A vehicle lighting control method and system

Abstract

This application relates to the field of lighting control technology, and more particularly to a vehicle lighting control method and system. The proposed solution involves acquiring dynamic information from multiple target objects to determine their motion trends and behavioral intentions, and mapping these intentions to a direction-sensitive objective function. Based on the motion trends, orientation alignment and spatial scaling are performed within a local reference coordinate system to construct a behavioral potential field model. Multiple behavioral potential field models are spatially superimposed to generate a spatially continuous illuminance weight map, which is then mapped to the lighting group to achieve point-to-point brightness adjustment. This application improves the accuracy of lighting response in complex traffic scenarios, avoids lighting interference, and ensures driving safety.

Description

Technical Field

[0001] This application relates to the field of lighting control technology, and in particular to a vehicle lighting control method and system. Background Technology

[0002] In existing technologies, vehicle lighting control is mostly based on target detection results to open or close the lighting area as a whole, block certain areas, or switch between high and low beams. Shading strategies are often implemented using rule thresholds, fixed angle control, or by using cameras to identify oncoming vehicles and pedestrians. However, in real-world road environments, traffic participants are diverse and their behaviors are complex, especially at night or in low visibility conditions. Problems such as dense target groups, intersecting behavioral paths, and unstable movement exist. Traditional control methods based on static position or simple behavioral judgments struggle to accurately match actual risk areas, easily leading to delayed illumination adjustment, misjudgments, or excessive shading, causing safety hazards such as lost vision and glare.

[0003] For example, Chinese patent CN111746381B discloses a vehicle lighting control system and a vehicle lighting control method. This vehicle lighting control system includes a lamp assembly, a lamp assembly controller, a smart camera, and a body controller. The smart camera is communicatively connected to both the body controller and the lamp assembly controller, and the lamp assembly controller is connected to the lamp assembly. The lamp assembly includes low beams and a predetermined number of high beams, with each high beam arranged in a stepped manner at a predetermined position at the front of the vehicle, such that the optical axis of each high beam forms a predetermined angle with the vehicle body. The body controller outputs vehicle information to the smart camera. The smart camera is positioned at a predetermined location at the front of the vehicle and is used to collect image information of the current lane and adjacent lanes in the vehicle's direction of travel, determine road information based on the image information, and output a lighting switching signal to the lamp assembly controller based on the road information. The lamp assembly controller switches the on / off state of the high beams in the lamp assembly in response to the lighting switching signal.

[0004] All of the above patents suffer from the problem described in the background: it is difficult to achieve finer-grained and more sensitive lighting response control in multi-objective dynamic environments, which limits the further improvement of lighting systems in intelligent driving. To solve the above problems, this application designs a vehicle lighting control method and system. Summary of the Invention

[0005] The technical problem this application aims to solve is to address the shortcomings of existing technologies by providing a vehicle lighting control method and system. This method involves acquiring dynamic information from multiple target objects, determining their motion trends and behavioral intentions, and mapping these intentions to a direction-sensitive objective function. Based on the motion trends, the method performs direction alignment and spatial scaling within a local reference coordinate system to construct a behavioral potential field model. Multiple behavioral potential field models are spatially superimposed to generate a spatially continuous illuminance weight map, which is then mapped to the lighting group to achieve point-to-point brightness adjustment.

[0006] To achieve the above objectives, this application provides the following technical solution:

[0007] A vehicle lighting control method is applied to a vehicle lighting assembly, the vehicle lighting assembly being disposed on the front structure of the vehicle and having its illumination direction covering the main lane and adjacent lane areas in front of the vehicle, the method comprising:

[0008] Acquire dynamic information of multiple target objects in the environment in front of the vehicle;

[0009] A behavioral potential field model is constructed based on the dynamic information, and the behavioral potential field model is spatially superimposed to generate an illuminance weight map of the vehicle's forward lighting area, wherein the illuminance weight map is a two-dimensional distribution map with spatial continuity.

[0010] The illuminance weight map is mapped to the vehicle headlight assembly so that the vehicle headlight assembly can perform point-by-point brightness adjustment.

[0011] Constructing a behavioral potential field model based on the dynamic information includes:

[0012] Based on the dynamic information, the movement trend and behavioral intent of the target object are obtained;

[0013] The behavioral intent is mapped to the target function type in a preset set of potential field constructors, wherein the potential field constructors are spatially distributed according to direction sensitivity, decay gradient and radius of action;

[0014] Based on the motion trend, the corresponding objective function types are oriented and spatially scaled within the vehicle's forward lighting area, and then superimposed to generate a behavioral potential field model.

[0015] Based on the aforementioned motion trend, the corresponding objective function type is directionally aligned and spatially scaled within the vehicle's forward lighting area, including:

[0016] Establish a local reference coordinate system with the current position of the target object as the origin and the direction corresponding to the movement trend as the main axis;

[0017] Based on the stability of the motion trend, the local reference coordinate system is divided into multiple unequal-distance lighting control sub-segments, and scaling scales and shape modulation parameters are set for the target function type in different sub-segments to make the function present a non-uniform distribution in the principal axis direction.

[0018] The superposition process to generate a behavioral potential field model includes:

[0019] The local deformation results of the objective function type in multiple unequal-distance lighting control sub-segments are connected and combined according to their spatial order in the local reference coordinate system to form an overall function structure spliced ​​together from multiple unequal-scale function segments.

[0020] The overall function structure is mapped to the vehicle's forward lighting area to construct a behavioral potential field model of the target object.

[0021] The scaling scale and shape modulation parameters are adjusted based on local features of sub-segments within a local reference coordinate system, including illuminance attenuation rate, coverage radius, boundary ambiguity coefficient, and gradient directionality coefficient.

[0022] Based on the dynamic information, the movement trend and behavioral intent of the target object are obtained, including:

[0023] By extracting and analyzing the dynamic information of the target object within a continuous time window, the trajectory direction, acceleration rate of change, and heading angle change trends are obtained to determine the motion trend;

[0024] By combining the relative position of the target object, the intersection relationship between its trajectory and the vehicle's path, and the continuous characteristics of its movement trend, the behavioral intent is determined based on preset behavior pattern matching rules.

[0025] The behavioral potential field models are spatially superimposed to generate an illuminance weight map of the vehicle's forward lighting area, including:

[0026] Select the overlapping region of multiple behavioral potential field models;

[0027] Within the overlapping area, the dominance of the overlapping area is determined based on the risk priority weight of each model, and the area is assigned a value according to the illuminance suppression value of the dominant model. At the same time, the boundary areas of other models within the overlapping area are subjected to gradual transition processing.

[0028] The non-overlapping areas and the processed overlapping areas are combined according to their spatial location, and the boundary is fused by a continuous interpolation algorithm on the area edges to generate an illuminance weight map of the vehicle's forward lighting area.

[0029] Mapping the illuminance weight map to the vehicle headlight assembly to enable point-by-point brightness adjustment of the vehicle headlight assembly includes:

[0030] The illuminance weight map is mapped to the spatial arrangement of the light-emitting units in the vehicle light group according to its two-dimensional coordinates. The mapping relationship is a one-to-one correspondence or regional correspondence between each pixel in the illuminance weight map and the corresponding light-emitting unit in the vehicle light group.

[0031] The weight values ​​of each pixel in the illuminance weight map are converted into the target brightness values ​​of the corresponding light-emitting units, and the target brightness values ​​are normalized in combination with the current ambient light level.

[0032] The normalized brightness value is sent to the drive control circuit of each light-emitting unit.

[0033] The vehicle lighting assembly further includes position lights, front fog lights, rear fog lights, and taillights, and the method further includes:

[0034] Based on the vehicle's current operating status, external environment information, and the target object's behavioral potential field model, the on / off status, brightness level, and control timing of the vehicle's lighting group are jointly scheduled.

[0035] A vehicle lighting control system, the system comprising:

[0036] The target recognition module is used to acquire dynamic information of multiple target objects in the environment in front of the vehicle;

[0037] The behavior analysis module is used to obtain the movement trends and behavioral intentions of each target object based on the dynamic information.

[0038] The potential field construction module is used to map behavioral intentions to target function types in a preset potential field constructor set, and to perform directional alignment and spatial scaling of the target function types according to the motion trend to generate a behavioral potential field model of the target object.

[0039] The potential field combination module is used to map the behavioral potential field models of multiple target objects to a unified coordinate system of the vehicle's forward lighting area, and to perform regional assignment and boundary transition processing based on the risk priority weights of each model in the overlapping area to generate a two-dimensional illuminance weight map with spatial continuity.

[0040] The lamp group control module is used to map the illuminance weight map to each light-emitting unit in the vehicle lamp group and control the light-emitting unit to perform point-by-point brightness adjustment.

[0041] The potential field construction module includes:

[0042] The function selection unit is used to map the behavioral intent of the target object to the target function type in the preset potential field constructor set. The target function type has spatial distribution characteristics of direction sensitivity, decay gradient and radius of action.

[0043] The coordinate alignment unit is used to establish a local reference coordinate system in the vehicle's forward lighting area with the current position as the origin and the direction of movement as the main axis, based on the movement trend of the target object.

[0044] A partitioned modulation unit is used to divide the local reference coordinate system into multiple unequal-distance lighting control sub-segments, and to set the scaling scale and shape modulation parameters for the objective function type in different sub-segments respectively;

[0045] The function splicing unit is used to splice and combine the function fragments after sub-segment modulation in spatial order to generate an overall function structure, and then map the structure onto the vehicle's forward lighting area to form a behavioral potential field model of the corresponding target object.

[0046] Compared with the prior art, the beneficial effects of this application are:

[0047] 1. The vehicle lighting control method provided in this application achieves dynamic behavioral response control of multiple targets in complex traffic environments by introducing a behavioral potential field modeling mechanism. Compared with traditional lighting adjustment methods based on target position or simple detection results, this application can not only construct a highly directional and structurally adaptive illuminance suppression region according to the movement trend and behavioral intention of each target, but also realize the conflict fusion and continuous transition of multi-target behavioral potential fields in space, generating a high-resolution, behavior-driven illuminance weight map. Finally, through pixel-level brightness adjustment, it achieves precise avoidance of high-risk targets and ensures lighting in low-interference areas, effectively improving the intelligence level, response accuracy, and traffic safety of vehicle forward lighting. Attached Figure Description

[0048] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0049] Figure 1 This is a schematic diagram illustrating an exemplary application scenario provided in an embodiment of this application;

[0050] Figure 2 A schematic flowchart illustrating a vehicle lighting control method provided in an embodiment of this application;

[0051] Figure 3 A flowchart illustrating the method for constructing a behavioral potential field model provided in this application embodiment;

[0052] Figure 4 A schematic diagram illustrating the principle of function deformation provided in the embodiments of this application;

[0053] Figure 5 A flowchart illustrating a method for processing an objective function provided in an embodiment of this application;

[0054] Figure 6 This is a flowchart illustrating another objective function processing method provided in an embodiment of this application. Detailed Implementation

[0055] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0056] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0057] Example 1

[0058] Please see Figure 1 This figure is a schematic diagram of an exemplary application scenario provided by an embodiment of this application.

[0059] like Figure 1 As shown, the application scenario includes the object to be controlled, a road scene, and multiple target objects. The front of the vehicle is equipped with adjustable lights and a perception module for target perception. Preferably, the vehicle also includes a processing module for outputting control commands based on the information from the perception module. The processing module can be located at the vehicle terminal.

[0060] In one example, the perception module may include sensors such as cameras, millimeter-wave radar, lidar, IMU, and GPS, for acquiring dynamic target object information on the road ahead in real time.

[0061] Figure 1 The image shows a traffic participant riding an electric bike and a pedestrian crossing the road in the lane where the target object is located. In the adjacent opposite lane, there is also an oncoming vehicle and a pedestrian, located at the far end and intersection of the current lane's illuminated area, respectively. These target objects differ in type, direction of movement, and intentions, constituting a set of traffic participants with multiple types, directions, and behavioral states. The sensor module acquires dynamic information such as the movement trajectory, speed, and acceleration of these targets, and combines this information with the vehicle's status (e.g., speed, steering wheel angle) to determine movement trends and behavioral intentions.

[0062] Figure 1The illumination range is shown to cover the main lane and adjacent lanes, with the dashed sector representing the illumination range. When facing different types of targets, the system adjusts the light output based on the suppression strength and spatial location of the behavioral potential field model. The illuminance weight map is mapped to the headlight pixel array to achieve point-by-point brightness output control in different areas, ensuring optimal compatibility between safety and driving visibility. For example:

[0063] Create a light-suppressing zone for electric vehicles and pedestrians to avoid glare;

[0064] Increase the occlusion priority for pedestrians with a high probability of path intersection;

[0065] For oncoming vehicles that are far from the path or have a stable speed, only edge softening is applied;

[0066] Maintain normal illumination output in non-threat areas to ensure the vehicle's visibility.

[0067] In one example, a pedestrian in the oncoming lane is about to cross the road, and is positioned close to the edge of the current lane, posing a high risk of entering the lane. At the same time, an oncoming vehicle is approaching. Due to the close distance and low angle of the vehicle, if the oncoming vehicle's headlights are still shining normally at high beams, it may cause severe glare to the driver, affecting the driver's judgment of the pedestrian crossing and braking reaction, thus inducing a potential safety hazard.

[0068] For example, in another area, an electric vehicle in the current lane is moving in the same direction as the vehicle. Its speed is low and its path is stable, but its position is exactly in the center of the headlight illumination area. If no adjustment is made, it can easily interfere with the driver, especially in low visibility conditions at night, causing visual pressure or even obstructing the driver's vision.

[0069] Next, with reference to the accompanying drawings, a vehicle lighting control method provided by an embodiment of this application will be described.

[0070] Please see Figure 2 The figure is a schematic flowchart of a vehicle lighting control method provided in an embodiment of this application. Figure 2 The method shown can be applied to vehicle headlights, which are mounted on the front structure of a vehicle and illuminate the main lane and adjacent lanes in front of the vehicle. Figure 2 The method shown includes the following S1-S4:

[0071] S1: Obtain dynamic information of multiple target objects in the environment in front of the vehicle;

[0072] In this embodiment, an information acquisition device is installed at the front of the vehicle, including multiple sensing components such as an image sensor, millimeter-wave radar, lidar, and inertial measurement unit. This device is used to collect dynamic target object information in the environment ahead of the vehicle in real time during vehicle operation. Target objects include various traffic participants such as pedestrians, non-motorized vehicles, and oncoming vehicles in the current or oncoming lanes. The information acquisition device extracts dynamic parameters such as the trajectory, speed, acceleration, and heading angle of the target objects by fusing image and radar data within a continuous time window, providing a basic information source for subsequent behavior trend analysis and lighting control. In this way, accurate identification and status perception of multiple types of targets in complex road environments can be achieved, improving the reliability and accuracy of lighting control decisions.

[0073] S2: Construct a behavioral potential field model based on the dynamic information;

[0074] In this embodiment, for each target object, a set of function templates with attributes of direction sensitivity, attenuation gradient, and spatial radius of action are invoked for modeling based on its movement trend and behavioral intent. Direction alignment and non-uniform scaling within sub-segments are then performed on the target function in a local reference coordinate system. By dynamically setting the illuminance attenuation rate, coverage area, and boundary ambiguity parameters within the local sub-segments, a fitting description of the target's behavioral path is achieved, constructing a behavioral potential field model with an asymmetric spatial structure. This model effectively reflects the risk distribution characteristics of different traffic targets in different directions and areas, thus providing a more differentiated and semantically recognizable control basis for subsequent illumination adjustment.

[0075] S3: Spatially superimpose the behavioral potential field model to generate an illuminance weight map of the vehicle's forward lighting area;

[0076] In this embodiment, the behavioral potential field models of all target objects are mapped to a unified vehicle forward lighting coordinate system. For spatial regions where potential field models overlap, risk priority weights are assigned based on the risk level represented by the behavioral intent, and illuminance coverage is performed according to the suppression value of the dominant weight model. Simultaneously, a gradual transition is applied to the boundary regions of secondary models. Non-overlapping regions retain their original structure, and boundary interpolation fusion is performed between all regions, ultimately generating a two-dimensional illuminance weight map that is spatially continuous and semantically differentiated in terms of behavioral control. This processing method enables concurrent responses to multiple targets and multiple risk behaviors in illumination control, avoiding visual conflicts and accidental lighting, and improving overall driving safety.

[0077] S4: Map the illuminance weight map to the vehicle headlight assembly so that the vehicle headlight assembly can perform point-by-point brightness adjustment;

[0078] In this embodiment, the illuminance weight map, through spatial coordinate mapping logic, corresponds to the physical arrangement of the light-emitting units in the vehicle's front headlight assembly, establishing a one-to-one or regional mapping relationship between each pixel in the illuminance map and the corresponding light-emitting unit within the headlight assembly. The weight value of each pixel in the illuminance map is converted into a target brightness output value and normalized in conjunction with the current ambient light level. Finally, it is sent to the light-emitting unit control and drive module according to a set refresh cycle, realizing dynamic adjustment of brightness in different spatial areas. This method can control the target object at the pixel level, precisely avoid and flexibly illuminate the area around the target object, while preserving the key illuminance area required for the vehicle's passage, optimizing visibility and safety while driving.

[0079] In complex traffic environments, especially at night or in low visibility conditions, vehicles often face multiple dynamic targets of different types, such as pedestrians, electric vehicles, non-motorized vehicles crossing the road, and oncoming vehicles. These targets often have different behavioral intentions, speed characteristics, and path prediction trends, and their behavior is highly uncertain. They are also distributed in different spatial areas, with the possibility of occlusion, overlap, or even intersection between them. In such scenarios, traditional lighting control methods can only adjust brightness based on single target or local environmental information, generally lacking behavioral recognition capabilities and regionally differentiated control methods. This makes it difficult to ensure the vehicle's lighting needs while also considering the visual safety and behavioral interference control of other road users, easily leading to problems such as "lighting accidental injury," glare interference, or waste of lighting resources.

[0080] To address this issue, the vehicle lighting control method proposed in this embodiment introduces a behavioral potential field model construction mechanism to represent the illuminance influence requirement area of ​​a single target object in space. Furthermore, by establishing a local reference coordinate system, behavioral trends are mapped to asymmetric structure functions, and differentiated scaling and modulation strategies are applied in multiple sub-segments to accurately generate an illuminance suppression structure reflecting the directionality of behavioral intent. The behavioral potential field models of multiple targets are uniformly mapped to the lighting control coordinate system. When spatially superimposed, risk priority weight control and boundary continuity fusion strategies are introduced, effectively solving the regional dominance problem in multi-potential field conflicts. The final generated two-dimensional illuminance weight map not only possesses spatial continuity but also target semantic responsiveness, enabling the lighting group to achieve point-by-point brightness adjustment.

[0081] Taking a typical urban intersection at night as an example, there is an electric scooter driver about to change lanes in the main lane ahead of the vehicle, an oncoming vehicle in the opposite lane, and a pedestrian quickly crossing the road at the intersection. Traditional methods usually employ a uniform brightness reduction or blind spot darkening strategy for the illuminated area, failing to distinguish which areas require priority avoidance and which areas need to be kept illuminated, easily leading to an overall decrease in illuminance or neglect of key areas. However, the technology in this application, through multi-target behavior recognition and risk analysis, generates a dynamically extended potential field for pedestrians, constructing a wide-range suppression area in their forward direction; it applies a directional light suppression structure to oncoming vehicles, controlling the brightness of the light clusters only within their forward line of sight; and it adjusts the illumination gradient for lane-changing electric scooters based on their trajectory stability, while ensuring that the visibility of the vehicle's main lane edge is not weakened. Through this comprehensive light control output involving multiple behaviors, strategies, and structures, it achieves the dual objectives of precise obstacle avoidance and visual visibility protection, which are difficult to achieve simultaneously with traditional solutions.

[0082] In this embodiment, dynamic information includes, but is not limited to, parameters such as the target object's position coordinates, velocity, acceleration, heading angle, trajectory curvature, and relative distance and angular relationship with the vehicle over a continuous time period. This information can be acquired collaboratively by multiple sensors, such as cameras, millimeter-wave radar, lidar, and inertial measurement units, and extracted and fused through time-series modeling to reflect the continuous motion state of the target object. This dynamic information provides a foundation for judging the target object's motion trend. Combined with its relative intersection with the vehicle's current path, the target's velocity change trend, and motion stability characteristics, it can be used to infer its possible behavioral intentions, such as crossing, changing lanes, moving to the side of the road, or going straight. Through the complete acquisition and processing of this dynamic information, accurate prediction of the behavior of multiple target objects in complex traffic scenarios and basic determination of lighting response can be achieved.

[0083] It is important to note that the selection of dynamic information can be based on the specific vehicle application scenario, sensing device configuration, and processing capabilities, and is not limited to a specific type or quantity. This application does not limit the specific structure of the dynamic information; any parameter that can be used in the prior art to characterize the motion state and behavioral trends of a target object can be used as a component of the dynamic information. For example, but not limited to, data sources such as velocity vectors, trajectory point sequences, path prediction results, and lane change signals can all be used as the basis for determining motion trends and behavioral intentions in this embodiment. Therefore, the specific form of the dynamic information can be flexibly adapted to different sensing schemes and algorithm architectures.

[0084] Please see Figure 3 The figure is a flowchart illustrating the behavioral potential field model construction method provided in this embodiment of the application. Figure 3 The method shown can be applied to step S2 of the aforementioned method, and the specific steps are as follows:

[0085] S2.1: Obtain the movement trend and behavioral intention of the target object based on the dynamic information;

[0086] Specifically, complex road environments contain various types of dynamic targets with highly uncertain behavior. For example, pedestrians may cross, stop, or turn back; electric vehicles may change lanes, run alongside, or pull over; and oncoming vehicles may suddenly change lanes and enter the lane. If the current trends and future behavioral intentions of these targets cannot be accurately identified, the lighting control system cannot respond appropriately, potentially leading to problems such as glare, wasted illumination resources, or insufficient lighting in key areas. Therefore, it is necessary to extract the motion trend characteristics of each target in the early stages through continuous dynamic information analysis, and further combine this with the vehicle's own path, target behavior patterns, and intersection relationships to infer its short-term behavioral intentions.

[0087] In this embodiment, the acquisition of motion trends is based on the extraction and analysis of multi-dimensional dynamic information of the target object within a continuous time window, including but not limited to position sequence, speed changes, acceleration direction, and heading angle deviation. By performing time-series fitting and statistical evaluation on these data, the target's current main direction of movement, speed stability, and path fluctuations can be identified, determining whether it is in a stable straight-line state, accelerating into the lane, deviating from its course, or stopped. This trend judgment result provides structural parameters such as the principal axis of direction and the radius of extension when constructing the behavioral potential field model. Next, based on the target's position relative to the vehicle, the spatial intersection relationship between its trajectory and the vehicle's predicted path (whether it will enter the lane or form an intersection point), and the persistence of the motion trend (whether the trend changes drastically in the short term), a preset behavior pattern matching rule library is invoked to determine the target's behavioral intent. Behavioral intent types may include, but are not limited to, crossing, changing lanes, merging, moving away, moving to the side, and remaining stationary. Matching methods can be implemented using trajectory template recognition, rule tree filtering, or lightweight classifier modeling.

[0088] Furthermore, after determining the target's behavioral intent, a risk level label is assigned to each intent type, serving as an important control parameter when constructing the behavioral potential field function. For example, for crossing or merging behaviors, a forward-expanding illuminance suppression function is generated with a high illuminance reduction weight; while for targets relatively far from the path or with stable trajectories, a marginal weak suppression function is constructed to ensure that the illuminance in the vehicle's main driving area is not disturbed. Through this trend recognition and intent judgment process, subsequent lighting control becomes more behaviorally sensitive and spatially structure-matched, not only improving lighting efficiency but also reducing unnecessary interference to other traffic participants.

[0089] In one example, the specific steps to obtain the movement trend and behavioral intent of a target object are as follows:

[0090] S2.1.1: By extracting and analyzing the dynamic information of the target object within a continuous time window, the trajectory direction, acceleration rate of change, and heading angle change trends are obtained to determine the motion trend;

[0091] Specifically, in multi-target dynamic environments, precise lighting adjustment requires prior assessment of the current movement state and short-term motion trends of each target object. Motion trends form the basis for constructing a behavioral potential field model, directly determining the directionality, range, and attenuation method of the illuminance control area. Without accurate motion trend assessment, the subsequently generated illuminance suppression area will fail to align with the target's behavioral path, easily leading to occlusion misalignment and accidental lighting injuries. Therefore, a time-series model needs to be established during the continuous movement of the target to dynamically capture its motion state.

[0092] In this embodiment, dynamic information of the target object within a continuous time window is collected and analyzed, including at least the position coordinates, velocity vector, acceleration vector, and heading angle (i.e., the direction angle relative to the reference coordinate system) at several moments. This information can be obtained through image detection, radar tracking, inertial measurement devices, etc. By constructing an interpolated trajectory of the position sequence, the rate of change of velocity and the trend of heading angle change at each moment are calculated, thereby determining whether the target is in a stable straight-line movement, a turning transition, a deceleration and change of direction, or a cross-movement. For example, if the target's rate of change of acceleration is positive for several consecutive moments and the trend of heading angle change tends to be stable, it indicates that the target is in the acceleration straight-line movement phase; if the heading angle fluctuates greatly and the speed decreases, it can be determined that it is in a deceleration transition state with uncertain direction. The above trend analysis process not only provides a basis for judging the main axis of the target path but also provides a logical premise for predicting subsequent behavioral intentions.

[0093] S2.1.2: Based on the relative position of the target object, the intersection relationship between its trajectory and the vehicle's path, and the continuity characteristics of its movement trend, determine the behavioral intent according to the preset behavior pattern matching rules;

[0094] Specifically, judging the movement trend of a target only reflects its current physical state. However, the essential requirement of vehicle lighting control lies in predicting behavior, that is, determining whether the target will intersect with the vehicle's path, whether avoidance is necessary, or whether it will enter a danger zone. Therefore, it is necessary to further deduce the target's behavioral intention. Behavioral intention refers to the specific actions the target may perform in a short period of time (such as crossing the road, changing lanes, parking on the side of the road, overtaking, or stopping). Only after identifying this behavioral trend can a behavioral potential field with a targeted illuminance suppression structure be constructed to avoid false illumination and waste of illuminance resources.

[0095] In this embodiment, the target's behavioral intent is determined by acquiring the relative position between the target and the vehicle (including relative distance, relative speed, and angle) and the spatial relationship between the target's trajectory and the vehicle's path in the coordinate system (such as intersection angle, intersection point distance, and time arrival overlap). This determination is combined with the target's motion trend label. This determination process is based on behavior pattern matching rules, which pre-set multiple traffic behavior scenario models. For example, a pedestrian moving slowly from the roadside to the middle of the lane with a heading angle pointing towards the center of the lane can be matched as crossing the road; an electric vehicle in the front lane suddenly slows down and deviates to the right, which can be matched as parking on the side of the road; an oncoming vehicle is close to the vehicle and has a stable trajectory but no intersection with the vehicle's trajectory, which can be matched as moving away from the vehicle. The matching rules are modeled based on large sample traffic behavior data and transformed into a classifier. They can be implemented using rule trees, pattern templates, or lightweight neural network models, which will not be elaborated here.

[0096] S2.2: Map the behavioral intent to a target function type in a preset set of potential field constructors, wherein the potential field constructors are spatially distributed according to direction sensitivity, decay gradient and radius of action;

[0097] Specifically, the construction quality of the behavioral potential field model directly determines the spatial accuracy of the final illuminance weight map and the lighting control effect. After identifying the behavioral intent of the target object, a corresponding objective function type needs to be selected from a set of pre-built potential field constructors based on the identified intent type. This objective function type must have spatial perception capabilities and illuminance suppression distribution patterns that match the logical structure of the behavioral intent, ensuring that the final generated illuminance control area not only structurally conforms to the target behavior direction, but also that the response intensity and illuminance gradient match its risk level and behavioral uncertainty, thereby achieving optimization of individual lighting suppression effects and minimization of multi-target illuminance conflicts.

[0098] In this embodiment, the potential field constructor set is designed as a categorized, multi-parameter, and composable structural template library, containing several basic function prototypes. Each basic function prototype has well-defined spatial attenuation characteristics, directional response models, and modulation control parameters. The basic function prototypes include, but are not limited to, elliptic Gaussian functions, conical functions, edge diffusion functions, asymmetric multipole functions, and local gradient transition functions. Elliptic Gaussian functions are suitable for constructing illumination regions with stable paths and stationary or slowly approaching targets. Their main characteristics are high central suppression strength, rapid edge attenuation, and strong axial symmetry. Conical functions are used for high-speed approach to or traversal of targets, exhibiting a directionally sensitive structure with slow attenuation along the principal axis and rapid weakening in the lateral direction. Multipole distribution functions are suitable for targets with high behavioral uncertainty, such as targets with large trajectory disturbances and low prediction reliability. By distributing multiple secondary suppression kernels along multiple possible path directions, they achieve a non-centralized expansion of the coverage area, improving the inclusiveness and robustness of control.

[0099] As an example, the function prototype library not only covers spatial distribution forms but also includes built-in default configurations for function parameters. Parameter dimensions include center strength value, principal axis length, secondary axis width, boundary ambiguity coefficient, and directional gradient weights. During the mapping process from behavioral intent to target function type, the matching logic module is first invoked. Based on the behavioral intent type and the function label matching rules in the prototype library, the function prototype with the structural form closest to the target behavioral path is selected. For example, if the target is determined to be a pedestrian crossing, a conical eccentric function is used; if it is determined to be a merging electric vehicle, an elliptical off-axis Gaussian structure is used; and if it is determined to be a stationary obstacle, a symmetric center function is used. This mapping process can employ semantic label matching and rule tree structures, learning the matching mapping relationship between behavioral intent and spatial inhibition structures based on training data, thereby achieving automated and real-time function type selection.

[0100] In one example, the mapping between behavioral intent and function prototype employs a structured semantic classification decision-making process. First, the dynamic information of the target object within a continuous time window is abstracted and transformed into a set of behavioral description feature vectors. These features include, but are not limited to: ① trajectory direction change rate, ② mean acceleration, ③ heading angle stability, ④ target relative speed, ⑤ intersection angle with the vehicle path, ⑥ prediction time from the intersection point, ⑦ target size and type encoding (e.g., pedestrian / non-motorized vehicle / motorized vehicle), and ⑧ historical behavior label frequency. These features are normalized to form a multi-dimensional input vector. After initially identifying the target behavioral intent label (e.g., crossing, changing lanes, remaining stationary, moving to the side), this label and the multi-dimensional input vector are used as joint input and fed into the matching decision module. The matching module can be based on various implementable strategies, such as structured rule tree matching, matching vector lookup tables, and classifier models. After matching, the obtained function type label is used as an index to extract the function structure template and its default parameters from the prototype function library. Subsequently, the extracted template parameters can be fine-tuned by incorporating the current motion trend variables of the target object (speed, predicted path length, etc.), for example:

[0101] Adjust the function's main axis length to the predicted path length plus the buffer length;

[0102] Adjust the function center strength by multiplying the base value by the risk level coefficient;

[0103] Adjust the boundary ambiguity coefficient to be the inverse function of the current target trajectory stability.

[0104] As an example, after selecting the basic function prototype, the prototype itself is not used directly. Instead, its structural parameters need to be adjusted to enable it to dynamically adapt to the current target individual. This adjustment process combines the target object's state parameters within the current frame and past time windows, mainly including: current velocity magnitude, acceleration direction, trajectory curvature change, heading angle stability, and historical behavioral intent change frequency. These dimensions can be used to determine whether the target's behavioral trend is stable, whether the behavior deviates, and whether the path oscillates violently. For example, when the target trajectory is stable but close to the vehicle's path, the main axis direction can be lengthened to create an advance illumination avoidance zone; if the target's acceleration is unstable, it indicates high behavioral uncertainty, and lateral coverage needs to be expanded to avoid false illumination. Dynamic adjustment can be handled using linear scaling, nonlinear piecewise modulation, and morphological reconstruction methods, and compensation functions (such as edge-weighted diffusion functions) can be introduced to enhance multi-directional responsiveness.

[0105] Preferably, for certain complex behavioral intentions, such as a combination of merging to the left and then accelerating insertion, a single function form is insufficient to accurately represent the range of possible future paths. Therefore, this embodiment also includes a function combination construction mechanism, which allows multiple structural prototypes to be combined to form a composite function body under the same objective. Combination methods include directional superposition, axial splicing, and weight superposition. For example, a conical main function can be spliced ​​in the main direction, and two eccentrically distributed auxiliary function kernels can be placed on the flanks to construct a linkage suppression region.

[0106] Taking the pedestrian coordinate system as an example, you can refer to... Figure 4 To understand, Figure 4 A schematic diagram of the function deformation principle provided in the embodiments of this application, such as Figure 4 As shown, the pedestrian is currently located at the origin O of the coordinate system. Their current direction of motion is defined as the principal axis, represented by a straight line with an arrow in the two-dimensional coordinate system. This principal axis serves as the reference direction for constructing the behavioral potential field, and multiple lighting control sub-segments are divided along the principal axis within the local reference coordinate system. The figure illustrates two unequally spaced lighting control sub-segments, namely lighting control sub-segment 1 and lighting control sub-segment 2. The different sub-segments have different lengths along the principal axis and cover different angular ranges, reflecting their responsiveness to changes in the stability of the motion trend.

[0107] Figure 4 The diagram shows a set of conical eccentric functions arranged within each sub-segment to simulate the illumination suppression requirements of the target object in that area. These conical functions exhibit directional sensitivity, meaning they extend further in the direction of pedestrian movement and attenuate more rapidly laterally, demonstrating a clear behavioral correlation. Function parameters are adjusted based on local characteristics of each sub-segment, such as velocity change rate, trajectory curvature, and prediction uncertainty. Figure 4 It further demonstrates the gradual change of the function's form across multiple sub-segments, including changes in dimensions such as the function's principal axis length, lateral expansion width, and edge transition smoothness, showcasing the flexible response of the lighting control area to behavior prediction.

[0108] It is important to note that Figure 4 The shaded area in the diagram represents the spatial distribution of function values ​​formed by the conical eccentric function along the principal axis, used to express the distribution trend of illumination control intensity. This function structure is directionally sensitive; it is stretched relatively long along the principal axis and decays rapidly in the vertical direction, forming a control mode with strong forward suppression and weak lateral response. This spatial distribution can accurately adapt to behavioral intentions such as pedestrians crossing, changing lanes, or approaching, providing suppression responses to key risky behavior areas without interfering with the vehicle's main field of vision.

[0109] S2.3: Based on the motion trend, the corresponding objective function type is oriented and spatially scaled within the vehicle's forward lighting area, and then superimposed to generate a behavioral potential field model;

[0110] Please see Figure 5 This figure is a schematic flowchart of an objective function processing method provided in an embodiment of this application. Figure 5 The method shown can be applied to step S2.3 of the aforementioned method. In one example, the specific steps for directional alignment and spatial scaling of the corresponding objective function type within the vehicle's forward lighting area are as follows:

[0111] S2.3.1: Establish a local reference coordinate system with the current position of the target object as the origin and the direction corresponding to the movement trend as the main axis;

[0112] Specifically, to achieve illumination response capability in response to the target's behavioral direction, the target information in the original global coordinate system must be transformed into a local structured control framework. Since the target object's position and orientation vary within the global illumination area, directly generating the illumination function in the global coordinate system will result in a discrepancy between the function direction and the actual behavioral direction, leading to deviations or misalignment of the control area. Therefore, before constructing the behavioral potential field function, an independent local reference coordinate system must be built using the target object as a reference point. Its current position is taken as the origin, and the principal axis direction is defined by its current motion trend direction, thereby establishing structural consistency between the function construction and the behavioral space.

[0113] In this embodiment, the current position coordinates of the target object are obtained, and its motion trend direction is calculated using a set of trajectory points within a continuous time window. This direction can be obtained by fitting the linear principal axis of the target trajectory or predicting the path vector, and is defined as the principal axis in this local reference coordinate system. A two-dimensional Cartesian coordinate system is constructed with the target's current position as the origin, where the principal axis points towards the motion trend direction, the X-axis represents the lane orientation, and the Y-axis represents the opposing direction (which can be set according to the lane orientation). The Y-axis is perpendicular to the X-axis, forming an orthogonal basis. This coordinate system is used to express all spatial operations, including the construction of the objective function, parameter adjustment, and region division, ensuring that the function structure has directional consistency and response specificity. The existence of this local coordinate system provides a structural reference for subsequent functions during interpolation, deformation, and mapping, avoiding offset between the illumination area and the behavioral direction, and improving the directional adaptation accuracy of lighting control.

[0114] Furthermore, this local reference coordinate system will serve as the transformation benchmark during the construction of the behavioral potential function. The constructed function structure will undergo orientation alignment, scale calculation, and spatial mapping operations within this coordinate system. This coordinate system can also serve as a dynamically updated structure, reconstructing in real time based on trajectory changes during target motion. This ensures that the function possesses spatial consistency and temporal coherence across consecutive frames, thereby supporting the dynamic and stable output of the behavioral potential field.

[0115] S2.3.2: Based on the stability of the motion trend, the local reference coordinate system is divided into multiple unequal-distance lighting control sub-segments, and scaling scales and shape modulation parameters are set for the objective function type in different sub-segments.

[0116] The function exhibits a non-uniform distribution along the principal axis, and the scaling scale and shape modulation parameters are adjusted based on local features of sub-segments within a local reference coordinate system. These local features include illuminance attenuation rate, coverage radius, boundary ambiguity coefficient, and gradient directionality coefficient.

[0117] Specifically, during lighting control, the illuminance adjustment requirements for different targets at various locations along their behavioral paths are not consistent. This is especially true in areas where behavioral trends are unstable, acceleration fluctuates drastically, or paths are about to intersect, requiring a broader, softer, or higher-intensity illuminance suppression response. Therefore, to precisely express this difference spatially, the function control area along the main axis of motion needs to be divided into multiple sub-segments, and modulation parameters need to be set within each sub-segment to achieve local structural deformation of the function. This results in a non-uniform distribution of the final behavioral potential field, improving overall lighting adaptability.

[0118] In this embodiment, based on the principal axis direction, the local reference coordinate system is divided into multiple continuous but unequally spaced lighting control sub-segments along the travel direction. The division criteria are based on the calculation results of the target object's motion trend stability, specifically using parameters such as trajectory change rate, acceleration disturbance rate, and heading angle fluctuation for evaluation. If the target trajectory is stable and the direction is stable, a longer sub-segment range can be set to reduce the computational overhead of function reconstruction; however, if the target path experiences directional jumps, trajectory reversals, or velocity oscillations, the sub-segment length needs to be shortened to improve the local structure representation capability. The length and number of each sub-segment can be adaptively adjusted according to preset dynamic rules and are not fixed to a uniform distribution.

[0119] Furthermore, after dividing the function into sub-segments, it is necessary to set the scaling factor and shape modulation parameters of the function for the local behavioral characteristics within each sub-segment. The scaling factor refers to the structural dimensions of the function, such as its length of action and coverage width, in the principal axis and lateral directions. The modulation parameters include the illuminance decay rate (controlling the spatial decay rate of the function intensity), coverage radius (controlling the physical boundary of the area of ​​action), boundary ambiguity coefficient (controlling the transition gradient from the function edge to the main illumination area), and gradient directionality coefficient (controlling the degree of unevenness in illuminance distribution in different directions). These parameters can be adjusted based on the dynamic characteristics within the sub-segment, for example:

[0120] If the current sub-segment is located near the turning point of the target path, the heading angle fluctuation rate is large and the prediction path uncertainty is high, then the lateral coverage radius will be widened and the boundary ambiguity coefficient will be increased to form a light suppression zone with strong inclusiveness.

[0121] If the current sub-segment is a stable acceleration segment with a clear direction of motion, the edge transition area can be reduced and the intensity of illuminance suppression along the main axis can be enhanced.

[0122] If there is a predicted intersection point between the vehicle and the target path in the current area, the main shaft length and illuminance intensity are increased to suppress the lighting output in the intersection risk area in advance.

[0123] Please see Figure 6 This figure is a schematic flowchart of another objective function processing method provided in an embodiment of this application. Figure 6 The method shown can be applied to step S2.3 of the aforementioned method. In one example, the specific steps for superimposing functions to generate a behavioral potential model are as follows:

[0124] S2.3.3: Connect and combine the local deformation results of the objective function type in multiple unequal-distance lighting control sub-segments according to their spatial order in the local reference coordinate system to form an overall function structure spliced ​​together from multiple unequal-scale function segments;

[0125] Specifically, in order to construct a lighting control area that can realistically express the target behavior path and its changing trends, the local deformation results of the functions previously generated in each lighting control sub-segment must be organized into an overall structure according to a certain logic. In different sub-segments, the objective function structure usually exhibits significant differences in parameters such as length, width, boundary transition, and illuminance weight in spatial distribution. If it is directly superimposed or averaged, it will not only cause abrupt boundary changes and light field breaks, but also weaken the targeted response of each segment to its corresponding local dynamic behavior, making it difficult for the illuminance control area to have a consistent behavioral structure.

[0126] In this embodiment, the sub-segments are first spatially sorted according to the principal axis direction in the local reference coordinate system of the target object. The function structure within each sub-segment has been parameter-modulated based on local features, including but not limited to: principal axis length adjustment, lateral expansion adjustment, center intensity factor setting, and boundary ambiguity coefficient setting. In the combination stage, function fusion is achieved using intra-segment interpolation and interval splicing. That is, a transition function band is generated between two adjacent sub-segments, and the function parameters within the transition band transition smoothly using linear or nonlinear interpolation. For example, the principal axis length transitions from L1 in the previous segment to L2 in the next segment, which can be continuously changed by setting a weighted transition function in the intermediate segment. In addition, the illuminance center value can be weighted and averaged according to the risk level difference between the two segments, so that the illuminance control intensity has a risk-driven continuous adjustment characteristic in space.

[0127] Furthermore, to avoid issues such as enhanced illumination overlap, attenuation cliffs, and boundary asymmetry at the function splicing edges, the combined structure also includes a boundary fusion zone processing mechanism. The boundary fusion zone is defined as the transition region at the intersection of the edges of the two sub-segment function structures. In this region, a boundary fuzzing function superposition operation is performed. The fuzzing function can use edge Gaussian, bilateral adjustment templates, directional weight kernels, etc., to smooth the region based on the illuminance gradient, thereby avoiding the appearance of illuminance structure break zones. Simultaneously, the behavior label index of each local function segment is retained in the combined function as the basis data for subsequent illuminance weight map synthesis and risk level coding.

[0128] S2.3.4: Map the overall function structure to the vehicle's forward lighting area to construct the behavioral potential field model of the target object;

[0129] Specifically, the behavioral potential field model cannot be effective simply by constructing it in a local reference coordinate system. Its spatial structure must be accurately mapped back to the vehicle's forward lighting area coordinate system so that it truly corresponds to the spatial position of the physical road scene and thus affects the illuminance control module within the lighting assembly. If the function structure is not accurately mapped or the mapping error is too large, its position in the illuminance map will be offset, causing the actual lighting area to fail to maintain spatial alignment with the target path, affecting the accuracy of the behavioral response. Therefore, the complete function structure constructed in the local reference coordinate system needs to be transformed and aligned before being embedded into the vehicle's global coordinate system to complete the spatial embedding of the behavioral potential field model.

[0130] In this embodiment, the mapping of the function structure is based on the relative spatial relationship between the target object and the vehicle, including the target's position vector in the vehicle coordinate system, the principal axis direction angle, and the attitude angle between the current target and the vehicle. By establishing a transformation matrix between the target's local coordinate system and the global lighting control coordinate system, including rotation and translation transformations, the constructed overall function structure is projected onto the vehicle's front lighting plane. The projection process must preserve the function structure's directional features, intensity attenuation characteristics, and fuzzy boundaries to ensure that the mapped illuminance control area possesses the original behavioral semantic expression capability within the vehicle lighting system.

[0131] The specific steps for S3 are as follows:

[0132] S3.1: Select the overlapping region of multiple behavioral potential field models;

[0133] Specifically, when multiple targets exist simultaneously in the area forward of a vehicle, their corresponding behavioral potential field models may exhibit partial or complete spatial overlap, meaning that the control functions of different models intersect at certain coordinate positions. Such overlapping areas have a high probability of control conflicts: different targets may have opposing illumination requirements; for example, one target might want localized dimming, while another might want the area to remain bright or be further illuminated. Therefore, without screening, analysis, and subsequent adjudication mechanisms for overlapping areas, it is easy for the illumination map to exhibit control failures or inconsistencies at critical locations.

[0134] In this embodiment, to identify and extract overlapping regions among multiple models, it is first necessary to traverse the domain of the function mapped by each potential field model and convert it into a spatial mask form, that is, to encode the effective range of the function in a two-dimensional Boolean matrix. Under the unified illumination coordinate system, all model mask matrices are superimposed at the pixel level, and all spatial coordinate points simultaneously covered by two or more models are recorded to form an overlapping point set. Further, the overlapping point set is clustered to divide neighboring pixels into several spatial regions.

[0135] S3.2: Within the overlapping area, the dominant position of the overlapping area is determined based on the risk priority weight of each model, the area is assigned a value according to the illuminance suppression value of the dominant model, and the boundary areas of other models within the overlapping area are subjected to gradual transition processing.

[0136] Specifically, the overlapping region contains the superposition of multiple behavioral potential field control functions. Each function represents the behavioral prediction and illuminance demand of a target object. Due to limited lighting resources, it is impossible to simultaneously satisfy the suppression or retention requirements of all models. Therefore, it is necessary to divide the conflict region into dominant and subordinate models, i.e., to determine which model's control effect has priority within the overlapping region, with other models serving as references or suppressed secondary participants. If a dominant mechanism is not established and instead numerical weighted averaging is performed, important behavioral signals are easily diluted or covered, reducing the rationality of illuminance response decisions.

[0137] In this embodiment, risk priority weight is used as the indicator for determining dominance. Each model is pre-associated with behavioral intent labels and risk levels during construction, such as high-risk traversing behavior, medium-risk lane-changing behavior, and low-risk stable departure behavior, thereby assigning a risk weight value to the function's area of ​​effect. In each overlapping region, all functions covering that region are sorted according to their risk weight values, and the one with the highest risk weight is selected as the dominant model. The dominant model provides the illuminance suppression intensity value for that region, which is used for the initial assignment of the illuminance map for that region.

[0138] S3.3: Combine the non-overlapping areas and the processed overlapping areas according to their spatial locations, and use a continuous interpolation algorithm to fuse the boundaries of the areas to generate an illuminance weight map of the vehicle's forward lighting area.

[0139] Specifically, in the behavioral potential field model, there are regions that belong to the single-target influence region, i.e., non-overlapping regions. These regions already possess complete function structures and intensity values, eliminating the need for conflict judgment. However, there is a risk of abrupt changes in illuminance gradients at the boundaries between these regions and overlapping regions. Direct combination may result in abrupt changes in illuminance and edge spikes at the region connections, affecting the final lighting control quality. Therefore, a boundary fusion mechanism needs to be introduced when spatially synthesizing non-overlapping and overlapping regions to achieve gradient transitions of the illuminance function between regions, ensuring spatial continuity and visual logic consistency in the output image.

[0140] In this embodiment, all overlapping regions that have completed the dominant adjudication and boundary processing, along with the non-overlapping regions of each model, are uniformly mapped to a two-dimensional illuminance map in the illumination coordinate system. Spatial location information is used as a combination index to insert data from different source regions into the illuminance map function matrix in blocks. Edge fusion zones are set between the boundaries of each region, and a gradient-direction-aware interpolation algorithm is executed. The interpolation weights can be adjusted based on parameters such as the difference in illuminance values ​​between the regions on both sides of the boundary, boundary curvature, and function continuity requirements. Common strategies include directional bilateral weighted interpolation and boundary gradient template interpolation.

[0141] The specific steps for S4 are as follows:

[0142] S4.1: Establish a mapping relationship between the illuminance weight map and the spatial arrangement of the light-emitting units in the vehicle light group according to its two-dimensional coordinates. The mapping relationship is a one-to-one correspondence or regional correspondence between each pixel in the illuminance weight map and the corresponding light-emitting unit in the vehicle light group.

[0143] Specifically, the illuminance weight map is a two-dimensional continuous distribution map generated based on the fusion of behavioral potential field models. Its coordinate space represents the physical location of the vehicle's forward lighting area and its corresponding illuminance suppression requirement intensity. However, the actual objects of vehicle control are multiple light-emitting units arranged in the front of the vehicle. These light-emitting units exhibit a certain regular spatial arrangement, usually a two-dimensional array or a multi-segment beam group. Therefore, it is necessary to establish an accurate mapping relationship between the two-dimensional spatial pixels in the illuminance map and each light-emitting unit in the actual lamp group to ensure that the illuminance control command can be implemented in the specific physical hardware, achieving point-by-point fine adjustment of the target lighting area. Without this mapping structure, the illuminance map is only auxiliary information at the visual layer and cannot truly guide pixel-level lighting behavior, resulting in the entire control strategy failing to close the loop.

[0144] In this embodiment, the vehicle headlight assembly adopts a matrix structure, with internal light-emitting units divided into several rows and columns of grids, each unit possessing independent brightness adjustment capabilities. First, the array is coordinate-encoded to establish a light-emitting unit index matrix. Each headlight sub-unit has its two-dimensional coordinates in the vehicle space, such as defining left-right and up-down coordinates with the vehicle center as the origin. Then, the pixel coordinate space of the illuminance weight map and the light-emitting unit coordinate space are geometrically calibrated. A one-to-one correspondence between the two is established using projection mapping or proportional mapping mechanisms. Specifically, based on a physical size calibration function, the image pixel positions in the illuminance map are converted into actual illumination landing areas. Then, through an optical transformation model, such as a Fresnel projection model, the corresponding light-emitting unit coordinates are derived, forming a mapping index table. If the pixel density is greater than the light-emitting unit resolution, a region correspondence method can be used, where multiple image pixels converge into one light-emitting unit region, using the average, maximum, or weighted value as the brightness input.

[0145] S4.2: Convert the weight values ​​of each pixel in the illuminance weight map into the target brightness values ​​of the corresponding light-emitting units, and perform brightness normalization processing on the target brightness values ​​in combination with the current ambient light level;

[0146] Specifically, each pixel value in the illuminance weighting map represents the illuminance adjustment weight to be achieved at that spatial location. This value is essentially a dimensionless relative control quantity and cannot be directly used as the driving current or brightness command for the light-emitting unit. Therefore, it needs to be converted into a target brightness value with physical meaning, for example, in cd / m². 2Alternatively, the PWM duty cycle can be considered. This conversion process requires taking into account multiple factors: the light intensity required at the current target point, the background light level of the environment, and the current upper and lower limits of the brightness of the light-emitting unit. In addition, the weight expression methods of the potential field function of different targets may differ during construction. To achieve unified control, all weight values ​​need to be normalized to generate a target output matrix adapted to the current lighting strategy.

[0147] In this embodiment, the value range of the illuminance weight map is typically defined within the range of [0,1] or [0,100]. A larger value indicates stronger occlusion and lower illuminance, while a smaller value indicates that illumination should be maintained or enhanced. The conversion process includes the following steps: First, the background illumination value collected by the ambient light sensor is read as the illumination reference parameter for the current vehicle operating environment; then, based on the vehicle lighting mode, such as urban roads, highways, tunnels, etc., a brightness output strategy template is set for the current frame transmission time; finally, the weight value of each pixel in the illuminance map is mapped to brightness according to this strategy, and the brightness output strategy template achieves non-linear mapping through a lookup table function.

[0148] Furthermore, to improve the smoothness of illuminance output and human visual adaptability, the normalization process can also introduce a dynamic range compression strategy. This involves applying convolutional blurring or median filtering to the target luminance value in neighboring regions to suppress light field breaks caused by pixel-level abrupt changes. Additionally, if the vehicle is currently in high-speed driving mode, the luminance baseline level can be increased to ensure visibility; if a vehicle ahead or pedestrian crossing the lane is detected, the upper limit of the suppression value is increased accordingly, making the main suppression area darker. This normalization strategy transforms the control intent in the illuminance map into the physical control target of specific luminous units, providing standardized input for subsequent actual luminous control.

[0149] S4.3: Send the normalized brightness value to the drive control circuit of each light-emitting unit;

[0150] Specifically, after establishing the mapping relationship between the illuminance weighting map and the light-emitting units and calculating the target brightness, the target brightness value corresponding to each light-emitting unit needs to be sent as an instruction parameter to its underlying control circuit. The driver chip or controller then modulates the current, voltage, or PWM waveform of the light-emitting module to complete the final physical response of the lighting output. This step is the landing conversion process from control logic to the actual execution layer in the lighting control link. If the accuracy of the instruction issuance is insufficient or the response delay is too large, it will directly affect the real-time performance and accuracy of the lighting system.

[0151] In this embodiment, the lamp group control module includes a brightness command encoding / decoding module, a frame-level buffer control module, and a light-emitting unit control interface. Normalized brightness values ​​are encapsulated into brightness command packets, each containing the brightness value, target address, refresh flag, and other information. The command packets are sent to the lamp group main control chip via a high-speed serial interface (such as CAN, LIN, or Ethernet), where the chip decodes the command based on the target address and matches it to the specific light-emitting unit control channel. Control commands can be converted into PWM waveforms, current-driven data, or analog voltage levels to achieve point-to-point control of LED luminous intensity.

[0152] In one example, the vehicle lighting assembly of this application further includes position lights, front fog lights, rear fog lights, and taillights.

[0153] It is understood that the lighting control method of this application not only adjusts the brightness of the main lights step by step, but also controls the auxiliary lights in combination based on environmental perception information, vehicle operating status and the output results of the behavioral potential field model. The vehicle operating status includes, but is not limited to, vehicle speed, turning signals and weather information.

[0154] It is easy to understand that when low visibility or foggy weather scenes are detected, the front fog lights and rear fog lights are automatically turned on, and their brightness or activation sequence is adjusted according to the relative position of the target object; when the vehicle enters a narrow or poorly lit area, the parking lights are automatically activated to alert other road users; when parking at night or slowing down suddenly, the rear warning effect is enhanced by linking the taillights and parking lights.

[0155] Furthermore, based on the vehicle's current operating status, external environmental information, and the target object's behavioral potential field model, the on / off state, brightness level, and control timing of the vehicle's lighting group can be jointly scheduled, specifically including:

[0156] The timing for turning on and off each auxiliary light is determined based on the operating parameters of vehicle speed, turn signals, and rain / fog sensor data.

[0157] For example, when the vehicle is turning at low speed and the rain and fog sensor detects low visibility, the front fog lights and side marker lights are automatically turned on; when the vehicle is traveling at high speed and there are no obstructions, unnecessary lights are turned off to reduce power consumption.

[0158] Based on the output of the aforementioned behavioral potential field model, the brightness of each lamp is adjusted.

[0159] For example, when a pedestrian or an electric vehicle crossing an irregular path is detected in front, the illuminance of the relevant area of ​​the main headlights is reduced, while the lateral visibility of the side marker lights is enhanced; the brightness of the rear fog lights is adjusted according to the distance of the vehicle approaching from behind, so as to avoid interference caused by strong light at close range.

[0160] By integrating the results of target object trajectory prediction and vehicle path planning, the timing of issuing control commands for each lighting fixture is dynamically adjusted to achieve predictive response of light signals.

[0161] How to predict the trajectory and how to plan the path have been described in the foregoing and are existing technologies, so this application will not repeat them here.

[0162] It should be noted that when the behavioral potential field model determines that multiple target objects constitute potential interference in different directions, this application can simultaneously schedule multiple lamps to form a composite light field.

[0163] For example, in a nighttime urban intersection scenario, if the system detects both oncoming vehicles and pedestrians crossing the intersection, the headlights will dim in the direction of the target vehicle, the taillights will flash more intensely as a warning, and the system will decide whether to activate the front fog lights simultaneously to improve short-range illumination based on the vehicle speed.

[0164] As an example, this application provides a vehicle lighting control system, the system comprising:

[0165] The target recognition module is used to acquire dynamic information of multiple target objects in the environment in front of the vehicle;

[0166] The behavior analysis module is used to obtain the movement trends and behavioral intentions of each target object based on the dynamic information.

[0167] The potential field construction module is used to map behavioral intentions to target function types in a preset potential field constructor set, and to perform directional alignment and spatial scaling of the target function types according to the motion trend to generate a behavioral potential field model of the target object.

[0168] The potential field combination module is used to map the behavioral potential field models of multiple target objects to a unified coordinate system of the vehicle's forward lighting area, and to perform regional assignment and boundary transition processing based on the risk priority weights of each model in the overlapping area to generate a two-dimensional illuminance weight map with spatial continuity.

[0169] The lamp group control module is used to map the illuminance weight map to each light-emitting unit in the vehicle lamp group and control the light-emitting unit to perform point-by-point brightness adjustment.

[0170] The potential field construction module includes:

[0171] The function selection unit is used to map the behavioral intent of the target object to the target function type in the preset potential field constructor set. The target function type has spatial distribution characteristics of direction sensitivity, decay gradient and radius of action.

[0172] The coordinate alignment unit is used to establish a local reference coordinate system in the vehicle's forward lighting area with the current position as the origin and the direction of movement as the main axis, based on the movement trend of the target object.

[0173] A partitioned modulation unit is used to divide the local reference coordinate system into multiple unequal-distance lighting control sub-segments, and to set the scaling scale and shape modulation parameters for the objective function type in different sub-segments respectively;

[0174] The function splicing unit is used to splice and combine the function fragments after sub-segment modulation in spatial order to generate an overall function structure, and then map the structure onto the vehicle's forward lighting area to form a behavioral potential field model of the corresponding target object.

[0175] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A vehicle lighting control method, applied to a vehicle lighting assembly, wherein the vehicle lighting assembly is disposed on the front structure of the vehicle and its illumination direction covers the main lane and adjacent lane areas in front of the vehicle, characterized in that, The method includes: Acquire dynamic information of multiple target objects in the environment in front of the vehicle; A behavioral potential field model is constructed based on the dynamic information, and the behavioral potential field model is spatially superimposed to generate an illuminance weight map of the vehicle's forward lighting area, wherein the illuminance weight map is a two-dimensional distribution map with spatial continuity. The illuminance weight map is mapped to the vehicle headlight assembly so that the vehicle headlight assembly can perform point-by-point brightness adjustment. Constructing a behavioral potential field model based on the dynamic information includes: Based on the dynamic information, the movement trend and behavioral intent of the target object are obtained; The behavioral intent is mapped to the target function type in a preset set of potential field constructors, wherein the potential field constructors are spatially distributed according to direction sensitivity, decay gradient and radius of action; Based on the motion trend, the corresponding objective function types are oriented and spatially scaled within the vehicle's forward lighting area, and then superimposed to generate a behavioral potential field model. Based on the aforementioned motion trend, the corresponding objective function type is directionally aligned and spatially scaled within the vehicle's forward lighting area, including: Establish a local reference coordinate system with the current position of the target object as the origin and the direction corresponding to the movement trend as the main axis; Based on the stability of the motion trend, the local reference coordinate system is divided into multiple unequal-distance lighting control sub-segments, and scaling scales and shape modulation parameters are set for the objective function type in different sub-segments respectively. The superposition process to generate a behavioral potential field model includes: The local deformation results of the objective function type in multiple unequal-distance lighting control sub-segments are connected and combined according to their spatial order in the local reference coordinate system to form an overall function structure spliced ​​together from multiple unequal-scale function segments. The overall function structure is mapped to the vehicle's forward lighting area to construct a behavioral potential field model of the target object; The behavioral potential field models are spatially superimposed to generate an illuminance weight map of the vehicle's forward lighting area, including: Select the overlapping region of multiple behavioral potential field models; Within the overlapping area, the dominance of the overlapping area is determined based on the risk priority weight of each model, and the area is assigned a value according to the illuminance suppression value of the dominant model. At the same time, the boundary areas of other models within the overlapping area are subjected to gradual transition processing. The non-overlapping areas and the processed overlapping areas are combined according to their spatial location, and the boundary is fused by a continuous interpolation algorithm on the area edges to generate an illuminance weight map of the vehicle's forward lighting area.

2. The vehicle lighting control method according to claim 1, characterized in that, The scaling scale and shape modulation parameters are adjusted based on local features of sub-segments within a local reference coordinate system, including illuminance attenuation rate, coverage radius, boundary ambiguity coefficient, and gradient directionality coefficient.

3. The vehicle lighting control method according to claim 1, characterized in that, Based on the dynamic information, the movement trend and behavioral intent of the target object are obtained, including: By extracting and analyzing the dynamic information of the target object within a continuous time window, the trajectory direction, acceleration rate of change, and heading angle change trends are obtained to determine the motion trend; By combining the relative position of the target object, the intersection relationship between its trajectory and the vehicle's path, and the continuous characteristics of its movement trend, the behavioral intent is determined based on preset behavior pattern matching rules.

4. The vehicle lighting control method according to claim 1, characterized in that, Mapping the illuminance weight map to the vehicle headlight assembly to enable point-by-point brightness adjustment of the vehicle headlight assembly includes: The illuminance weight map is mapped to the spatial arrangement of the light-emitting units in the vehicle light group according to its two-dimensional coordinates. The mapping relationship is a one-to-one correspondence or regional correspondence between each pixel in the illuminance weight map and the corresponding light-emitting unit in the vehicle light group. The weight values ​​of each pixel in the illuminance weight map are converted into the target brightness values ​​of the corresponding light-emitting units, and the target brightness values ​​are normalized in combination with the current ambient light level. The normalized brightness value is sent to the drive control circuit of each light-emitting unit.

5. A vehicle lighting control method according to claim 4, characterized in that, The vehicle lighting assembly further includes position lights, front fog lights, rear fog lights, and taillights, and the method further includes: Based on the vehicle's current operating status, external environment information, and the target object's behavioral potential field model, the on / off status, brightness level, and control timing of the vehicle's lighting group are jointly scheduled.

6. A vehicle lighting control system for implementing the vehicle lighting control method as described in any one of claims 1-5, characterized in that, The system includes: The target recognition module is used to acquire dynamic information of multiple target objects in the environment in front of the vehicle; The behavior analysis module is used to obtain the movement trends and behavioral intentions of each target object based on the dynamic information. The potential field construction module is used to map behavioral intentions to target function types in a preset potential field constructor set, and to perform directional alignment and spatial scaling of the target function types according to the motion trend to generate a behavioral potential field model of the target object. The potential field combination module is used to map the behavioral potential field models of multiple target objects to a unified coordinate system of the vehicle's forward lighting area, and to perform regional assignment and boundary transition processing based on the risk priority weights of each model in the overlapping area to generate a two-dimensional illuminance weight map with spatial continuity. The lamp group control module is used to map the illuminance weight map to each light-emitting unit in the vehicle lamp group and control the light-emitting unit to perform point-by-point brightness adjustment.

Images