving trip, and finally forms a database of commonly used historical routes.

[0063] Step S7, as follows Figure 1 As shown, when the vehicle starts, it is judged based on the real-time positioning of the vehicle based on a high-precision map to determine whether the current real-time path is a path in the frequently used historical route database.

[0064] If the determination is yes, the relevant operation is executed by directly calling the vehicle lighting control strategy in the commonly used historical route database. This can avoid the repeated acquisition and calculation of information by various sensors and controllers, thereby further reducing system power consumption.

[0065] If the judgment is negative, then based on the driver's needs and combined with environmental perception and decision control results, a new route is added and saved to the frequently used historical route database.

[0066] The following analysis will take the addition of a new route as an example:

[0067] After the vehicle starts moving, it obtains the current vehicle location based on GPS signals and high-precision maps, and determines whether the current route is a frequently used route. If not, a pop-up window appears in the human-machine interface, asking the driver whether to record the trip data. If the driver selects yes, a new route data file is created, recording the driver's operation and vehicle lighting control strategy into the current route data for direct retrieval next time.

[0068] The automotive lighting control strategy in this embodiment includes:

[0069] Set the light intensity of the vehicle's headlights to 100% when all LED lights are on.

[0070] 1. Current time period headlight control strategy:

[0071] When there is sufficient natural light during the day, there is no need to turn on the headlights;

[0072] As the light intensity gradually changes between dawn and dusk, the light intensity changes inversely with the light intensity.

[0073] When driving at night, the headlight intensity is turned on to 100%, and the specific time period is adaptively adjusted based on the measurement results of the light intensity sensor.

[0074] 2. Current road conditions and headlight control strategy:

[0075] When the current road section is a well-lit urban road section, turn on the lights at 50% intensity;

[0076] When the current road section is a rural road section with poor lighting conditions, the light intensity is set to 100%, and the specific setting is adaptively adjusted based on the environmental perception results and the light intensity sensor measurement results.

[0077] 3. Current weather conditions and vehicle light control strategy:

[0078] When the weather is clear and visibility is high, the light intensity should be turned up to 70%.

[0079] In rain, fog, or extreme weather, the light intensity is turned up to 100%.

[0080] 4. Current driving situation headlight control strategy:

[0081] Based on the current vehicle speed, determine whether there are vehicles traveling at a corresponding distance ahead. If there are vehicles traveling ahead, the current driving situation is to follow the vehicle in front. The headlight illumination distance is adjusted according to the actual distance to the vehicle.

[0082] The following section analyzes the distance-based control strategies for automotive lighting:

[0083] like Figure 5 The image shows the route data for a car traveling from the east gate of Xingyu Center to Gate 2 of Xincheng Gymnasium at 8:30 PM on a clear summer night. The data is generated through GNSS positioning, a perception system identifying lane markings and following distance, and a light intensity sensor providing feedback on streetlight illumination conditions. The data is then comprehensively analyzed to output the light intensity. Figure 5It can be seen that the street lighting conditions in sections 3, 5, and 6 are good, so the number of lights turned on is 50% regardless of whether there is a vehicle following. However, in section 4, the number of lights turned on is about 90% when the lighting conditions are poor.

[0084] In addition, the vehicle in this embodiment includes an autonomous driving domain controller, a communication module, a central control screen, switches, and other systems on the vehicle side. The autonomous driving domain controller is a control module that implements the vehicle's driving assistance or autonomous driving system functions; the communication module is used to enable communication between the vehicle and the cloud; the central control screen is a human-machine interface; and the switches are physical or analog switches on the vehicle side used to activate the self-learning function of the autonomous driving system. Figure 4 As shown, in the technical solution of this embodiment, the vehicle's sensors and modules first collect user driving data, upload it from the vehicle to the cloud server, complete model optimization learning and training on the server, and return the updated performance parameters to the vehicle to control the vehicle to realize autonomous driving-related functions in order to optimize the user experience.

[0085] The specific embodiments described above further illustrate the technical problems, technical solutions, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-sensor fusion based vehicle light control method, characterized by, It comprises: Step S1, vehicle matching positioning is performed through an inertial navigation system to obtain vehicle self-positioning data; Step S2, the environment of the road in front of the vehicle is collected through a camera to obtain environment information of the road in front of the vehicle; Step S3, modeling is performed according to the environment information of the road in front of the vehicle, and then the vehicle self-positioning data is matched to complete real-time positioning of the vehicle based on a high-precision map; Step S4, historical driving data in a historical trip of the user driving the vehicle and real-time driving data in a current trip of the user driving the vehicle are obtained; Step S5, the obtained historical driving data and real-time driving data are subjected to abnormal data elimination to form a new training data set; Step S6, deep learning is performed on the data to be processed in the new training data set, and a commonly used historical route database is established through the learning result; Step S7, when the vehicle starts, it is judged according to the real-time positioning of the vehicle based on the high-precision map whether the current real-time path is a path in the commonly used historical route database; If yes, the vehicle lighting control strategy in the commonly used historical route database is directly called to perform related operations; If no, a new route is added to the commonly used historical route database according to the needs of the driver combined with the environment perception and decision control result; The step S1 specifically comprises the following steps: The GNSS+IMU combined positioning mode is adopted, the GNSS sensor provides absolute positioning, the position of the GNSS sensor is (x i , y i , z i ), the positioning satellite position is (x p , y p , z p ), and a positioning model formed by pseudo distances S1 is as follows: wherein t p is the time of the GNSS sensor clock module, t i is the time of the satellite clock module; A plurality of positioning satellite data information is received to establish a plurality of pseudo-range observation equation groups, and then a least square linearization equation is constructed to solve the positioning point coordinates to obtain the vehicle self-positioning data; The vehicle lighting control strategy comprises a current time period vehicle light control strategy, which is: When the natural light is sufficient during the day, the high beam is not needed to be turned on; When the light intensity gradually changes during the morning and evening, the light intensity changes inversely with the light intensity; When driving at night, the light intensity is turned on to 100%; The vehicle lighting control strategy comprises a current road condition vehicle light control strategy, which is: When the current road section is a city road section with good lighting conditions, the light intensity is turned on to 50%; When the current road section is a rural road section with poor lighting conditions, the light intensity is set to 100%; The vehicle lighting control strategy comprises a current weather condition vehicle light control strategy, which is: When the weather is sunny and the visibility is high, the light intensity is turned on to 70%; When it is rainy, foggy or extreme weather, the light intensity is turned on to 100%. 2.The multi-sensor fusion based vehicle lamp control method according to claim 1, characterized in that: The driving data comprises vehicle driving speed, engine speed, front vehicle distance, throttle brake and steering wheel data, vehicle light state, safety belt state, airbag state and driving environment information. 3.The multi-sensor fusion based vehicle lamp control method of claim 1, wherein: The driving environment information is detected through a front camera and a surround camera, and the driving environment information comprises road section traffic light information, road section zebra crossing information, road section congestion information and front vehicle information. 4.The multi-sensor fusion based vehicle lamp control method according to claim 1, wherein, The step S5 specifically comprises the following steps: The abnormal data elimination is performed on the driving data of the non-design operation domain, the driving data of the non-normal driving scene, and the driving data of the accident scene in the obtained historical driving data and real-time driving data, and the remaining normal driving data is formed into a new training data set. 5.The multi-sensor fusion based vehicle lamp control method according to claim 1, wherein, The step S6 specifically comprises the following steps: According to the to-be-processed data in the new training data set, a to-be-learned driving parameter is selected, the to-be-processed data is calculated according to a calculation rule corresponding to the driving parameter, a learning result of the driving parameter is obtained, the automatic driving system is instructed to set the driving parameter according to the learning result in an automatic driving trip according to the learning result, and finally a commonly used historical route database is formed.

6. The multi-sensor fusion based vehicle light control method of claim 1, wherein: The automobile lighting control strategy comprises a current driving condition vehicle light control strategy, and the current driving condition vehicle light control strategy is: According to the current vehicle speed, it is judged whether a vehicle is driving within a corresponding distance in front, if a vehicle is driving, the current driving condition is a following driving condition, and the light illumination distance of the headlamp is adjusted according to the actual vehicle distance.

Images