A method for estimating vehicle sliding parameters

Abstract

The present invention relates to a method for estimating vehicle sliding parameters, and relates to the field of intelligent vehicle technology. It addresses the existing difficulty in predicting vehicle sliding parameters in off-road scenarios. The method includes obtaining point cloud data, image data, and vehicle trajectory data of a target scene; and inputting the target scene point cloud data, image data, and vehicle trajectory data into a pre-trained sliding parameter estimation model to obtain an estimated sliding parameter result.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent vehicles, and in particular to a method for estimating vehicle sliding parameters. Background Art

[0002] Slip and roll are the relative movement of the grounded portion of a vehicle during driving. Slip and roll (collectively referred to as "slip") are common and unavoidable parameters that significantly impact vehicle operation. Therefore, properly estimating a vehicle's slip parameters is crucial for improving autonomous driving safety.

[0003] Sliding parameter estimation refers to the process of quickly estimating sliding parameters by utilizing measurement data from various sensors. This is very helpful in improving vehicle driving stability and safety. The main sliding parameter estimation methods include proprioception and model-based reasoning. The proprioception method uses proprioception sensors to perceive ground properties and estimate sliding parameters through direct contact. Although this method can accurately obtain wheel information for sliding parameter estimation, it has a high risk of siltation due to the need for contact before measurement. The model-based reasoning sliding method is a wheel terrain analysis based on terrain mechanics. It infers sliding parameters by constructing a wheel-ground interaction model. Its disadvantage is that it is only valid for the current sliding parameter value and cannot predict the sliding parameters of future terrain.

[0004] Current research on sliding parameter estimation focuses primarily on analyzing sliding parameters in current and historical states. Traditional research focuses on using proprioception and model reasoning to estimate sliding parameters. While these methods offer promising results in estimating real-time sliding parameters, they are limited by their principles and make it difficult to reliably estimate future sliding parameters. Furthermore, current sliding parameter estimation techniques primarily focus on wheeled vehicles and structured roads, with limited research on tracked platforms and off-road scenarios, which are more significantly affected by sliding parameters. However, tracked platforms often play an irreplaceable role in off-road scenarios due to their unique advantages, making research on sliding parameter estimation for tracked platforms in off-road scenarios crucial. Summary of the Invention

[0005] In view of the above analysis, an embodiment of the present invention aims to provide a method for estimating vehicle sliding parameters, so as to solve the problem in the prior art of difficulty in predicting the sliding parameters of a vehicle in an off-road scenario.

[0006] In a first aspect, an embodiment of the present invention provides a method for estimating a vehicle sliding parameter, comprising the following steps:

[0007] Obtain point cloud data, image data, and vehicle trajectory data of the target scene;

[0008] The point cloud data, image data and vehicle trajectory data of the target scene are input into a pre-trained sliding parameter estimation model to obtain an estimation result of the sliding parameter; wherein,

[0009] The sliding parameter estimation model includes a feature extraction network, a feature fusion network, a sliding parameter identification network, and a linear layer connected in sequence;

[0010] The feature extraction network includes a first residual network for extracting point cloud features, a second residual network for extracting image features, and a trajectory feature extraction network;

[0011] The feature fusion network includes a first fusion network for fusing point cloud features and trajectory features and a second fusion network for fusing image features and trajectory features;

[0012] The sliding parameter recognition network includes a first convolutional long short-term memory network for performing sliding parameter learning on the fused point cloud features and trajectory features, and a second convolutional long short-term memory network for performing sliding parameter learning on the fused image features and trajectory features.

[0013] Based on a further improvement of the above method, the training method of the sliding parameter estimation model includes:

[0014] Obtain the target scene’s point cloud data, image data, vehicle trajectory data, and corresponding sliding parameter labels;

[0015] Inputting the point cloud data, image data and vehicle trajectory data of the target scene into an initial model of a sliding parameter estimation model to obtain a preliminary estimation result of the sliding parameter;

[0016] Based on the comparison between the preliminary estimation result of the sliding parameter and the sliding parameter label, an initial model of the sliding parameter estimation model is updated.

[0017] Based on a further improvement of the above method, obtaining point cloud data of the target scene includes:

[0018] The point cloud block data of the target area is intercepted from the point cloud map according to the geometric size of the vehicle and the vehicle's driving trajectory.

[0019] Based on a further improvement of the above method, the step of acquiring image data of the target scene includes:

[0020] Image data representing the same position is determined from the camera image according to the position information of the target area.

[0021] Based on a further improvement of the above method, before inputting the point cloud data of the target scene into the pre-trained sliding parameter estimation model, the following pre-processing operations are performed on the point cloud data of the target scene:

[0022] The point cloud data of the target scene is projected into a two-channel image containing height information and intensity information.

[0023] Based on a further improvement of the above method, the vehicle trajectory data includes: vehicle speed information, acceleration information and steering degree information.

[0024] Based on a further improvement of the above method, the first residual network is a ResNet-18 network model, and the second residual network is a ResNet-34 network model.

[0025] Based on a further improvement of the above method, before inputting the two-channel image and image data of the target scene into the pre-trained sliding parameter estimation model, the following operations are performed on the two-channel image and image data of the target scene:

[0026] Segmenting the two-channel image and the image data of the target scene according to the geometric size of the vehicle and the grid resolution to obtain the two-channel image and the image data of the target scene in a grid format;

[0027] Encoding the two-channel image of the target scene in a raster format and the raster representing the same position in the image data;

[0028] Furthermore, the estimation method further includes:

[0029] The estimated results of the sliding parameters are restored to the grids at corresponding positions based on the grid encoding, and a sliding parameter prediction grid map of the target scene is generated.

[0030] Based on the further improvement of the above method, the sliding parameter label is calculated by the following method:

[0031]

[0032] Among them, s is the sliding parameter label, v q is the vehicle drag speed, v s is the winding speed.

[0033] Based on a further improvement of the above method, the loss function used when training the sliding parameter estimation model is the mean square error.

[0034] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0035] 1. This invention utilizes image color information and three-dimensional point cloud data, combined with the vehicle's driving state, to develop a groundbreaking sliding parameter estimation network model. This network model, combining the strengths of residual networks and long-short-term memory networks, accurately predicts sliding parameters in a dataset constructed from a real off-road environment, achieving both sliding parameter prediction and estimation in off-road scenarios.

[0036] 2. While predicting sliding parameters using a dataset constructed in a real off-road environment, the present invention can also generate a sliding parameter map, improving the accuracy and comprehensiveness of the map, which is conducive to the safe and efficient driving of the vehicle platform in complex off-road environments.

[0037] In the present invention, the above-mentioned technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of the present invention will be described in the following description, and some advantages will become apparent from the description or be learned through practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the contents particularly pointed out in the description and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The accompanying drawings are only used for the purpose of illustrating specific embodiments and are not to be considered as limiting the present invention. Throughout the drawings, the same reference symbols denote the same components.

[0039] Figure 1 A schematic flow chart of a method for estimating vehicle sliding parameters according to an embodiment of the present invention is shown.

[0040] Figure 2 A schematic diagram of acquiring point cloud data of a target scene according to an embodiment of the present invention is shown.

[0041] Figure 3 A schematic diagram is shown for temporally synchronizing point cloud data, image data, and vehicle trajectory data in an embodiment of the present invention.

[0042] Figure 4 A schematic diagram of a network framework of a sliding parameter estimation model according to an embodiment of the present invention is shown.

[0043] Figure 5 A schematic diagram of the network structure of a sliding parameter estimation model according to another embodiment of the present invention is shown.

[0044] Figure 6 A schematic diagram of a point cloud projection image according to an embodiment of the present invention is shown.

[0045] Figure 7 A schematic diagram of a process for generating a sliding parameter map according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0046] The preferred embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, and are not used to limit the scope of the present invention.

[0047] Figure 1 FIG. 1 is a flow chart showing a method for estimating vehicle sliding parameters according to an embodiment of the present invention. Figure 1 As shown, the method for estimating the vehicle sliding parameter includes the following steps:

[0048] Step S101: Acquire point cloud data, image data, and vehicle trajectory data of a target scene.

[0049] In this embodiment, the vehicle's own sensors can be used to collect environmental information around the vehicle and vehicle trajectory data. Specifically, sensors such as a lidar, a monocular RGB camera, an inertial navigation system, a GPS receiver, and vehicle underlying controller information can be used to collect point cloud maps and image data of the vehicle's surroundings, three-axis acceleration information from the inertial navigation system, positioning information from the GPS system, and vehicle underlying speed and steering information.

[0050] In this embodiment, the point cloud data of the target scene refers to the point cloud data of the target area in the point cloud map collected by the vehicle's own laser radar. For example, the target area can be a block area located a certain distance in front of the vehicle.

[0051] In some embodiments, the point cloud data of the target scene can be obtained by the following method:

[0052] The point cloud block data of the target area is intercepted from the point cloud map according to the geometric size of the vehicle and the vehicle's driving trajectory.

[0053] Figure 2 FIG. 1 shows a schematic diagram of obtaining point cloud data of a target scene according to an embodiment of the present invention. Figure 2 As shown, the geometric center of the vehicle is the origin o, the forward direction of the vehicle is the x-axis, the speed of the vehicle is v (on the x-axis), and the vehicle coordinate system xoy is established with the geometric center of the vehicle as the origin. In this embodiment, the corresponding point cloud block can be extracted according to the geometric dimensions of the vehicle. The size of the point cloud block is obtained based on the rectangular box surrounding the outer contours of the left and right tracks of the vehicle. In order to ensure that the vehicle can completely fall within the point cloud block and reserve a certain error range, the length and width of the vehicle can be rounded up. Figure 2 In the figure, the length of the rectangular box is [length] and the width is [width]. According to the vehicle's driving trajectory, the point cloud block at a distance d in front of the vehicle can be intercepted. It is known that the vehicle's geometric center point o is the current discrete trajectory point, and the heading angle at this point is θ. The coordinates of the vertex of the target area can be obtained as A(xa ,y a )、B(x b ,y b )、C(x c ,y c ) and D(x d ,y d ), as shown in the following table:

[0054]

[0055]

[0056] It is known that the coordinates of the origin o in the xoy coordinate system are (x o ,y o ), the heading angle is θ, and the point A (x A ,y A ),B(x B ,y B ),C(x C ,y C ),D(x D ,y D ) is shown in the following formula:

[0057]

[0058] Finally, according to the geodetic coordinate system, point A(x A ,y A ),B(x B ,y B ),C(x C ,y C ),D(x D ,y D )’s real coordinates to intercept the point cloud and obtain the point cloud data of the target scene.

[0059] In this embodiment, by extracting corresponding point cloud blocks according to the geometric dimensions of the vehicle, the quality and relevance of the features expressed by the point cloud data can be improved.

[0060] In this embodiment, the image data of the target scene and the spatiotemporally aligned vehicle trajectory data can be obtained by temporally and spatially aligning the point cloud data, the image data, and the vehicle trajectory data. The spatiotemporal alignment of the point cloud data, the image data, and the vehicle trajectory data is described below.

[0061] When collecting data, cameras capture image frames through instantaneous exposure, resulting in a higher frequency of image data. LiDAR, on the other hand, captures point cloud data through rotating laser beams, resulting in a lower frequency of point cloud data. Furthermore, GPS and inertial navigation systems also collect data at a higher frequency, resulting in a higher frequency of vehicle trajectory data. Due to the differences in sampling frequencies between these three types of data, the point cloud data, image data, and vehicle trajectory data must first be synchronized.

[0062] Figure 3 FIG. 1 shows a schematic diagram of time synchronization of point cloud data, image data and vehicle trajectory data in an embodiment of the present invention. Figure 3 As shown, a multi-threading and queue caching method can be used to simultaneously open three threads to receive point cloud, image, and trajectory data. The timestamp of the lower-frequency point cloud data can be used as the message synchronization benchmark, and the image and trajectory data can be compared with the benchmark timestamp until the time difference is within the set threshold, thus completing the time synchronization of one frame of data.

[0063] After completing the time synchronization of the point cloud data, image data, and vehicle trajectory data, it is necessary to obtain the pose transformation relationship of the LiDAR and camera coordinate systems to achieve spatial alignment of the image data and point cloud data. It should be noted that the pose transformation relationship of the LiDAR and camera coordinate systems is a hyperparameter that can be learned during the training phase of the sliding parameter estimation model described below. During the application phase of the sliding parameter estimation model, as long as the position of the LiDAR and camera on the vehicle does not change, the pose transformation relationship of the LiDAR and camera coordinate systems learned during the training phase can be used.

[0064] The following describes the process of learning the pose transformation relationship between the LiDAR and camera coordinate systems. Existing calibration methods can be used to quickly jointly calibrate the camera and LiDAR. For example, features can be placed at key corner points on the ground and real-time image data can be acquired online. The camera's internal calibration parameters can then be obtained using the Harris corner and boundary tracking corner detection algorithm in the OpenCV library. The four corner points of the calibration object are then selected as seed points for the point cloud data, and the point cloud data for the key point area is obtained using a region growing algorithm. Subsequently, plane fitting based on a randomized consistency algorithm can be performed to obtain multiple sets of pose data representations for the same location area for the image data and point cloud data, thereby solving the spatial transformation relationship between the image data and point cloud data.

[0065] After learning the spatial transformation relationship between image data and point cloud data, we can use the position information of key points in the point cloud data of the target scene (e.g. Figure 2Through the spatial transformation relationship between the two, the matching position of the key points in the image data can be obtained, and then the image data representing the same position area can be obtained.

[0066] For the sliding parameters, in addition to being reflected to a certain extent by image data and point cloud data, the driving state of the vehicle will also have a certain impact on the sliding parameters. Therefore, in order to achieve accurate estimation of the sliding parameters, this embodiment also takes into account the trajectory data of the vehicle, such as the speed and acceleration information of the vehicle. Since the image data and point cloud data of the target scene obtained according to this embodiment are low-frequency signals, and the inertial navigation signal and GPS signal are high-frequency signals, when matching the trajectory data, the speed information, acceleration information, and vehicle steering degree information within a fixed time window near the most recent time corresponding to a frame of point cloud data of the target scene can be used. The speed information, acceleration information, and vehicle steering degree information within this fixed time window are the trajectory data of the vehicle after spatiotemporal alignment. For the speed, acceleration, and steering degree information x(i) containing N sampling signals in the time window, the following five commonly used statistics can be used to characterize its characteristics:

[0067] (1) Mean:

[0068]

[0069] (2) Standard deviation:

[0070]

[0071] (3) Square root amplitude:

[0072]

[0073] (4) Root mean square value:

[0074]

[0075] (5) Maximum absolute value:

[0076] f5=max|x(i)|

[0077] In some embodiments, the acceleration information can be selected from the longitudinal acceleration and the vertical acceleration for analysis. The longitudinal acceleration reflects the acceleration or deceleration of the vehicle, during which the sliding parameter changes greatly. The vertical acceleration is a measure of the acceleration of the vehicle in the vertical direction, and the gravity acceleration a needs to be eliminated in advance. g =9.8m / s 2The influence of the vehicle's vibration during driving can, to a certain extent, reflect the current ground properties. The speed information can be analyzed by selecting the three-axis speeds. The longitudinal speed represents the vehicle's driving speed, the lateral speed represents the vehicle's steering and other operations, and the vertical speed represents the vehicle's vertical climb. The steering degree information is the vehicle's real-time steering curvature calculated by combining the left and right speeds of the underlying motor with the inertial navigation system. For the above six variables, each variable can generate five statistical features, for a total of 30 statistical features.

[0078] Step S102: inputting the point cloud data, image data and vehicle trajectory data of the target scene into a pre-trained sliding parameter estimation model to obtain an estimation result of the sliding parameter.

[0079] Figure 4 FIG. 1 shows a schematic diagram of a network framework of a sliding parameter estimation model according to an embodiment of the present invention. Figure 4 As shown, the sliding parameter estimation model includes a sensor data feature extraction network, a feature fusion network, a parameter (i.e., sliding parameter) identification network, and a hierarchical linearization layer connected in sequence. The sensor data feature extraction network includes a point cloud data residual network, a camera data residual network, and a trajectory feature extraction network. The point cloud data residual network is used to extract point cloud features from the point cloud data of the target scene. The camera data residual network is used to extract image features from the image data of the target scene. The trajectory feature extraction network is used to extract trajectory features from the vehicle's trajectory data. The feature fusion network includes a first fusion network and a second fusion network ( Figure 4 The first fusion network is used to fuse the point cloud features and the trajectory features. The second fusion network is used to fuse the image features and the trajectory features. The parameter (i.e. sliding parameter) recognition network includes a point cloud data convolutional long short-term memory network (i.e. Figure 4 Point cloud data ConvLSTM) and camera data convolutional long short-term memory network (i.e. Figure 4 The output of the first fusion network is fed into the point cloud data convolutional long short-term memory network, while the output of the second fusion network is fed into the camera data convolutional long short-term memory network. The point cloud data convolutional long short-term memory network learns sliding parameters for the fused point cloud and trajectory features, while the camera data convolutional long short-term memory network learns sliding parameters for the fused image and trajectory features. The outputs of the point cloud data ConvLSTM and the camera data ConvLSTM are fed into a hierarchical linearization layer. Through fusion and extraction of multiple linearization layers, the sliding parameter estimates are ultimately output.

[0080] Figure 5FIG. 1 shows a schematic diagram of a network structure of a sliding parameter estimation model according to another embodiment of the present invention. Figure 5 As shown in the figure, the input data of the sliding parameter estimation model consists of three parts: three-channel image information (RGB) representing the image data of the target scene, two-channel image information (height and reflection intensity) representing the point cloud data of the target scene, and speed, acceleration, and steering information representing the vehicle's trajectory data.

[0081] In this embodiment, before inputting the target scene's point cloud data into the sliding parameter estimation model, the target scene's point cloud data can be preprocessed to obtain two-dimensional image information of the point cloud data. Specifically, each point cloud block can be projected into a two-channel image containing height and intensity information. Figure 6 FIG. 1 shows a schematic diagram of a point cloud projection image according to an embodiment of the present invention. Figure 6 As shown, each point in the 3D point cloud block is assigned to the nearest pixel in the image, and each pixel unit corresponds to many point cloud data points p i , called a point cloud cluster. Calculate the height mean h of the point cloud cluster in each pixel mean and the mean reflection intensity i mean , and assign it to the red and blue channels of the corresponding pixels in the image. For areas where the point cloud is missing, linear interpolation is performed through the surrounding pixel blocks.

[0082] In this embodiment, before inputting the vehicle trajectory data into the sliding parameter estimation model, feature extraction can be performed on the vehicle trajectory data. For example, for the longitudinal acceleration and vertical acceleration in the acceleration information, the following five features are calculated respectively:

[0083] (1) Mean, (2) Standard deviation, (3) RMS amplitude, (4) RMS value, (5) Maximum absolute value.

[0084] For the longitudinal velocity, lateral velocity, and vertical velocity in the velocity information, the following five features are calculated respectively:

[0085] (1) Mean, (2) Standard deviation, (3) RMS amplitude, (4) RMS value, (5) Maximum absolute value.

[0086] For the steering degree information, the following five features are calculated respectively:

[0087] (1) Mean, (2) Standard deviation, (3) RMS amplitude, (4) RMS value, (5) Maximum absolute value.

[0088] For the above six variables, each variable can generate 5 statistical features, and a total of 30 statistical features can be generated.

[0089] Back to Figure 5The input parameters for three-channel image information are [b, 3, a, b']. The first dimension parameter b represents the batch size, the second dimension parameter represents the three-channel image, and the third and fourth dimension parameters represent the image size information. The input parameters for vehicle trajectory data are [b, 6*5]. The first dimension parameter b represents the batch size, and 6*5 represents "six variables, each of which can generate 5 statistical features." The input parameters for point cloud data are [b, 3, a, b']. The first dimension parameter b represents the batch size, the second dimension parameter represents the two-channel image, and the third and fourth dimension parameters represent the image size information.

[0090] The three-channel image information is input into a residual network module (i.e. Figure 5 The two-channel point cloud information is input to another residual network module (i.e. Figure 5 (The "Resnet" in the lower middle section) The trajectory features are expanded to [b, w*h] through successive linearization layers and reshaped to [b, w, h] through a reshape operation. The upper residual network module extracts image features from the three-channel image information, while the lower residual network module extracts point cloud features from the two-channel point cloud information. The image and point cloud features are then fused with the trajectory features through element-wise multiplication. The fused results are then subjected to convolution, batch normalization, activation, and pooling. Figure 5 The "CBRP" modules in the figure represent the convolutional layer (C), batch normalization layer (B), activation layer (R), and pooling layer (P). The fused features output by the "CBRP" modules are fed into the ConvLSTM modules for sliding parameter learning. Finally, the outputs of the two ConvLSTM modules are flattened and concatenated. The data is then subjected to layer-wise linearization followed by channel-wise linearization, ultimately regressing to sliding parameter estimates.

[0091] In some embodiments, the ResNet-34 network is used for image feature extraction, and the ResNet-18 network is used for point cloud feature extraction. Image data contains rich environmental features, and the more complex ResNet-34 network can fully utilize the features contained in the image.

[0092] In some embodiments, before inputting the two-channel image and image data of the target scene into the pre-trained sliding parameter estimation model, the following operations may be performed on the two-channel image and image data of the target scene:

[0093] Step S111: segmenting the two-channel image and image data of the target scene according to the geometric size of the vehicle and the grid resolution to obtain the two-channel image and image data of the target scene in a grid format.

[0094] Step S222: Encode the two-channel image of the target scene in the grid format and the grid representing the same position in the image data.

[0095] and, Figure 1 The estimation methods in also include:

[0096] Step S103: restoring the estimated result of the sliding parameter to the grid at the corresponding position based on the grid coding, and generating a sliding parameter prediction grid map of the target scene.

[0097] Figure 7 A schematic diagram of a process for generating a sliding parameter map according to an embodiment of the present invention is shown.

[0098] like Figure 7 As shown, the color image on the left represents the two-channel image information of the point cloud data representing the target scene, the gray image on the left represents the three-channel image information of the image data representing the target scene, and the function "f 1.2.3.4.5 (v x.v , α x.v.z , δ steer )" represents the "six variables, each of which can generate five statistical features" in the trajectory features described above. The middle module "module" represents the sliding parameter estimation model in an embodiment of the present invention. First, the two images on the left can be segmented by position, combining the vehicle size and grid resolution, to obtain a series of point cloud data and image data representing the same location. In the process of generating the raster map, each location can be encoded to ensure that the sliding parameter prediction results output by the sliding parameter estimation model can match the corresponding location. Finally, the point cloud data and image data are combined with the vehicle trajectory features as input, and estimation is performed using the trained sliding parameter estimation model. The estimation results are restored to the corresponding grid position based on the position encoding to generate the sliding parameter prediction grid map for the current scene.

[0099] The following describes the training process of the sliding parameter estimation model in the embodiment of the present invention.

[0100] The training method of the sliding parameter estimation model in the embodiment of the present invention includes:

[0101] Step S100: Obtain point cloud data, image data, vehicle trajectory data and corresponding sliding parameter labels of the target scene.

[0102] The method for obtaining the target scene's point cloud data, image data, and vehicle trajectory data has been described in step S101 and will not be repeated here. The processing of the sliding parameter label in this embodiment is as follows:

[0103] The generation of sliding parameters is expressed in terms of the absolute value of the slip, as follows:

[0104]

[0105] Among them, s is the sliding parameter label; v q is the vehicle's drag speed, i.e. the vehicle's actual speed; v s is the winding speed, is the driving speed of the vehicle motor. When the vehicle does not have slip, v q =v s , at this time the sliding parameter is 0; when the driving wheel is locked and fully braked, v s =0, it is a pure slip state and the sliding parameter is 1.

[0106] Step S200: inputting the point cloud data, image data and vehicle trajectory data of the target scene into an initial model of a sliding parameter estimation model to obtain a preliminary estimation result of the sliding parameter.

[0107] Step S300: Based on the comparison between the preliminary estimation result of the sliding parameter and the sliding parameter label, the initial model of the sliding parameter estimation model is updated.

[0108] In this embodiment, the training process of the initial model of the sliding parameter estimation model can use the mean squared error (MSE). The mean squared error is a loss function in regression tasks. It is the mean of the sum of squares of the differences between the predicted value f(x) and the target value y. It is used to measure the difference between the predicted value of the initial model of the sliding parameter estimation model and the true value. The calculation formula is as follows:

[0109]

[0110] Where n is the number of samples.

[0111] Compared with the prior art, the embodiments of the present invention can achieve at least one of the following beneficial effects:

[0112] 1. This invention utilizes image color information and three-dimensional point cloud data, combined with the vehicle's driving state, to develop a groundbreaking sliding parameter estimation network model. This network model, combining the strengths of residual networks and long-short-term memory networks, accurately predicts sliding parameters in a dataset constructed from a real off-road environment, achieving both sliding parameter prediction and estimation in off-road scenarios.

[0113] 2. While predicting sliding parameters using a dataset constructed in a real off-road environment, the present invention can also generate a sliding parameter map, improving the accuracy and comprehensiveness of the map, which is conducive to the safe and efficient driving of the vehicle platform in complex off-road environments.

[0114] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or replacements that can be easily conceived by any technician familiar with this technical field within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. A method for estimating vehicle sliding parameters, characterized in that: The following steps are involved: Obtain point cloud data, image data, and vehicle trajectory data of the target scene; The point cloud data, image data and vehicle trajectory data of the target scene are input into a pre-trained sliding parameter estimation model to obtain an estimation result of the sliding parameter; wherein, The sliding parameter estimation model includes a feature extraction network, a feature fusion network, a sliding parameter identification network, and a linear layer connected in sequence; The feature extraction network includes a first residual network for extracting point cloud features, a second residual network for extracting image features, and a trajectory feature extraction network; The feature fusion network includes a first fusion network for fusing point cloud features and trajectory features and a second fusion network for fusing image features and trajectory features; The sliding parameter recognition network includes a first convolutional long short-term memory network for performing sliding parameter learning on the fused point cloud features and trajectory features, and a second convolutional long short-term memory network for performing sliding parameter learning on the fused image features and trajectory features; The training method of the sliding parameter estimation model includes: Obtain the target scene’s point cloud data, image data, vehicle trajectory data, and corresponding sliding parameter labels; Inputting the point cloud data, image data and vehicle trajectory data of the target scene into an initial model of a sliding parameter estimation model to obtain a preliminary estimation result of the sliding parameter; updating an initial model of the sliding parameter estimation model based on a comparison between the preliminary estimation result of the sliding parameter and the sliding parameter label; The sliding parameter label is calculated as follows: Among them, s is the sliding parameter label, v q is the vehicle drag speed, v s is the winding speed.

2. The estimation method according to claim 1, characterized in that The step of obtaining the point cloud data of the target scene includes: The point cloud block data of the target area is intercepted from the point cloud map according to the geometric size of the vehicle and the vehicle's driving trajectory.

3. The estimation method according to claim 2, characterized in that The acquiring of image data of the target scene comprises: Image data representing the same position is determined from the camera image according to the position information of the target area.

4. The estimation method according to claim 2, characterized in that Before inputting the point cloud data of the target scene into the pre-trained sliding parameter estimation model, the following pre-processing operations are performed on the point cloud data of the target scene: The point cloud data of the target scene is projected into a two-channel image containing height information and intensity information.

5. The estimation method according to claim 1, wherein: The vehicle trajectory data includes: vehicle speed information, acceleration information and steering degree information.

6. The estimation method according to claim 1, wherein: The first residual network is a ResNet-18 network model, and the second residual network is a ResNet-34 network model.

7. The estimation method according to claim 4, characterized in that Before inputting the two-channel image and image data of the target scene into the pre-trained sliding parameter estimation model, the following operations are performed on the two-channel image and image data of the target scene: Segmenting the two-channel image and the image data of the target scene according to the geometric size of the vehicle and the grid resolution to obtain the two-channel image and the image data of the target scene in a grid format; Encoding the two-channel image of the target scene in a raster format and the raster representing the same position in the image data; Furthermore, the estimation method further includes: The estimated results of the sliding parameters are restored to the grids at corresponding positions based on the grid encoding, and a sliding parameter prediction grid map of the target scene is generated.

8. The estimation method according to claim 1 or 7, characterized in that: The loss function used when training the sliding parameter estimation model is the mean square error.

Images