A mobile luminaire motion control method and system based on environmental perception

Abstract

The present application relates to the technical field of intelligent lighting control, and particularly relates to a mobile lamp motion control method and system based on environment perception, comprising: S1. extracting independent space-time topology features of environment perception data, and converting environment perception data in each dimension into corresponding topology feature nodes; S2. taking the spatial distribution of a dynamic target as a topology reference. The present application completes the whole process conversion of environment perception information to lamp motion control through a topological processing mode, converts multi-source heterogeneous environment perception data into topology feature nodes with independent space-time attributes, completes global space-time topology reconstruction taking the spatial distribution of a dynamic target as a reference, constructs an environment perception unified topology network, unifies the space-time reference of multi-source perception data, eliminates the natural time difference and spatial offset in the data collection process of different perception components, and solves the problem of implicit bias in motion control caused by the space-time mismatch characteristics of perception data.

Description

Technical Field

[0001] This invention relates to the field of intelligent lighting control technology, specifically to a method and system for controlling the motion of mobile lighting fixtures based on environmental perception. Background Technology

[0002] Mobile lighting fixtures are widely used in outdoor dynamic operations, emergency rescue, and intelligent inspection scenarios. They rely on environmental perception technology to capture multi-dimensional information about the scene, target, and surrounding environment, and combine this with a control unit to adjust the motion state and lighting direction to achieve continuous and stable illumination of dynamic targets. Current industry control technologies generally establish a basic process of environmental information acquisition, multi-source data processing, and motion command output. The core focus is on improving target recognition accuracy and optimizing command generation efficiency to meet the basic dynamic lighting needs of mobile lighting fixtures.

[0003] Patent CN119313917A discloses an intelligent control method and system for a mobile lighting device. This solution extracts target features through image preprocessing technology, combines the real-time positioning data of the target and the lamp, generates motion control commands through algorithm fitting, and simultaneously matches lighting parameters to complete directional adjustment. The core of this solution realizes the basic tracking and lighting adaptation of the lamp to the target, which is a typical technical solution in this field.

[0004] While existing technologies achieve basic environmental perception and command output, they fail to address the spatiotemporal heterogeneity of multi-source sensing data. Different types of sensing components inherently exhibit time differences and spatial offsets during data acquisition. For example, environmental image acquisition components and airflow sensing components have different response rhythms, and the movement of the luminaire itself further amplifies these spatiotemporal differences, making it impossible to form a unified reference benchmark from the acquired multi-source data. Existing control processes directly fuse uncalibrated heterogeneous data, failing to eliminate the inherent spatiotemporal biases. These biases transform into implicit motion trajectory offsets during command processing. These offsets are not directly caused by external disturbances or response lags but rather stem from the spatiotemporal mismatch characteristics of the sensing data itself. As the operation continues, these implicit biases accumulate, ultimately leading to a continuous decrease in the alignment between the luminaire's motion trajectory and the target area, while also implicitly affecting the stability of the lighting direction. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a motion control method and system for mobile lighting fixtures based on environmental perception, which solves the problems of spatiotemporal mismatch of multi-source heterogeneous sensing data causing accumulation of control deviations and insufficient accuracy and stability of dynamic lighting tracking in existing technologies.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a motion control method for mobile lighting fixtures based on environmental perception, comprising: S1. Extract independent spatiotemporal topological features from environmental perception data and transform environmental perception data of each dimension into corresponding topological feature nodes; S2. Using the spatial distribution of dynamic targets as the topological benchmark, the topological feature nodes are reconstructed in a global spatiotemporal manner to construct a unified topological network for environmental perception. S3. Establish a lighting fixture motion control topology model, perform bidirectional topology mapping between the unified environmental perception topology network and the lighting fixture motion control topology model, complete the motion control logic calculation through topology node matching, and generate motion control commands; S4. Convert the motion control command into a drive signal to drive the lamp to adjust the motion trajectory and lighting direction; S5. Collect the actual operating topology state after the lamp moves, and send the actual operating topology state back to the topology reconstruction stage. Based on the matching degree between the actual operating topology state and the unified topology network for environmental perception, iteratively optimize the topology reconstruction rules and mapping logic.

[0007] Furthermore, in S1, the extraction of independent spatiotemporal topological features from environmental perception data includes: acquiring multi-source heterogeneous data using image perception components, laser scanning components, airflow perception components, and inertial measurement components; analyzing the rate of change of multi-source heterogeneous data on the time axis and its distribution density in the three-dimensional spatial coordinate system using feature dimensionality reduction algorithms; defining the timestamp identifier and spatial pose parameters of each data cluster; and forming topological feature nodes with independent spatiotemporal attributes.

[0008] Furthermore, the topological feature node includes: time identification information, spatial location vector information, and environmental energy distribution intensity information. During the extraction process, the acquisition frequency of the sensing data in each dimension is normalized to eliminate the difference in data sampling interval between different sensors, so that each topological feature node has comparable attribute parameters under a unified time slice.

[0009] Furthermore, in S2, using the spatial distribution of the dynamic target as a topological reference includes: identifying the geometric center and motion vector of the dynamic target in the environment, mapping the motion path of the dynamic target into a three-dimensional topological skeleton, constructing a relative coordinate system with the three-dimensional topological skeleton as the origin, and mapping the topological feature nodes into the relative coordinate system.

[0010] Furthermore, in S2, the global spatiotemporal topology reconstruction of the topological feature nodes includes: calculating the spatiotemporal offset of each topological feature node relative to the dynamic target topological reference, correcting the logical association strength between each node through a spatiotemporal smoothing algorithm, eliminating abnormal nodes with spatiotemporal heterogeneity, and establishing a unified environmental perception topology network describing the environmental perception state and the logical association between the dynamic target.

[0011] Furthermore, in S3, the establishment of the lamp motion control topology model includes: constructing a motion control topology model that reflects the lamp's execution capability based on the degree-of-freedom parameters of the lamp's motion mechanism, the motor response curve, and the lighting spot diffusion model. The motion control topology model consists of multiple control nodes representing displacement, rotational speed, and pitch angle.

[0012] Furthermore, in S3, the step of performing bidirectional topology mapping between the unified environmental perception topology network and the lighting motion control topology model includes: calculating the spatial distance correlation coefficient and time series correlation coefficient between the sensing nodes in the unified environmental perception topology network and the control nodes in the motion control topology model, establishing a cross-domain mapping matrix, and realizing real-time mapping of sensing information to control commands through the cross-domain mapping matrix.

[0013] Furthermore, in S4, the driving signal includes: a motor stepping pulse signal and a gimbal rotation angle control signal. During the adjustment process, the convergence adjustment and optical axis pointing adjustment of the illumination beam are performed simultaneously to keep the center position of the illumination spot coincide with the spatial topological center point of the dynamic target.

[0014] Furthermore, in S5, the iterative optimization topology reconstruction rules and mapping logic include: defining the actual running topology state as the observation vector, defining the ideal state of the environment-aware unified topology network as the target vector, calculating the Euclidean distance error between the observation vector and the target vector, and adjusting the weight coefficients of the cross-domain mapping matrix through a backpropagation algorithm when the Euclidean distance error exceeds a preset threshold.

[0015] The present invention also provides a motion control system for mobile lighting fixtures based on environmental perception, comprising: The topology feature extraction module is used to extract independent spatiotemporal topology features from environmental perception data, transforming environmental perception data of various dimensions into corresponding topology feature nodes. The spatiotemporal topology reconstruction module is used to reconstruct the topology feature nodes in the whole domain using the spatial distribution of dynamic targets as the topology benchmark, and to build a unified topology network for environmental perception. The bidirectional topology mapping module is used to establish a lighting motion control topology model and perform bidirectional topology mapping between the unified environmental perception topology network and the lighting motion control topology model. The motion command solution module is used to solve the motion control logic by matching topology nodes and generate motion control commands. The execution drive module is used to convert the motion control command into a drive signal to drive the lamp to complete the adjustment of the motion trajectory and lighting direction; The topology status acquisition module is used to acquire the actual operating topology status after the lamps have moved. The iterative optimization module is used to iteratively optimize the topology reconstruction rules and mapping logic based on the matching degree between the actual running topology state and the environment-aware unified topology network. The global loop control module is used to control the above modules to continuously execute the process of processing the environmental perception data and generating motion control commands.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention completes the entire process of transforming environmental perception information into lighting fixture motion control through topology processing. It transforms multi-source heterogeneous environmental perception data into topological feature nodes with independent spatiotemporal attributes. Based on the spatial distribution of dynamic targets, it completes full-domain spatiotemporal topology reconstruction, constructs a unified environmental perception topology network, unifies the spatiotemporal reference of multi-source perception data, eliminates the inherent time difference and spatial offset during data acquisition from different perception components, and solves the problem of implicit deviations in motion control caused by the spatiotemporal mismatch of the perception data itself. Through bidirectional topology mapping between the unified environmental perception topology network and the lighting fixture motion control topology model, it completes the accurate calculation and transformation of perception information into motion control commands, synchronously collects the actual operating status of the lighting fixture and transmits it back to the front end of the control link, and iteratively optimizes the topology reconstruction rules and mapping logic based on the matching degree between the operating status and the ideal state, avoiding the continuous accumulation of deviations in the control process, improving the fit between the lighting fixture motion trajectory and the target lighting area, ensuring the continuous stability of lighting direction in dynamic operation scenarios, and providing mobile lighting fixtures with complete motion control logic with spatiotemporal consistency, realizing continuous and stable lighting control of targets in dynamic scenarios. Attached Figure Description

[0017] Figure 1 This is a closed-loop diagram of the entire process of motion control of mobile lighting fixtures based on environmental perception according to the present invention. Figure 2 This is a flowchart of the environmental perception data topology feature extraction process of the present invention; Figure 3 This is a flowchart of the global spatiotemporal topology reconstruction and unified topology network construction of the present invention; Figure 4 This is a flowchart of the bidirectional topology mapping and motion control command solving process of the present invention; Figure 5 This is a flowchart of the lamp motion execution and closed-loop iterative optimization process of the present invention; Figure 6 This is a diagram illustrating the architecture of the motion control system for mobile lighting fixtures based on environmental perception, as described in this invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1-5 This invention provides a motion control method for mobile lighting fixtures based on environmental perception, comprising: S1. Extracting independent spatiotemporal topological features from environmental perception data, and converting environmental perception data of each dimension into corresponding topological feature nodes; S2. Using the spatial distribution of dynamic targets as the topological benchmark, performing global spatiotemporal topological reconstruction on the topological feature nodes to construct a unified environmental perception topological network; S3. Establishing a lighting fixture motion control topological model, performing bidirectional topological mapping between the unified environmental perception topological network and the lighting fixture motion control topological model, and completing the motion control logic calculation through topological node matching to generate motion control commands; S4. Converting the motion control commands into drive signals to drive the lighting fixture to complete the adjustment of its motion trajectory and lighting direction; S5. Collecting the actual operating topological state of the lighting fixture after movement, transmitting the actual operating topological state back to the topological reconstruction stage, and iteratively optimizing the topological reconstruction rules and mapping logic based on the matching degree between the actual operating topological state and the unified environmental perception topological network.

[0020] This solution completes the entire process of transforming environmental perception information into luminaire motion control through topology processing. It constructs a complete control link from perception data feature extraction, spatiotemporal benchmark unification, control logic mapping to closed-loop iterative optimization. This solves the problem of spatiotemporal mismatch in multi-source heterogeneous perception data in existing technologies, which leads to the accumulation of implicit deviations and insufficient stability of lighting direction in the motion control process. It provides mobile luminaires with spatiotemporally consistent motion control logic, enabling continuous and stable lighting control of mobile luminaires in dynamic scenarios.

[0021] In one specific embodiment, please refer to Figure 2 In S1, independent spatiotemporal topological features of environmental perception data are extracted, including: acquiring multi-source heterogeneous data using image perception components, laser scanning components, airflow perception components and inertial measurement components; analyzing the rate of change of multi-source heterogeneous data on the time axis and the distribution density in the three-dimensional spatial coordinate system through feature dimensionality reduction algorithm; defining the timestamp identifier and spatial pose parameters of each data cluster; and forming topological feature nodes with independent spatiotemporal attributes.

[0022] Specifically, the image perception component uses visible light or infrared cameras to collect visual image data of the environment, capturing the contours and positional information of dynamic targets, adapting to scene perception needs under different lighting conditions. The laser scanning component uses single-line or multi-line LiDAR to acquire three-dimensional spatial point cloud data of objects in the environment, accurately representing the distance information between spatial obstacles and targets, compensating for the shortcomings of visual perception in depth measurement. The airflow perception component uses wind pressure and wind speed sensors to collect airflow parameters in the environment, providing data support for predicting wind resistance disturbances in outdoor lighting scenarios, and proactively avoiding the impact of environmental disturbances on lighting movement. The inertial measurement unit uses a six-axis or nine-axis IMU module, installed on the lighting body and motion actuator, to collect the lighting's own motion attitude, acceleration, and angular velocity data, eliminating the interference of the lighting's own motion on the perception data and ensuring the baseline stability of the perception data. The feature reduction algorithm employs principal component analysis (PCA), retaining principal components with a cumulative variance contribution rate greater than or equal to 95%. The maximum number of principal components is set to eight. This algorithm performs dimensionality reduction on multi-source heterogeneous data, eliminating redundant dimensions and retaining core principal components that characterize the spatiotemporal properties of the data, thus reducing the computational burden of subsequent data processing. The rate of change on the time axis refers to the magnitude of numerical change of data within consecutive time slices of the same dimension. The distribution density in the three-dimensional spatial coordinate system refers to the degree of clustering of data points in three-dimensional space within the same time slice. Through dual filtering using the rate of change and distribution density, data clusters are divided, with each cluster corresponding to a set of data possessing the same spatiotemporal variation characteristics. Timestamps are calibrated using a unified system clock to ensure consistent time references across all data clusters. Spatial pose parameters are represented using three-dimensional coordinate vectors and attitude angles in the world coordinate system. The resulting topological feature nodes each correspond to unique spatiotemporal attributes, achieving standardized topological transformation of multi-source heterogeneous data and laying the foundation for subsequent spatiotemporal reference unification.

[0023] In one specific embodiment, please refer to Figure 2 The topological feature nodes include: time identification information, spatial location vector information, and environmental energy distribution intensity information. During the extraction process, the acquisition frequency of the sensing data in each dimension is normalized to eliminate the difference in data sampling interval between different sensors, so that each topological feature node has comparable attribute parameters under a unified time slice.

[0024] Specifically, time stamp information is recorded using nanosecond-level timestamps from the system's global clock, ensuring that the accuracy of the time stamp meets the real-time control requirements of dynamic scenarios and providing a unified time reference for all sensing data. Spatial position vector information is represented using coordinate vectors in a three-dimensional Cartesian coordinate system. The origin of the coordinate system is the world coordinate system origin corresponding to the initial installation position of the lighting fixture, and the vector direction follows the right-hand coordinate system rule, achieving a unified representation of the spatial position of all sensing data. Environmental energy distribution intensity information is represented according to the type of different sensing components. The energy distribution intensity corresponding to image sensing components is the grayscale value of image pixels or infrared radiation intensity; the energy distribution intensity corresponding to laser scanning components is the intensity value of laser echo signals; the energy distribution intensity corresponding to airflow sensing components is the dynamic pressure value of airflow; and the energy distribution intensity corresponding to inertial measurement components is the amplitude of motion acceleration, achieving attribute unification for different types of sensing data. The frequency normalization process is implemented using a resampling algorithm. Sensing data with different sampling frequencies are uniformly resampled. For example, if a uniform sampling frequency of 100Hz is set, sensor data with sampling frequencies higher than 100Hz are downsampled using mean filtering, while sensor data with sampling frequencies lower than 100Hz are upsampled using linear interpolation. This processing method eliminates the sampling interval differences between different sensors, ensuring that all topological feature nodes correspond to the same time slice and have comparable attribute parameters under the same time reference. This eliminates the impact of spatiotemporal heterogeneity at the data acquisition level and avoids deviations caused by asynchronous data in subsequent processing.

[0025] In one specific embodiment, please refer to Figure 3 In S2, the spatial distribution of dynamic targets is used as the topological reference, including: identifying the geometric center and motion vector of dynamic targets in the environment, mapping the motion path of dynamic targets into a three-dimensional topological skeleton, constructing a relative coordinate system with the three-dimensional topological skeleton as the origin, and mapping topological feature nodes to the relative coordinate system.

[0026] Specifically, dynamic target identification is achieved through a pre-trained target detection network. For different scenarios such as emergency rescue and intelligent inspection, corresponding target detection models can be trained to identify dynamic targets requiring continuous illumination within the scene, such as inspection personnel, rescue vehicles, and mobile work equipment. The target identification process simultaneously distinguishes target types, adapting to the lighting needs of different targets. The geometric center of the dynamic target is calculated using the center coordinates of the target's bounding rectangle obtained from target identification, corresponding to the center point coordinates in the three-dimensional coordinate system, serving as the core reference point for target lighting. The motion vector of the dynamic target is calculated through the positional changes of the geometric center within continuous time slices, including two core parameters: direction and speed of motion. This is used to predict the target's subsequent movement trend and pre-adapt the lighting control parameters. The motion path of the dynamic target is obtained by fitting the geometric center position points within continuous time slices. A cubic B-spline curve is used to fit the discrete position points, resulting in a smooth motion path curve. This motion path curve is mapped to a three-dimensional topological skeleton. The nodes of the topological skeleton correspond to key position points on the motion path, and the orientation of the skeleton corresponds to the direction of the motion path, comprehensively representing the target's motion pattern. A relative coordinate system is constructed using a 3D topological skeleton as the origin. The X-axis of the relative coordinate system is along the direction of the motion vector of the dynamic target, the Y-axis is perpendicular to the X-axis and parallel to the horizontal plane, and the Z-axis is perpendicular to the horizontal plane and upwards. By constructing this relative coordinate system, the topological feature nodes that were originally based on the world coordinate system are transformed into nodes based on the relative coordinate system of the dynamic target. This allows all topological feature nodes to be spatially represented with the dynamic target as the core reference, eliminating the reference offset caused by the movement of the lamp itself and changes in the environmental space, and providing a unified spatial reference for subsequent global spatiotemporal topological reconstruction.

[0027] In one specific embodiment, please refer to Figure 3 In S2, a global spatiotemporal topology reconstruction is performed on the topological feature nodes, including: calculating the spatiotemporal offset of each topological feature node relative to the dynamic target topology benchmark, correcting the logical association strength between each node through a spatiotemporal smoothing algorithm, eliminating abnormal nodes with spatiotemporal heterogeneity, and establishing a unified environmental perception topology network that describes the environmental perception state and the logical association between the dynamic target.

[0028] Specifically, the calculation of spatiotemporal offset is divided into two parts: time offset and spatial offset. The time offset is the difference between the time identifier of the topological feature node and the reference time of the current time slice. The formula for calculating the combined spatiotemporal offset is as follows:

[0029] in, To integrate spatiotemporal offset, This is the time dimension weighting coefficient, with a value of 0.4. This is the spatial dimension weighting coefficient, with a value of 0.6. This is the normalized time offset. This is the normalized spatial offset. Normalized time offset. The calculation formula is:

[0030] in, For the first The time identifier of each topological feature node The base time for the current time slice. The maximum allowed time offset is set to 0.02 s. Normalized spatial offset. The calculation formula is:

[0031] in, For the first Spatial location vectors of topological feature nodes The spatial position vector of the geometric center of the dynamic target. The maximum permissible spatial offset is set to 10m. The offsets in two dimensions are used to fully characterize the spatiotemporal correlation between topological feature nodes and the dynamic target topological reference. The spatiotemporal smoothing algorithm employs a four-dimensional spatiotemporal Gaussian kernel smoothing algorithm. The four-dimensional spatiotemporal Gaussian kernel = three-dimensional space (x, y, z) + one-dimensional time (t), suitable for point clouds, three-dimensional pose, and other stereoscopic spatiotemporal data. The spatial kernel bandwidth is set to 0.5m, the temporal kernel bandwidth to 0.01s, the smoothing window size to a 3×3×3×3 spatiotemporal cube, and the smoothing coefficient to 0.8. This smooths the spatiotemporal offsets of each topological feature node. Based on the magnitude of the spatiotemporal offset, the logical correlation strength between nodes is corrected. The smaller the spatiotemporal offset of a node, the stronger its correlation with the dynamic target topological reference, and the greater its weight in the topological network. Conversely, the larger the spatiotemporal offset, the weaker its correlation with the dynamic target topological reference, and the smaller its weight in the topological network. By correcting the correlation strength, core perceptual information relevant to the dynamic target lighting requirements is highlighted, while interference from irrelevant information is weakened. Anomaly nodes exhibiting spatiotemporal heterogeneity refer to nodes whose temporal or spatial offsets exceed a preset reasonable range. For example, nodes with temporal offsets exceeding a single time slice interval, or nodes with spatial offsets exceeding the dynamic target lighting requirements. These anomaly nodes are removed to prevent abnormal data from interfering with the construction of the topology network and to ensure its accuracy. The unified environmental perception topology network uses the topology nodes corresponding to dynamic targets as core nodes and other topology feature nodes as edge nodes. The connection edges between nodes correspond to the logical association strength between them. Through this topology network, a unified representation of all environmental perception data under the same spatiotemporal reference is achieved, solving the spatiotemporal heterogeneity problem of multi-source perception data and providing a unified and accurate perception data foundation for subsequent control command calculations.

[0032] In one specific embodiment, please refer to Figure 4 In S3, a motion control topology model for the lighting fixture is established, including: based on the degree-of-freedom parameters of the lighting fixture's motion mechanism, the motor response curve, and the lighting spot diffusion model, a motion control topology model reflecting the lighting fixture's execution capability is constructed. The motion control topology model consists of multiple control nodes representing displacement, rotational speed, and pitch angle.

[0033] Specifically, the degrees of freedom parameters of the luminaire's motion mechanism are determined based on the actual structure of the luminaire. For example, the walking mechanism of the moving luminaire has translational degrees of freedom in the X and Y axes, while the pan-tilt mechanism has rotational degrees of freedom in the horizontal and pitch directions. The corresponding degrees of freedom parameters are the range of motion, maximum speed, and acceleration limit for each degree of freedom, fully characterizing the executable range of the luminaire's motion mechanism. The motor response curve is obtained by actually calibrating the drive motor corresponding to the luminaire's motion mechanism. During the calibration process, parameters such as the motor's speed response time, steady-state speed error, and stall torque under different control signals are recorded, and the motor's response curve is fitted to characterize the motor's actual execution capability, avoiding control commands exceeding the motor's response capability range and ensuring the executability of the control commands. The lighting spot diffusion model is constructed based on the luminaire's light source parameters and optical lens parameters, characterizing the correspondence between lighting distance, spot diameter, and spot illuminance. This model can predict the lighting effect under different luminaire poses, ensuring that the actual coverage effect of the lighting spot is considered during the generation of control commands, meeting the illuminance requirements of the target lighting. The control nodes of the motion control topology model correspond to the control parameters of each degree of freedom of the lamp's motion mechanism. The control node representing displacement corresponds to the translational displacement parameter of the walking mechanism, the control node representing rotational speed corresponds to the rotational speed parameter of the drive motor, and the control node representing pitch angle corresponds to the pitch and horizontal rotation angle parameters of the pan-tilt unit. The connection relationship between each control node corresponds to the coupling relationship between each degree of freedom of motion. Through this topology model, the actual execution capability of the lamp is topologically represented, forming an isomorphic topology structure with the unified topology network of environmental perception, laying the foundation for subsequent bidirectional topology mapping.

[0034] In one specific embodiment, please refer to Figure 4 In S3, a bidirectional topology mapping is performed between the unified environmental perception topology network and the lighting motion control topology model. This includes: calculating the spatial distance correlation coefficient and time series correlation coefficient between the sensing nodes in the unified environmental perception topology network and the control nodes in the motion control topology model; establishing a cross-domain mapping matrix; and realizing the real-time mapping of sensing information to control commands through the cross-domain mapping matrix.

[0035] Specifically, the spatial distance correlation coefficient characterizes the degree of correlation between sensing nodes and control nodes in the spatial dimension, while the time series correlation coefficient characterizes the degree of correlation between sensing nodes and control nodes in the time dimension. By calculating the correlation coefficients in both dimensions, a complete correlation between sensing information and control parameters is established. The formula for calculating the spatial distance correlation coefficient uses the Pearson correlation coefficient, and the formula is as follows:

[0036] in, Let be the spatial distance correlation coefficient between the i-th sensing node and the j-th control node. Let be the coordinates of the i-th sensing node in the k-th spatial dimension. Let be the mean coordinates of the i-th sensing node across all spatial dimensions. Let be the corresponding parameter value of the j-th control node in the k-th spatial dimension. Let be the average parameter of the j-th control node across all spatial dimensions, and n be the total number of spatial dimensions.

[0037] The formula for calculating the correlation coefficient of time series also uses the Pearson correlation coefficient, which is calculated for the node parameters within a continuous time series. The formula is as follows:

[0038] in, Let be the time series correlation coefficient between the i-th sensing node and the j-th control node. Let be the parameter value of the i-th sensing node in the m-th time slice. Let be the mean of the parameters of the i-th sensing node across all time slices. Let be the parameter value of the j-th control node in the m-th time slice. Let M be the average parameter value of the j-th control node across all time slices, and M be the total number of time slices.

[0039] The cross-domain mapping matrix is ​​constructed based on a weighted sum of spatial distance correlation coefficients and time series correlation coefficients. The weighting coefficients are set according to scenario requirements; for example, in highly dynamic scenarios, the weight of the time series correlation coefficient can be increased, while in static scenarios, the weight of the spatial distance correlation coefficient can be increased. In the final cross-domain mapping matrix, each element corresponds to the mapping weight from the sensing node to the control node. Through this matrix, the sensing information in the unified environmental sensing topology network can be directly mapped to the control parameters in the lighting motion control topology model, achieving rapid and accurate conversion from sensing information to control commands, avoiding the accumulation of biases caused by multiple data processing steps in existing technologies. During the bidirectional topology mapping process, a reverse mapping from the control node to the sensing node is simultaneously implemented to verify the rationality of the mapping results. When the deviation between the reverse mapping result and the original sensing node exceeds a reasonable range, the mapping matrix is ​​initially corrected to ensure the accuracy of the mapping logic.

[0040] In one specific embodiment, please refer to Figure 5 In S4, the driving signals include: motor stepping pulse signals and gimbal rotation angle control signals. During the adjustment process, the convergence adjustment and optical axis pointing adjustment of the illumination beam are performed simultaneously to keep the center position of the illumination spot coincide with the spatial topological center point of the dynamic target.

[0041] Specifically, the motor stepping pulse signal corresponds to the stepper motor drive of the lamp's walking mechanism and the pan-tilt rotation mechanism. The frequency of the pulse signal corresponds to the motor's rotation speed, and the number of pulse signals corresponds to the motor's rotation angle, which is then converted into the lamp's translational displacement and the pan-tilt rotation angle. The output of the pulse signal is completed by the motor driver chip, which converts the digital parameters in the control command into pulse signals that the motor can recognize, ensuring the consistency between the motor's execution action and the control command. The pan-tilt rotation angle control signal corresponds to the pan-tilt's horizontal and pitch rotation degrees of freedom, and is output using PWM control signals. The pan-tilt horizontal rotation angle has a linear mapping relationship with the PWM duty cycle, with 0° corresponding to a 5% duty cycle and 360° corresponding to a 95% duty cycle; the pan-tilt rotation angle also has a linear mapping relationship with the PWM duty cycle, with -90° corresponding to a 5% duty cycle and 90° corresponding to a 95% duty cycle. Closed-loop feedback through the encoder ensures the consistency between the pan-tilt rotation angle and the control command, eliminating angle deviations during the pan-tilt rotation process. According to the backpropagation algorithm, the convergence of the light beam is adjusted by controlling the relative position of the optical lens group of the lamp. The displacement of the lens group is driven by a micro stepper motor. The nonlinear relationship between the lens displacement and the beam divergence angle is as follows:

[0042] in, This represents the displacement of the lens group relative to the reference position, in mm, with a value range of 0~20mm; The divergence angle of the illumination beam, measured in degrees, is determined by adjusting the position of the lens group according to changes in the illumination distance. This alters the divergence angle, ensuring the size of the illumination spot matches the size of the dynamic target and guarantees adequate illumination in the target area. The optical axis pointing is adjusted via the horizontal and vertical rotation of the pan-tilt unit. The optical axis pointing directly corresponds to the center position of the illumination spot. During adjustment, based on control parameters obtained through cross-domain mapping, optical axis pointing and beam convergence adjustments are performed simultaneously. This ensures the center position of the illumination spot always coincides with the spatial topological center point of the dynamic target, achieving precise tracking illumination of the dynamic target, preventing the illumination spot from deviating from the target area, and eliminating illumination pointing jitter during movement, thus improving illumination stability.

[0043] In one specific embodiment, please refer to Figure 5 In S5, the iterative optimization of topology reconstruction rules and mapping logic includes: defining the actual running topology state as the observation vector, defining the ideal state of the environment-aware unified topology network as the target vector, calculating the Euclidean distance error between the observation vector and the target vector, and adjusting the weight coefficients of the cross-domain mapping matrix through the backpropagation algorithm when the Euclidean distance error exceeds a preset threshold.

[0044] Specifically, the actual operating topology is acquired through attitude sensors, position sensors, and encoders installed on the luminaires. This includes parameters such as the actual displacement of the luminaire, the actual rotation angle of the pan-tilt unit, and the actual position of the illumination spot. These parameters are arranged in a fixed order to form an observation vector, which fully characterizes the actual operating state of the luminaires. The target vector is composed of topological reference parameters of dynamic targets in the unified environmental perception topology network. This includes parameters such as the target position, target illumination angle, and target spot size of the dynamic target. Its dimension is completely consistent with the observation vector, ensuring comparability and providing a unified benchmark for error calculation.

[0045] The formula for calculating the Euclidean distance error is:

[0046] Where e is the Euclidean distance error between the observed vector and the target vector. Let be the parameter value of the k-th dimension of the observed vector. is the parameter value of the k-th dimension of the target vector, and N is the total number of dimensions of the vector.

[0047] The preset threshold is set according to the control precision requirements of the lighting fixtures. A smaller threshold can be set for high-precision lighting needs in emergency rescue scenarios, while a relatively larger threshold can be set for lighting needs in ordinary inspection scenarios, adapting to the control requirements of different scenarios. When the Euclidean distance error exceeds the preset threshold, it indicates a deviation between the current mapping logic and the topology reconstruction rules, making precise motion control impossible. In this case, the backpropagation algorithm is used, with the Euclidean distance error as the loss function, to adjust the weight coefficients of the cross-domain mapping matrix. During the adjustment process, the weight coefficients are updated along the gradient descent direction of the loss function, gradually converging the loss function to within the preset threshold range. Simultaneously, the topology reconstruction rules are optimized according to the source of the error. For example, when the error originates from the calculation deviation of the spatiotemporal offset, the kernel function parameters of the spatiotemporal smoothing algorithm are adjusted to correct the logical correlation strength between nodes, making the reconstructed unified topology network for environmental perception more closely match the actual environmental state. Through this iterative optimization process, closed-loop feedback of the control link is achieved, continuously correcting deviations in the control process, avoiding the accumulation of deviations, and continuously improving the motion control precision and lighting stability of the lighting fixtures as they operate, adapting to dynamic changes in different scenarios.

[0048] Please see Figure 6 The present invention also provides a motion control system for mobile lighting fixtures based on environmental perception, comprising: The topology feature extraction module is used to extract independent spatiotemporal topology features from environmental perception data, transforming environmental perception data of various dimensions into corresponding topology feature nodes. The spatiotemporal topology reconstruction module is used to reconstruct the topology feature nodes in the whole domain using the spatial distribution of dynamic targets as the topology benchmark, and to build a unified topology network for environmental perception. The bidirectional topology mapping module is used to establish a lighting motion control topology model and perform bidirectional topology mapping between the unified environmental perception topology network and the lighting motion control topology model. The motion command solution module is used to solve the motion control logic by matching topology nodes and generate motion control commands. The execution drive module is used to convert the motion control command into a drive signal to drive the lamp to complete the adjustment of the motion trajectory and lighting direction; The topology status acquisition module is used to acquire the actual operating topology status after the lamps have moved. The iterative optimization module is used to iteratively optimize the topology reconstruction rules and mapping logic based on the matching degree between the actual running topology state and the environment-aware unified topology network. The global loop control module is used to control the above modules to continuously execute the process of processing the environmental perception data and generating motion control commands.

[0049] Specifically, the topology feature extraction module is housed in an embedded data processing chip, connected to various environmental sensing components via a communication bus. It receives multi-source heterogeneous data collected by the sensing components, runs feature extraction algorithms to generate topology feature nodes, and achieves standardized processing of the sensing data. The spatiotemporal topology reconstruction module is connected to the topology feature extraction module via an on-chip bus, receives the generated topology feature nodes, runs a topology reconstruction algorithm to construct a unified environmental sensing topology network, and achieves a unified spatiotemporal reference for multi-source sensing data. The bidirectional topology mapping module is connected to the spatiotemporal topology reconstruction module, receives the constructed unified environmental sensing topology network, and pre-stores relevant parameters of the lighting fixture motion control topology model. It completes a bidirectional mapping between the two topologies, establishing a correlation between sensing information and control parameters. The motion command calculation module is connected to the bidirectional topology mapping module, calculates the control logic based on the mapping results, generates standardized motion control commands, and realizes the conversion of sensing information into control commands. The execution drive module is housed in a motor drive circuit and an optical adjustment drive circuit, connected to the motion command calculation module. It receives motion control commands, converts them into corresponding drive signals, and outputs them to the lighting fixture's motion execution mechanism and optical adjustment mechanism to complete the lighting fixture's motion and illumination adjustment. The hardware carrier of the topology status acquisition module consists of position sensors, attitude sensors, and encoders installed on each actuator of the luminaire. Connected to the iterative optimization module, it acquires and transmits the actual operating topology status of the luminaire, providing data support for closed-loop optimization. The iterative optimization module connects to the spatiotemporal topology reconstruction module and the bidirectional topology mapping module, respectively. Based on the transmitted actual operating topology status, it iteratively optimizes the topology reconstruction rules and mapping logic, continuously improving control accuracy. The global loop control module connects to all functional modules and, according to a preset control cycle, schedules each module to complete its corresponding function sequentially, achieving continuous loop execution of the entire control process. This ensures that the mobile luminaire maintains accurate and stable illumination of the dynamic target throughout the entire operation.

[0050] In summary, this invention completes the entire process of transforming environmental perception information into lighting fixture motion control through topology processing. It transforms multi-source heterogeneous environmental perception data into topological feature nodes with independent spatiotemporal attributes. Based on the spatial distribution of dynamic targets, it completes full-domain spatiotemporal topology reconstruction, constructs a unified environmental perception topology network, unifies the spatiotemporal reference of multi-source perception data, eliminates the inherent time difference and spatial offset during data acquisition from different perception components, and solves the problem of implicit deviations in motion control caused by the spatiotemporal mismatch of the perception data itself. Through bidirectional topology mapping between the unified environmental perception topology network and the lighting fixture motion control topology model, it completes the accurate calculation and transformation of perception information into motion control commands, synchronously collects the actual operating status of the lighting fixture and transmits it back to the front end of the control link, and iteratively optimizes the topology reconstruction rules and mapping logic based on the matching degree between the operating status and the ideal state, avoiding the continuous accumulation of deviations in the control process, improving the fit between the lighting fixture motion trajectory and the target lighting area, ensuring the continuous stability of lighting direction in dynamic operation scenarios, providing mobile lighting fixtures with complete motion control logic with spatiotemporal consistency, and realizing continuous and stable lighting control of targets in dynamic scenarios.

[0051] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0052] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for controlling the motion of mobile lighting fixtures based on environmental perception, characterized in that, include: S1. Extract independent spatiotemporal topological features from environmental perception data and transform environmental perception data of each dimension into corresponding topological feature nodes; S2. Using the spatial distribution of dynamic targets as the topological benchmark, the topological feature nodes are reconstructed in a global spatiotemporal manner to construct a unified topological network for environmental perception. S3. Establish a lighting fixture motion control topology model, perform bidirectional topology mapping between the unified environmental perception topology network and the lighting fixture motion control topology model, complete the motion control logic calculation through topology node matching, and generate motion control commands; S4. Convert the motion control command into a drive signal to drive the lamp to adjust the motion trajectory and lighting direction; S5. Collect the actual operating topology state after the lamp moves, and send the actual operating topology state back to the topology reconstruction stage. Based on the matching degree between the actual operating topology state and the unified topology network for environmental perception, iteratively optimize the topology reconstruction rules and mapping logic.

2. The motion control method for mobile lighting fixtures based on environmental perception according to claim 1, characterized in that, In S1, the extraction of independent spatiotemporal topological features from environmental perception data includes: acquiring multi-source heterogeneous data using image perception components, laser scanning components, airflow perception components, and inertial measurement components; analyzing the rate of change of multi-source heterogeneous data on the time axis and the distribution density in the three-dimensional spatial coordinate system using feature dimensionality reduction algorithms; defining the timestamp identifier and spatial pose parameters of each data cluster; and forming topological feature nodes with independent spatiotemporal attributes.

3. The motion control method for mobile lighting fixtures based on environmental perception according to claim 2, characterized in that, The topological feature nodes include: time identification information, spatial location vector information, and environmental energy distribution intensity information. During the extraction process, the acquisition frequency of the sensing data in each dimension is normalized to eliminate the difference in data sampling interval between different sensors, so that each topological feature node has comparable attribute parameters under a unified time slice.

4. The motion control method for mobile lighting fixtures based on environmental perception according to claim 1, characterized in that, In S2, using the spatial distribution of the dynamic target as a topological reference includes: identifying the geometric center and motion vector of the dynamic target in the environment, mapping the motion path of the dynamic target into a three-dimensional topological skeleton, constructing a relative coordinate system with the three-dimensional topological skeleton as the origin, and mapping the topological feature nodes to the relative coordinate system.

5. The motion control method for mobile lighting fixtures based on environmental perception according to claim 1, characterized in that, In S2, the global spatiotemporal topology reconstruction of the topological feature nodes includes: calculating the spatiotemporal offset of each topological feature node relative to the dynamic target topological benchmark, correcting the logical association strength between each node through a spatiotemporal smoothing algorithm, eliminating abnormal nodes with spatiotemporal heterogeneity, and establishing a unified environmental perception topology network describing the environmental perception state and the logical association between the dynamic target.

6. The motion control method for mobile lighting fixtures based on environmental perception according to claim 1, characterized in that, In S3, the establishment of the lamp motion control topology model includes: constructing a motion control topology model that reflects the lamp's execution capability based on the degree-of-freedom parameters of the lamp's motion mechanism, the motor response curve, and the lighting spot diffusion model. The motion control topology model consists of multiple control nodes representing displacement, rotational speed, and pitch angle.

7. The motion control method for mobile lighting fixtures based on environmental perception according to claim 1, characterized in that, In S3, the step of performing bidirectional topology mapping between the unified environmental perception topology network and the lighting motion control topology model includes: calculating the spatial distance correlation coefficient and time series correlation coefficient between the sensing nodes in the unified environmental perception topology network and the control nodes in the motion control topology model, establishing a cross-domain mapping matrix, and realizing real-time mapping of sensing information to control commands through the cross-domain mapping matrix.

8. The motion control method for mobile lighting fixtures based on environmental perception according to claim 1, characterized in that, In S4, the driving signal includes: motor stepping pulse signal and gimbal rotation angle control signal. During the adjustment process, the convergence adjustment and optical axis pointing adjustment of the illumination beam are performed simultaneously to keep the center position of the illumination spot coincide with the spatial topological center point of the dynamic target.

9. The motion control method for mobile lighting fixtures based on environmental perception according to claim 1, characterized in that, In S5, the iterative optimization topology reconstruction rules and mapping logic include: defining the actual running topology state as the observation vector, defining the ideal state of the environment-aware unified topology network as the target vector, calculating the Euclidean distance error between the observation vector and the target vector, and adjusting the weight coefficients of the cross-domain mapping matrix through a backpropagation algorithm when the Euclidean distance error exceeds a preset threshold.

10. A motion control system for mobile lighting fixtures based on environmental perception, used to execute the motion control method for mobile lighting fixtures based on environmental perception as described in any one of claims 1-9, characterized in that, include: The topology feature extraction module is used to extract independent spatiotemporal topology features from environmental perception data, transforming environmental perception data of various dimensions into corresponding topology feature nodes. The spatiotemporal topology reconstruction module is used to reconstruct the topology feature nodes in the whole domain using the spatial distribution of dynamic targets as the topology benchmark, and to build a unified topology network for environmental perception. The bidirectional topology mapping module is used to establish a lighting motion control topology model and perform bidirectional topology mapping between the unified environmental perception topology network and the lighting motion control topology model. The motion command solution module is used to solve the motion control logic by matching topology nodes and generate motion control commands. The execution drive module is used to convert the motion control command into a drive signal to drive the lamp to complete the adjustment of the motion trajectory and lighting direction; The topology status acquisition module is used to acquire the actual operating topology status after the lamps have moved. The iterative optimization module is used to iteratively optimize the topology reconstruction rules and mapping logic based on the matching degree between the actual running topology state and the environment-aware unified topology network. The global loop control module is used to control the above modules to continuously execute the process of processing the environmental perception data and generating motion control commands.

Images