Vehicle trajectory prediction method, electronic equipment and vehicle

Abstract

The invention provides a vehicle track prediction method, electronic equipment and a vehicle, and is applied to the technical field of automatic driving, and the method comprises the steps: obtaining the historical pose data of a vehicle within a preset duration before the current moment, and calculating and determining the motion change data within the preset duration according to the historical pose data of the vehicle. Calculating and determining a fusion weight coefficient according to the motion change data, and predicting a vehicle trajectory based on the fusion weight coefficient and a pre-constructed fusion model to obtain a vehicle predicted trajectory; wherein the fusion model is a model obtained by fusing a linear model and a steering model. When the vehicle trajectory prediction is carried out, the fusion model fusing the linear model and the steering model is adopted, the complex motion of the vehicle can be better adapted through the fusion model, and the problem that the vehicle trajectory is not matched with the actual vehicle motion trend due to the fact that only a single model is adopted for prediction is avoided.

Description

Technical Field

[0001] This application relates to the field of autonomous driving technology, and in particular to a vehicle trajectory prediction method, electronic device, and vehicle. Background Technology

[0002] Autonomous driving technology offers significant advantages such as improved traffic safety and reduced traffic congestion, leading to increasing attention. In real-world driving scenarios, autonomous vehicles require reliable predictions of their future trajectories to inform safe and efficient decision-making and planning. However, existing vehicle trajectory prediction methods rely on a single prediction model, resulting in a mismatch between the prediction and the actual vehicle's movement trends. Summary of the Invention

[0003] In view of this, the purpose of this application is to propose a vehicle trajectory prediction method, electronic device and vehicle to solve the problem of mismatch between predicted trajectory and vehicle motion trend caused by a single prediction model.

[0004] To achieve the above objectives, the first aspect of this application provides a vehicle trajectory prediction method, comprising: Obtain historical vehicle position and pose data within a preset time period prior to the current moment; Based on the vehicle's historical posture data, motion change data within a preset time period is calculated and determined. The fusion weighting coefficients are calculated and determined based on the motion change data. Based on the fusion weight coefficients and the pre-built fusion model, trajectory prediction is performed to obtain the vehicle's predicted trajectory. The fusion model is the model obtained by fusing the straight line model and the turning model.

[0005] Optionally, the motion change data includes the rate of change of curvature and the variance of acceleration; the fusion weighting coefficient includes a first weighting coefficient and a second weighting coefficient; the calculation of the fusion weighting coefficient based on the motion change data includes: calculating and determining the first weighting coefficient based on the rate of change of curvature; and calculating and determining the second weighting coefficient based on the variance of acceleration.

[0006] Optionally, the step of predicting the vehicle trajectory based on the fusion weight coefficients and the pre-built fusion model to obtain the predicted vehicle trajectory includes: The vehicle trajectory is predicted using the linear model to obtain a first state vector; The vehicle trajectory is predicted using the steering model to obtain the second state vector; The predicted vehicle trajectory is obtained based on the fusion weight coefficient, the first state vector, and the second state vector.

[0007] Optionally, the fusion weighting coefficients include a first weighting coefficient and a second weighting coefficient; obtaining the vehicle predicted trajectory based on the fusion weighting coefficients, the first state vector, and the second state vector includes: Calculate the product of the first weight coefficient and the first state vector, and use it as the first target state vector; Calculate the product of the second weighting coefficient and the second state vector, and use it as the second target state vector; The first target state vector and the second target state vector are fused to obtain the vehicle predicted trajectory.

[0008] Optionally, before calculating and determining the fusion weighting coefficients based on the motion change data, the following steps are included: The vehicle's motion type is determined based on the motion change data; In response to the vehicle's motion type being complex motion, a fusion weighting coefficient is calculated and determined based on the motion change data.

[0009] Optionally, determining the vehicle's motion type based on the motion change data includes: Based on the motion change data, determine whether the vehicle's motion type is linear motion; In response to determining that the motion type is not linear motion, the motion type of the vehicle is determined to be steering motion based on the motion change data; In response to determining that the motion type is not a steering motion, the motion type of the vehicle is determined to be a complex motion.

[0010] Optionally, the motion change data includes the rate of change of curvature and the variance of acceleration; determining whether the vehicle's motion type is linear motion based on the motion change data includes: In response to the number of frames in which the rate of curvature change is less than a first preset threshold and the acceleration variance is less than a second preset threshold within a preset time period reaching a first preset number of frames, the motion type of the vehicle is determined to be linear motion. Determining whether the vehicle's motion type is steering motion based on the motion change data includes: In response to the fact that within a preset time period, the number of frames in which the rate of curvature change is less than a first preset threshold and the acceleration variance is less than a second preset threshold does not reach a first preset number of frames, and the number of frames in which the rate of curvature change is less than a third preset threshold reaches a second preset number of frames, the motion type of the vehicle is determined to be steering motion; wherein the third preset threshold is greater than the first preset threshold.

[0011] Optionally, the motion change data includes vehicle speed variance; before calculating and determining the fusion weighting coefficients based on the motion change data, the following steps are included: The current time step is calculated using a dynamic step size algorithm based on the vehicle speed variance. The fusion weight coefficients are calculated and determined based on the motion change data according to the current time step.

[0012] Based on the same inventive concept, a second aspect of this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor, when executing the computer program, implements the method described in the first aspect.

[0013] Based on the same inventive concept, a third aspect of this application also provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method described in the first aspect.

[0014] As can be seen from the above, the vehicle trajectory prediction method, electronic device, and vehicle provided in this application include: acquiring historical vehicle pose data within a preset time period prior to the current moment, and calculating motion change data within the preset time period based on the historical vehicle pose data. Calculating motion change data from historical pose data preserves the continuity of vehicle motion, thus improving the accuracy of vehicle trajectory prediction. A fusion weight coefficient is calculated based on the motion change data, and the vehicle trajectory is predicted based on the fusion weight coefficient and a pre-built fusion model to obtain the predicted vehicle trajectory. The fusion model is a model obtained by fusing a straight-line model and a steering model. This application uses a fusion model that integrates a straight-line model and a steering model when predicting vehicle trajectory. This fusion model better adapts to the complex motion of the vehicle, avoiding the problem of mismatch between the vehicle trajectory and the actual vehicle motion trend caused by using only a single model for prediction. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the vehicle trajectory prediction method according to an embodiment of this application; Figure 2 This is a schematic diagram of the vehicle trajectory prediction device according to an embodiment of this application; Figure 3 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of this application. Detailed Implementation To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0017] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this application should have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0018] As described in the background section, vehicle trajectory prediction can provide reliable future motion trajectories for autonomous driving, providing a basis for safe and efficient decision-making and planning for autonomous vehicles. However, the main reasons for the inaccuracy of existing vehicle trajectory prediction methods include the following: 1. The prediction model is too simplistic. Traditional trajectory prediction methods often suffer from the drawback of static model selection. Their core decisions rely solely on the target's state information at a single instantaneous time slice (such as current velocity, instantaneous acceleration, or current curvature). This approach neglects in-depth analysis of historical motion patterns and trends. The lack of historical motion pattern analysis in the decision-making logic results in a model selection that lacks foresight and contextual awareness, only reflecting the current state and failing to predict trends.

[0019] Lacking historical context, the system cannot determine whether the current state is at the midpoint of a stable motion phase or a brief moment in a complex maneuver. For example, a high instantaneous curvature might represent a sustained turn or simply a slight and brief correction made by the vehicle to avoid a pothole. Traditional trajectory prediction methods cannot distinguish between these two situations, mechanically selecting the same model (such as a steering model) for both, resulting in a severe mismatch between the prediction model and the actual motion trend and intent. Especially when long-term predictions are required, the predicted trajectory can quickly deviate from the actual path, significantly reducing reliability.

[0020] 2. Critical state prediction of jump Current prediction models mainly include linear models and steering models. In the linear model, it is assumed that the vehicle maintains its current speed and heading angle while moving in a uniform straight line. The vehicle's motion in the two-dimensional plane (XY coordinate system) can be decomposed into two independent, uniform linear motions. The state vector is defined as [px, py, vx, vy]. T Let px and py represent the vehicle's position coordinates in the current coordinate system. Let vx and vy represent the vehicle's velocity components in the x and y directions, respectively. Assuming a prediction time interval of Δt, the state vector for the next moment is predicted based on the current state vector. The straight-line model is a linear model suitable for highway cruising scenarios and short-time predictions in urban roads. While the straight-line model is simple and efficient, it cannot predict turns and accumulates errors quickly. In the steering model, it is assumed that the vehicle moves with a constant turning rate and a constant speed. Compared to the straight-line model, the steering model introduces the vehicle's heading angle and turning rate. The state vector is defined as [px, py, v, θ, ω]. T px and py represent the vehicle's position coordinates in the current coordinate system, v represents the magnitude of the vehicle's resultant velocity, θ represents the vehicle's heading angle, and ω represents the vehicle's turning rate. The state vector for the next moment is predicted based on the current state vector. The steering model is a nonlinear model, suitable for cornering scenarios. Due to the assumption of a constant turning rate, the steering model cannot accurately predict complex movements such as entering a corner, exiting a corner, or making S-shaped lane changes.

[0021] The switching between the straight-line model and the steering model is based on a model switching mechanism with fixed rule thresholds (such as velocity thresholds and curvature thresholds). In critical states, there is a problem of discontinuous prediction output. When the target's motion parameters (such as lateral velocity and curvature) change abruptly near the preset threshold boundaries, the system will frequently switch between the "straight-line model" and the "steering model".

[0022] This switching directly leads to jumps, jitters, or obvious broken lines in the generated predicted trajectory. For example, if a car is slowly entering a curve with its curvature value fluctuating slightly around a threshold, the predicted trajectory will oscillate between a straight line and a circular arc with a fixed curvature, which is completely inconsistent with the smooth and continuous actual motion trajectory of the vehicle. This instability of the predicted trajectory will greatly interfere with the downstream decision-making and planning modules, seriously affecting the comfort of the ride, the safety of the system, and the rationality of the decisions. This model switching mechanism lacks a smooth transition and does not quantitatively model the continuity and stability of historical motion, processing what should be continuous physical motion with discrete decision logic, resulting in discontinuous output.

[0023] 3. Failure due to complex motion When faced with highly dynamic and complex driving scenarios (such as emergency obstacle avoidance, S-shaped lane changes, weaving, or sudden braking), single-model or fixed-combination-model approaches will completely fail. These complex movements are the result of the simultaneous and coordinated effects of multiple motion features.

[0024] For example, an S-shaped lane change involves both continuous changes in curvature (from turning left to turning right) and dramatic fluctuations in acceleration (potentially accompanied by acceleration or deceleration during the lane change). Fixed models cannot simultaneously capture these two variations: constant yaw rate models cannot handle changes in curvature direction; uniform speed models cannot handle fluctuations in acceleration. If a single model is used for fitting, the resulting predicted trajectory will differ significantly from the actual path.

[0025] Traditional vehicle trajectory prediction lacks collaborative analysis methods that can integrate multi-dimensional historical motion characteristics (such as acceleration fluctuations and curvature change trends). It can only analyze linear motion and steering motion, and cannot analyze complex motion.

[0026] In view of this, this application proposes a vehicle trajectory prediction method. It uses historical pose data to calculate motion change data and predicts the vehicle trajectory through a fusion model, solving the problem of model simplification and ensuring that the predicted vehicle trajectory matches the vehicle's motion trend. Simultaneously, it integrates a straight-line model and a steering model, allowing the outputs of the straight-line model and the steering model to work synergistically to obtain the final predicted vehicle trajectory, making it more suitable for complex motions. A fusion weighting coefficient is introduced to eliminate prediction jumps at critical states, ensuring the continuity and smoothness of the vehicle's motion trend.

[0027] The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0028] This application proposes a vehicle trajectory prediction method, referencing... Figure 1 This includes the following steps: Step 102: Obtain the vehicle's historical pose data within a preset time period prior to the current moment.

[0029] Specifically, vehicle pose data includes at least vehicle position, speed, and acceleration. Vehicle pose data can be acquired through various sensors mounted on the vehicle. Historical pose data refers to vehicle pose data over a period of time preceding the current moment. This historical pose data effectively reflects the vehicle's motion trend over a given period, providing an accurate data foundation for vehicle trajectory prediction.

[0030] The preset duration is determined based on actual needs. A longer preset duration results in a larger amount of historical pose data, allowing for a more accurate prediction of the vehicle trajectory, but this increases computational overhead and slows the prediction response. Conversely, a shorter preset duration results in a smaller amount of historical pose data, leading to a faster prediction response and lower computational overhead. However, the accuracy of the vehicle trajectory prediction may decrease. Therefore, an appropriate preset duration should be selected based on the actual vehicle movement.

[0031] Step 104: Calculate and determine the motion change data within a preset time period based on the vehicle's historical pose data.

[0032] Specifically, motion change data reflects the vehicle's motion trend over a preset time period. This can be achieved through metrics such as vehicle speed variance, acceleration variance, and rate of change of curvature. Vehicle speed variance characterizes the stability of the vehicle's longitudinal motion. A smaller vehicle speed variance indicates that the vehicle maintains a relatively stable speed, with a highly predictable motion trend; the vehicle is likely to continue its current straight-line or gentle curve. A larger vehicle speed variance indicates frequent speed fluctuations, potentially involving continuous acceleration and deceleration, and the driving scenario is usually more complex. Acceleration variance characterizes the abruptness and aggressiveness of the vehicle's driving force. A smaller acceleration variance indicates smooth acceleration and linear power output. The driver's driving style is stable, and the motion trend is smooth and easily predictable. A larger acceleration variance indicates drastic changes in acceleration, with abrupt changes in the vehicle's motion trend. For example, during an S-shaped lane change, there is usually rapid switching between accelerator and brake, resulting in high-frequency fluctuations in acceleration. The rate of change of curvature reflects the continuity of steering operations. A low rate of curvature change indicates a smooth and continuous steering action, suggesting the vehicle is steadily performing the steering maneuver and its tendency to continue along the current curve. A high rate of curvature change indicates a sharp, sudden, or discontinuous steering action. When changing lanes, the curvature initially increases to one side (initiating a turn), quickly returns to zero (entering the new lane), and may finally increase slightly to the other side (straightening the steering wheel). This process generates a brief but high-amplitude pulse in the rate of curvature change. During emergency obstacle avoidance, a sudden and large steering input will produce an extremely high rate of curvature change, indicating that the vehicle is deviating from its original path and the trend is changing drastically. When turning at an intersection, the curvature increases sharply from zero, indicating that the vehicle is about to execute a clear turning intention, rather than going straight.

[0033] By comprehensively analyzing the above motion change data, we can conduct in-depth analysis of the driver's intentions, behavioral continuity, and motion abrupt changes, thereby accurately predicting the vehicle's driving trajectory, which is more in line with the actual driving logic of the vehicle.

[0034] Step 106: Calculate and determine the fusion weight coefficient based on the motion change data.

[0035] Specifically, the fusion weighting coefficients include the weighting coefficients assigned to different prediction models. Since the fusion weighting coefficients are calculated based on motion change data within a preset time period, they are continuous values ​​with consistency. When the vehicle's motion state approaches a critical point (such as changing from linear motion to curvilinear motion), the fusion weighting coefficients transition smoothly, thus ensuring the continuity and smoothness of the final trajectory prediction and avoiding trajectory jumps.

[0036] Step 108: Based on the fusion weight coefficients and the pre-built fusion model, perform trajectory prediction to obtain the vehicle prediction trajectory; wherein, the fusion model is the model obtained by fusing the straight line model and the steering model.

[0037] Specifically, in the fusion model, different weight coefficients are assigned to different prediction models. These prediction models include a straight-line model and a steering model. The fusion model combines the trajectories predicted by the straight-line model and the steering model, along with their respective weight coefficients, to obtain the vehicle's predicted trajectory output by the fusion model. The calculated fusion weight coefficients differ depending on the vehicle's motion scenario; that is, the weight coefficients assigned to different prediction models vary, ensuring that the proportions of the trajectories output by different prediction models in the overall vehicle prediction trajectory differ. For example, when the vehicle is traveling in a straight line, the weight coefficient for the steering model decreases, while the weight coefficient for the straight-line model increases. When the vehicle is turning, the weight coefficient for the specialized model increases, while the weight coefficient for the straight-line model decreases. The fusion model can handle various driving scenarios, and compared to a single prediction model, the vehicle prediction trajectory obtained through the fusion model more closely approximates the vehicle's actual motion trend.

[0038] Based on steps 102 to 108 above, the vehicle trajectory prediction method proposed in this embodiment includes: acquiring historical vehicle pose data within a preset time period before the current moment, and calculating motion change data within the preset time period based on the historical vehicle pose data. Calculating motion change data using historical pose data preserves the continuity of vehicle motion, thus improving the accuracy of vehicle trajectory prediction. A fusion weight coefficient is calculated based on the motion change data, and the vehicle trajectory is predicted based on the fusion weight coefficient and a pre-built fusion model to obtain the predicted vehicle trajectory; wherein the fusion model is a model obtained by fusing a straight-line model and a steering model. In predicting vehicle trajectory, this application uses a fusion model that integrates a straight-line model and a steering model. This fusion model better adapts to the complex motion of the vehicle, avoiding the problem of mismatch between the vehicle trajectory and the actual vehicle motion trend caused by using only a single model for prediction.

[0039] In some embodiments, the motion change data includes the rate of change of curvature and the variance of acceleration; the fusion weighting coefficient includes a first weighting coefficient and a second weighting coefficient; the calculation of the fusion weighting coefficient based on the motion change data includes: calculating and determining the first weighting coefficient based on the rate of change of curvature; and calculating and determining the second weighting coefficient based on the variance of acceleration.

[0040] Specifically, the fusion model is a fusion of the linear model and the steering model. Correspondingly, the fusion weighting coefficients include a first weighting coefficient and a second weighting coefficient. The first weighting coefficient is the weighting coefficient assigned to the linear model, and the second weighting coefficient is the weighting coefficient assigned to the steering model. This is based on the rate of change of curvature. Calculate and determine the first weighting coefficient The calculation method is as follows: (1) Among them, the rate of change of curvature It is the average value of the K-frame curvature change rate calculated based on the vehicle's historical pose data, or a stable value obtained after filtering the K-frame curvature change rate. As can be seen from equation (1), the curvature change rate... With the first weighting coefficient They are negatively correlated. (Rate of change of curvature) When the value is large (due to turning or complex motion), the first weighting coefficient When the curvature changes near zero, the linear model has a lower weight, weakening its predictive role in the fusion model. When the value is small (linear motion), the first weighting coefficient When the value is close to 1, the linear model has a higher weight, which strengthens its predictive role in the fusion model.

[0041] Based on acceleration variance Calculate and determine the second weighting coefficient The calculation method is as follows: (2) Among them, acceleration variance It is the mean of the K-frame acceleration variance calculated based on the vehicle's historical pose data, or a stable value obtained after filtering the K-frame acceleration variance. This represents the maximum value of the acceleration variance in K frames. From equation (2), it can be seen that the acceleration variance... With the second weighting coefficient They are positively correlated. When When the acceleration is relatively small (stable acceleration), the second weighting coefficient When the value is close to zero, the steering model has a low weight, weakening its predictive role in the fusion model. When the acceleration is large (rapid changes), the second weighting coefficient When the value is close to 1, the steering model has a higher weight, which strengthens the predictive role of the steering model in the fusion model.

[0042] It should be noted that the rate of change of curvature can be replaced by the variance of the radius of the circle obtained by fitting historical trajectory points. The radius variance characterizes the fluctuation of the radius, i.e., the change in curvature. The acceleration variance can be replaced by jerk. Jerk is the rate of change of acceleration over time, describing how quickly acceleration changes. Rapid acceleration or braking produces high jerk, which occupants will feel significant discomfort. At low jerk, the vehicle travels more smoothly.

[0043] By using the calculation method of the first weighting coefficient and the second weighting coefficient in this embodiment, the first weighting coefficient is assigned to the straight line model and the second weighting coefficient is assigned to the steering model. This enables the accurate adjustment of the output trajectory weights of the straight line model and the steering model under different driving scenarios, so as to match the actual movement trend of the vehicle and improve the accuracy of vehicle trajectory prediction.

[0044] In some embodiments, predicting the vehicle trajectory based on the fusion weight coefficients and a pre-built fusion model to obtain the predicted vehicle trajectory includes: The vehicle trajectory is predicted using the linear model to obtain a first state vector; The vehicle trajectory is predicted using the steering model to obtain the second state vector; The predicted vehicle trajectory is obtained based on the fusion weight coefficient, the first state vector, and the second state vector.

[0045] Specifically, the fusion model is obtained by fusing the straight-line model and the steering model. Based on the vehicle's current state vector, the straight-line model predicts the vehicle's future state vector, which is used as the first state vector. Based on the vehicle's current state vector, the steering model predicts the vehicle's future state vector, which is used as the second state vector. After fusing the first and second state vectors using fusion weighting coefficients, the predicted vehicle trajectory is obtained. The predicted vehicle trajectory incorporates both the first and second state vectors, and the proportions of the first and second state vectors in the predicted trajectory are adjusted by the fusion weighting coefficients. The proportions of the first and second state vectors in the predicted vehicle trajectory differ in different driving scenarios. For example, when the vehicle is driving in a straight line, the first state vector has a larger proportion. When the vehicle is turning, the second state vector has a larger proportion. When the vehicle is in a complex driving scenario, the proportions of the first and second state vectors in the predicted vehicle trajectory will be dynamically adjusted accordingly. The state vector includes at least parameters such as vehicle position, speed, and acceleration.

[0046] The vehicle trajectory prediction method in this embodiment achieves the synergistic effect of the straight-line model and the steering model. By fusing the vehicle's future state vectors predicted by the straight-line model and the steering model respectively in a driving scenario, and adjusting the fusion weight coefficients in real time, it can adapt to various driving scenarios, especially the movement of vehicles in complex driving scenarios. This avoids the problem of a single prediction model failing in some driving scenarios, improves the universality of the prediction model, and enhances the accuracy of vehicle trajectory prediction.

[0047] Furthermore, the fusion weighting coefficients include a first weighting coefficient and a second weighting coefficient; obtaining the vehicle prediction trajectory based on the fusion weighting coefficients, the first state vector, and the second state vector includes: Calculate the product of the first weight coefficient and the first state vector, and use it as the first target state vector; Calculate the product of the second weighting coefficient and the second state vector, and use it as the second target state vector; The first target state vector and the second target state vector are fused to obtain the vehicle predicted trajectory.

[0048] Specifically, the fusion model is represented by the following formula: T_final = w1×T1 + w2×T2 (3) Where T1 represents the first state vector, w1 represents the first weight coefficient, T2 represents the second state vector, w2 represents the second weight coefficient, and T_final represents the fusion vector.

[0049] As can be seen from formula (3), based on the current state vector of the vehicle, the future state vector of the vehicle is predicted by the linear model and used as the first state vector T1. The product of the first state vector T1 and the first weight coefficient w1 is used as the first target state vector w1×T1. Based on the current state vector of the vehicle, the future state vector of the vehicle is predicted by the steering model and used as the second state vector T2. The product of the second state vector T2 and the second weight coefficient w2 is used as the second target state vector w2×T2. The sum of the first target state vector w1×T1 and the second target state vector w2×T2 is calculated and used as the fusion vector T_final. That is, the fusion vector is the vector obtained by fusing the first state vector and the second state vector. By using the fusion vector to predict the vehicle trajectory, the problems of trajectory jumps and discontinuities caused by model switching in the prior art can be avoided. The fusion vector can be adapted to any driving scenario of the vehicle. When the driving trajectory of the vehicle changes, the dynamic adjustment of the fusion vector is achieved by calculating the first weight coefficient and the second weight coefficient in real time, so that the predicted vehicle trajectory can conform to the actual movement trend of the vehicle.

[0050] It should be noted that, in addition to the methods described above for calculating and determining the first and second weight coefficients, the first and second weight coefficients can also be predicted using a trained neural network. The motion change data of the vehicle within a preset time period (K frames) is input into the trained neural network, which then outputs the first and second weight coefficients. For example, the neural network architecture can be a combination of a convolutional neural network and a long short-term memory network. Convolutional neural networks excel at extracting local time-series patterns, using one-dimensional convolutional layers sliding along the time dimension to efficiently capture these features. Long short-term memory networks excel at processing long sequences and remembering long-term dependencies.

[0051] When predicting weight coefficients using a neural network, a convolutional layer is first used to extract features from the input sequence (K frames of motion change data), outputting a higher-level feature sequence. This higher-level feature sequence is then input into a long short-term memory network, and finally, a fully connected layer is used to output the first and second weight coefficients.

[0052] When constructing the training set for the neural network, the vehicle's current state vector and its true state vector at future time points are collected. During model training, the current state vector is predicted using both a linear model and a steering model, resulting in a first state vector and a second state vector. These two vectors are then fused to obtain a fused vector. The error between the fused vector and the true state vector is calculated, and the network parameters are updated by minimizing this error, ultimately yielding the trained neural network model. This neural network model can learn more complex weight allocation strategies from the data, leading to higher accuracy in predicting weight coefficients.

[0053] In this embodiment, a first state vector and a second state vector are predicted using a linear model and a steering model, respectively. These two state vectors are then fused to obtain a fused vector. The weights of the first and second state vectors in the fused vector are adjusted using a first weighting coefficient and a second weighting coefficient. The vehicle trajectory predicted using the fused vector matches the actual movement trend of the vehicle, improving the accuracy of trajectory prediction.

[0054] The fusion model in the above embodiments can be adapted to various driving scenarios. However, to further reduce the computational overhead of the fusion model, the prediction model adapted to the vehicle in the current scenario can be determined first. If the current driving scenario is a straight-line motion scenario, only the straight-line model can be matched; if the current driving scenario is a steering motion scenario, only the steering model can be matched; if the current driving scenario is a complex motion scenario, the fusion model can be matched. In this way, the computational resource consumption on the vehicle side can be appropriately reduced. Specific embodiments are described below.

[0055] In some embodiments, before calculating and determining the fusion weighting coefficients based on the motion change data, the method includes: The vehicle's motion type is determined based on the motion change data; In response to the vehicle's motion type being complex motion, a fusion weighting coefficient is calculated and determined based on the motion change data.

[0056] Specifically, after determining the motion change data, the first step is to determine the vehicle's motion type. Different motion types are suited to different prediction models. For example, linear motion is suited to a linear model, steering motion to a steering model, and complex motion to a fusion model.

[0057] Since the fusion model combines the prediction results of the straight-line model and the steering model, calling the fusion model is equivalent to calling both the straight-line model and the steering model simultaneously. This requires real-time calculation of the weighting coefficients for the straight-line model and the steering model, resulting in relatively high computational overhead. In this embodiment, when matching the prediction model based on the motion type, if it is determined that the vehicle's current motion type is complex, the fusion model is then called, and the fusion weighting coefficients are calculated and determined based on the motion change data. If it is determined that the vehicle's current motion type is not complex, only the straight-line model or the steering model needs to be called. When only the straight-line model or the steering model is called, there is no need to call other prediction models or calculate fusion weighting coefficients, reducing the vehicle's computational overhead and improving the efficiency of model prediction.

[0058] Furthermore, determining the vehicle's motion type based on the motion change data includes: Based on the motion change data, determine whether the vehicle's motion type is linear motion; In response to determining that the motion type is not linear motion, the motion type of the vehicle is determined to be steering motion based on the motion change data; In response to determining that the motion type is not a steering motion, the motion type of the vehicle is determined to be a complex motion.

[0059] Specifically, determining the vehicle's motion type is crucial for accurately calling the prediction model and performing precise trajectory prediction. The motion type determination employs a hierarchical, progressive logical decision tree, precisely identifying the vehicle's current motion type based on real-time motion change data.

[0060] When determining the vehicle's motion type, the first step is to determine whether the motion is linear based on the motion change data. If it is not linear, then determine whether it is steering. If it is also not steering, then the vehicle's motion type can be classified as complex motion. When determining the vehicle's motion type based on motion change data, motion change data thresholds corresponding to each motion type can be pre-set. These thresholds are then used to determine whether the motion is linear or steering.

[0061] To determine whether a vehicle is moving in a straight line, its lateral motion data can be used. This data can include lateral acceleration, yaw rate, and curvature. In straight-line motion, the lateral motion data values ​​are typically small. By using the threshold values ​​corresponding to these lateral motion data, it can be determined whether the vehicle is moving in a straight line.

[0062] After ruling out straight-line vehicle movement, it is necessary to further determine whether the vehicle is turning. If the vehicle's lateral motion data remains at a high value, then the vehicle is determined to be turning. For example, if the yaw rate is consistently higher than a preset threshold and fluctuates little, or if the curvature remains near a stable non-zero value, then the vehicle can be determined to be turning.

[0063] After ruling out the possibility of linear motion and steering motion, the vehicle's motion type is determined to be complex motion. The progressive motion type decision method in this embodiment can accurately determine the vehicle's motion type, avoiding redundant judgments during the process. Since linear motion and steering motion have obvious characteristics and are easier to judge, excluding them allows for the determination of complex motion without additional judgment, making the identification of complex motion more convenient and faster.

[0064] Furthermore, the motion change data includes the rate of change of curvature and the variance of acceleration; determining whether the vehicle's motion type is linear motion based on the motion change data includes: In response to the number of frames in which the rate of curvature change is less than a first preset threshold and the acceleration variance is less than a second preset threshold within a preset time period reaching a first preset number of frames, the motion type of the vehicle is determined to be linear motion. Determining whether the vehicle's motion type is steering motion based on the motion change data includes: In response to the fact that within a preset time period, the number of frames in which the rate of curvature change is less than a first preset threshold and the acceleration variance is less than a second preset threshold does not reach a first preset number of frames, and the number of frames in which the rate of curvature change is less than a third preset threshold reaches a second preset number of frames, the motion type of the vehicle is determined to be steering motion; wherein the third preset threshold is greater than the first preset threshold.

[0065] Specifically, when determining whether a vehicle is moving in a straight line or turning, the determination is based on motion change data over a preset time period, rather than instantaneous motion change data. This improves the stability and robustness of motion type determination and avoids frequent fluctuations in motion type. To determine straight-line motion, the system needs to ensure that the number of frames with a curvature change rate less than a first preset threshold reaches a first preset number of frames, and simultaneously, the number of frames with an acceleration variance less than a second preset threshold reaches a first preset number of frames. A curvature change rate less than the first preset threshold indicates a very low and stable rate of change in the vehicle's direction of travel, without any abrupt turning movements. An acceleration variance less than the second preset threshold indicates very smooth acceleration / deceleration behavior, without sudden acceleration or braking. Only when both conditions are met simultaneously can the motion type be determined to be straight-line motion. For example, the preset time period can be K data frames. The first preset threshold is 0.1, the first preset number of frames is N data frames, and the second preset threshold is 0.5. Here, K and N are both positive integers greater than 1, with K being greater than N. In other words, if N out of K data frames have a curvature change rate of less than 0.1 and an acceleration variance of less than 0.5, then the vehicle's motion type is determined to be linear motion.

[0066] When determining steering motion, it's necessary to ascertain that the current motion type is not linear motion. Therefore, the conditions for linear motion cannot be met. Specifically, within a preset time period, the number of frames with a curvature change rate less than a first preset threshold and an acceleration variance less than a second preset threshold does not reach the first preset number of frames. If, however, the number of frames with a curvature change rate less than a third preset threshold reaches the second preset number of frames, it indicates that the vehicle's steering is smooth and stable, not abrupt or drastic, and the vehicle's motion type is determined to be steering motion. For example, the third preset threshold is 0.3, and the second preset number of frames is N data frames. That is, within K data frames, if the number of data frames with a curvature change rate less than 0.1 and an acceleration variance less than 0.5 does not reach N, but the number of data frames with a curvature change rate less than 0.3 reaches N, then the vehicle's motion type is determined to be steering motion.

[0067] When determining complex motion, it is necessary to ascertain that the current motion type is neither linear motion nor steering motion; otherwise, the current motion type is considered complex motion. In other words, the determination of complex motion is aided by the criteria for linear and steering motion. This is because the characteristics of complex motion are difficult to quantify, while the characteristics of linear and steering motion are relatively easy to quantify. Therefore, by excluding linear and steering motion, the vehicle's motion type can be determined to be complex motion.

[0068] In the actual judgment process, the curvature change rate and acceleration variance of K frames are first calculated based on the motion change data. If there are N data frames with a curvature change rate less than 0.1 and an acceleration variance less than 0.5, the motion type is determined to be linear motion. If there are no N data frames with a curvature change rate less than 0.1 and an acceleration variance less than 0.5, but there are N data frames with a curvature change rate less than 0.3, the motion type is determined to be turning motion. If there are no N data frames with a curvature change rate less than 0.3, the motion type is determined to be complex motion.

[0069] This embodiment provides a detailed method for determining linear motion and steering motion. By setting a preset number of frames, the stability of motion type determination is ensured, avoiding short-term jumps in motion type. It captures the overall trend and pattern of motion in the time domain, reducing the impact of jumpy data frames on motion type determination. By introducing multiple preset thresholds, different motion types are accurately distinguished, achieving precise motion type identification. Accurate motion type determination directly improves the accuracy of subsequent prediction model calls and vehicle trajectory prediction.

[0070] When predicting vehicle trajectories, if the vehicle's motion is relatively stable, the number of trajectory predictions can be reduced to lower computational resource costs. If the vehicle's motion is more volatile, the number of trajectory predictions can be increased to improve accuracy.

[0071] In some embodiments, the motion change data includes vehicle speed variance, and before calculating and determining the fusion weighting coefficients based on the motion change data, the following is included: The current time step is calculated using a dynamic step size algorithm based on the vehicle speed variance. The fusion weight coefficients are calculated and determined based on the motion change data according to the current time step.

[0072] Specifically, before calculating and determining the fusion weighting coefficients based on the motion change data, the current time step is first determined based on the motion change data. The time step is the time interval between two adjacent trajectory predictions. The longer the time step, the lower the trajectory prediction frequency; the shorter the time step, the higher the trajectory prediction frequency. The current time step is calculated as follows: (4) in, Indicates the basic step size. Represents the sensitivity coefficient. This represents the mean or filtered value of the variance of the vehicle speeds in K frames. For example, The value is 0.03s, meaning the time interval between predictions of adjacent trajectories is 0.03s. Sensitivity coefficient. The value is 1.2. As can be seen from formula (4), in this embodiment, the prediction or control step size is dynamically adjusted based on the stability of the vehicle speed. It can measure the magnitude of vehicle speed fluctuations within K frames. The larger the value, the more unstable the vehicle's speed, indicating frequent acceleration or deceleration. The vehicle is in a state of rapid and unpredictable change. The smaller the value, the more stable the vehicle speed, approaching a constant speed, indicating a smooth and predictable driving state. When the vehicle speed is stable, Approaching zero. Current time step. Approximately the base step size When the vehicle speed is unstable, As the value increases, the current time step... Smaller than the base step size This is equivalent to shortening the time step and increasing the trajectory prediction frequency. The more unstable the vehicle speed, the shorter the calculated current time step. The smaller.

[0073] It is an adjustable parameter used to control the system's sensitivity to vehicle speed fluctuations. As the value increases, The smaller the value, the weaker the influence of vehicle speed variance on step size, and the lower the sensitivity of vehicle trajectory prediction. The value decreases, making The larger the value, the greater the impact of the vehicle speed variance on the step size, thus increasing the sensitivity of vehicle trajectory prediction.

[0074] This embodiment achieves a balance between trajectory prediction accuracy and computational efficiency through dynamic adjustment of the time step. In complex conditions, a shorter time step results in denser trajectory prediction points and a higher trajectory prediction update frequency, enabling the capture of rapidly changing motion states and timely responses, thus improving vehicle driving safety. In simple conditions, a longer time step reduces the number of trajectory predictions required per unit time, saving computational resources. Simultaneously, the current time step... It changes continuously, rather than in a step-like manner, ensuring the smoothness of vehicle trajectory prediction commands and avoiding control jitter caused by sudden changes in time step.

[0075] It should be noted that the method in this embodiment can be executed by a single device, such as a computer or server. The method can also be applied in a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method in this embodiment, and the multiple devices will interact with each other to complete the method described.

[0076] It should be noted that some embodiments of this application have been described above. In some cases, the actions or steps described in the above embodiments can be performed in a different order than that shown in the above embodiments and the desired result can still be achieved. In addition, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0077] Based on the same inventive concept, corresponding to any of the above embodiments, this application also provides a vehicle trajectory prediction device.

[0078] refer to Figure 2 The vehicle trajectory prediction device includes: The acquisition module 202 is configured to acquire the vehicle's historical pose data within a preset time period prior to the current moment; The determination module 204 is configured to calculate and determine motion change data within a preset time period based on the vehicle's historical pose data; Calculation module 206 is configured to calculate and determine fusion weight coefficients based on the motion change data; The prediction module 208 is configured to perform trajectory prediction based on the fusion weight coefficients and the pre-built fusion model to obtain the vehicle prediction trajectory. The fusion model is the model obtained by fusing the straight line model and the turning model.

[0079] In some embodiments, the motion change data includes the rate of change of curvature and the variance of acceleration; the fusion weighting coefficients include a first weighting coefficient and a second weighting coefficient; the calculation module 206 is further configured to calculate and determine the first weighting coefficient based on the rate of change of curvature; and to calculate and determine the second weighting coefficient based on the variance of acceleration.

[0080] In some embodiments, the prediction module 208 is further configured to predict the vehicle trajectory using the straight line model to obtain a first state vector; predict the vehicle trajectory using the steering model to obtain a second state vector; and obtain the predicted vehicle trajectory based on the fusion weight coefficient, the first state vector, and the second state vector.

[0081] In some embodiments, the fusion weighting coefficients include a first weighting coefficient and a second weighting coefficient; the prediction module 208 is further configured to calculate the product of the first weighting coefficient and the first state vector as a first target state vector; calculate the product of the second weighting coefficient and the second state vector as a second target state vector; and fuse the first target state vector and the second target state vector to obtain the vehicle prediction trajectory.

[0082] In some embodiments, before calculating and determining the fusion weight coefficient based on the motion change data, the determining module 204 is further configured to determine the motion type of the vehicle based on the motion change data; and in response to the motion type of the vehicle being complex motion, to calculate and determine the fusion weight coefficient based on the motion change data.

[0083] In some embodiments, the determining module 204 is further configured to determine whether the motion type of the vehicle is linear motion based on the motion change data; in response to determining that the motion type is not linear motion, to determine whether the motion type of the vehicle is steering motion based on the motion change data; and in response to determining that the motion type is not steering motion, to determine that the motion type of the vehicle is complex motion.

[0084] In some embodiments, the motion change data includes the rate of change of curvature and the variance of acceleration; the determining module 204 is further configured to determine that the motion type of the vehicle is linear motion in response to the number of frames in which the rate of change of curvature is less than a first preset threshold and the variance of acceleration is less than a second preset threshold reaching a first preset number of frames within a preset time period; and to determine that the motion type of the vehicle is steering motion in response to the number of frames in which the rate of change of curvature is less than the first preset threshold and the variance of acceleration is less than the second preset threshold not reaching the first preset number of frames within a preset time period, and the number of frames in which the rate of change of curvature is less than a third preset threshold reaching a second preset number of frames; wherein the third preset threshold is greater than the first preset threshold.

[0085] In some embodiments, the motion change data includes vehicle speed variance; before calculating and determining the fusion weight coefficient based on the motion change data, the calculation module 206 is further configured to calculate the current time step based on the vehicle speed variance using a dynamic step size algorithm; and calculate and determine the fusion weight coefficient based on the motion change data according to the current time step.

[0086] For ease of description, the above devices are described in terms of function, divided into various modules. Of course, in implementing this application, the functions of each module can be implemented in one or more software and / or hardware.

[0087] The apparatus of the above embodiments is used to implement the corresponding vehicle trajectory prediction method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0088] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the vehicle trajectory prediction method described in any of the above embodiments.

[0089] Figure 3 This embodiment illustrates a more specific hardware structure of an electronic device. The device may include a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected internally via the bus 1050.

[0090] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0091] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0092] The input / output interface 1030 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components within the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touchscreens, microphones, various sensors, etc., while output devices may include displays, speakers, vibrators, indicator lights, etc.

[0093] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0094] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0095] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0096] The electronic devices described above are used to implement the corresponding vehicle trajectory prediction methods in any of the foregoing embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0097] Based on the same inventive concept, corresponding to the methods of any of the above embodiments, this application also provides a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the vehicle trajectory prediction method as described in any of the above embodiments.

[0098] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0099] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the vehicle trajectory prediction method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0100] Based on the same concept, corresponding to any of the above embodiments, this application also provides a computer program product, including computer program instructions, which, when run on a computer, cause the computer to perform the method described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0101] It is understood that before using the technical solutions of the various embodiments in this disclosure, users will be informed of the type, scope of use, and usage scenarios of the personal information involved in an appropriate manner, and user authorization will be obtained.

[0102] For example, upon receiving a user's active request, a prompt message is sent to the user to explicitly inform them that the requested operation will require the acquisition and use of the user's personal information. This allows the user to choose, based on the prompt message, whether to provide personal information to the software or hardware such as the electronic device, application, server, or storage medium performing the operations of this disclosed technical solution.

[0103] As an optional but not limited implementation, in response to a user's active request, sending a prompt message to the user can be done via a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" regarding the provision of personal information by the electronic device.

[0104] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this disclosure. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this disclosure.

[0105] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this application is limited to these examples; under the concept of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this application as described above, which are not provided in detail for the sake of brevity.

[0106] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this application, the well-known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided drawings. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this application, and this also takes into account the fact that the details of the implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this application will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuits) have been set forth to describe exemplary embodiments of this application, it will be apparent to those skilled in the art that the embodiments of this application can be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0107] Although this application has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0108] The embodiments of this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of this application.

Claims

1. A vehicle trajectory prediction method, characterized in that, include: Obtain historical vehicle position and pose data within a preset time period prior to the current moment; Based on the vehicle's historical posture data, motion change data within a preset time period is calculated and determined. The fusion weighting coefficients are calculated and determined based on the motion change data. Based on the fusion weight coefficients and the pre-built fusion model, trajectory prediction is performed to obtain the vehicle's predicted trajectory. The fusion model is the model obtained by fusing the straight line model and the turning model.

2. The method according to claim 1, characterized in that, The motion change data includes the rate of change of curvature and the variance of acceleration; the fusion weighting coefficients include a first weighting coefficient and a second weighting coefficient; the calculation of the fusion weighting coefficients based on the motion change data includes: The first weighting coefficient is determined based on the rate of change of curvature. The second weighting coefficient is determined based on the acceleration variance.

3. The method according to claim 1, characterized in that, The process of predicting vehicle trajectories based on the fusion weight coefficients and a pre-built fusion model to obtain the predicted vehicle trajectory includes: The vehicle trajectory is predicted using the linear model to obtain a first state vector; The vehicle trajectory is predicted using the steering model to obtain the second state vector; The predicted vehicle trajectory is obtained based on the fusion weight coefficient, the first state vector, and the second state vector.

4. The method according to claim 3, characterized in that, The fusion weighting coefficients include a first weighting coefficient and a second weighting coefficient; obtaining the vehicle prediction trajectory based on the fusion weighting coefficients, the first state vector, and the second state vector includes: Calculate the product of the first weight coefficient and the first state vector, and use it as the first target state vector; Calculate the product of the second weighting coefficient and the second state vector, and use it as the second target state vector; The first target state vector and the second target state vector are fused to obtain the vehicle predicted trajectory.

5. The method according to claim 1, characterized in that, Before calculating and determining the fusion weighting coefficients based on the motion change data, the following steps are included: The vehicle's motion type is determined based on the motion change data; In response to the vehicle's motion type being complex motion, a fusion weighting coefficient is calculated and determined based on the motion change data.

6. The method according to claim 5, characterized in that, Determining the vehicle's motion type based on the motion change data includes: Based on the motion change data, determine whether the vehicle's motion type is linear motion; In response to determining that the motion type is not linear motion, the motion type of the vehicle is determined to be steering motion based on the motion change data; In response to determining that the motion type is not a steering motion, the motion type of the vehicle is determined to be a complex motion.

7. The method according to claim 6, characterized in that, The motion change data includes the rate of change of curvature and the variance of acceleration; determining whether the vehicle's motion type is linear motion based on the motion change data includes: In response to the number of frames in which the rate of curvature change is less than a first preset threshold and the acceleration variance is less than a second preset threshold within a preset time period reaching a first preset number of frames, the motion type of the vehicle is determined to be linear motion. Determining whether the vehicle's motion type is steering motion based on the motion change data includes: In response to the fact that within a preset time period, the number of frames in which the rate of curvature change is less than a first preset threshold and the acceleration variance is less than a second preset threshold does not reach a first preset number of frames, and the number of frames in which the rate of curvature change is less than a third preset threshold reaches a second preset number of frames, the motion type of the vehicle is determined to be steering motion; wherein the third preset threshold is greater than the first preset threshold.

8. The method according to claim 1, characterized in that, The motion change data includes vehicle speed variance; before calculating and determining the fusion weighting coefficients based on the motion change data, the following steps are included: The current time step is calculated using a dynamic step size algorithm based on the vehicle speed variance. The fusion weight coefficients are calculated and determined based on the motion change data according to the current time step.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1 to 8.

10. A vehicle, characterized in that, The vehicle includes the electronic equipment as described in claim 9.

Images