An automatic driving vehicle path planning method, device, equipment and medium

Abstract

The application discloses an automatic driving vehicle path planning method and device, equipment and medium. The method comprises the following steps: acquiring the sensing information of an unmanned vehicle; determining the safety coefficient of the vehicle and at least one dynamic obstacle according to the sensing information; wherein the safety coefficient is the ratio of the current distance between the vehicle and the dynamic obstacle and the safety distance; determining the safety loss matched with each dynamic obstacle according to each safety coefficient, taking each safety loss as the prior feature of each dynamic obstacle, inputting the prediction model, and determining the planning path of the automatic driving vehicle. The technical scheme solves the problems of low accuracy and high time cost caused by the direct abstract feature extraction of the prediction model on the sensing information, and can improve the prediction accuracy and model generation efficiency of the prediction model while enhancing the explainability of the prediction model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned driving, and in particular to an automatic driving vehicle path planning method, device, equipment and medium. BACKGROUND

[0002] With the continuous progress of technology, the demand for navigation capability of unmanned vehicles under various road conditions is increasingly urgent. The navigation system of the unmanned vehicle needs to obtain radar, visual and other perception information, and make reasonable decisions based on the recognition results of the perception information, so as to reach the destination in time and accurately under the premise of ensuring driving safety.

[0003] At present, the navigation system of the unmanned vehicle mainly relies on Lyft, waymo and other prediction models to directly perform feature representation on the perception information based on a deep learning network, and makes driving decisions according to the abstract features extracted from the perception information.

[0004] However, the existing technology directly represents the perception information through a deep learning network, which needs to consume a large amount of time to train the deep learning network, and the stability of the training effect is difficult to guarantee, and the prediction model has poor interpretability. SUMMARY

[0005] The present application provides an automatic driving vehicle path planning method, device, equipment and medium to solve the problems of low accuracy and high time cost caused by directly extracting abstract features from perception information by a prediction model, which can improve the prediction accuracy and model generation efficiency of the prediction model while enhancing the interpretability of the prediction model.

[0006] According to an aspect of the present application, an automatic driving vehicle path planning method is provided, the method comprising:

[0007] obtaining perception information of the vehicle; wherein the perception information comprises speed information of the vehicle, speed information of dynamic obstacles, distance information between the vehicle and the dynamic obstacles, and response time;

[0008] determining a safety coefficient of the vehicle and at least one dynamic obstacle according to the perception information; wherein the safety coefficient is a ratio of the current distance between the vehicle and the dynamic obstacle to the safety distance;

[0009] determining a safety loss matched with each dynamic obstacle according to each safety coefficient, taking each safety loss as a prior feature of each dynamic obstacle, inputting the prediction model, and determining a planned path of the automatic driving vehicle.

[0010] According to another aspect of the present application, an automatic driving vehicle path planning device is provided, the device comprising:

[0011] The perception information acquisition module is configured to acquire perception information of the vehicle, wherein the perception information comprises speed information of the vehicle, speed information of the dynamic obstacle, distance information between the vehicle and the dynamic obstacle, and response time;

[0012] The safety factor determination module is configured to determine a safety factor between the vehicle and the at least one dynamic obstacle according to the perception information, wherein the safety factor is a ratio of a current distance between the vehicle and the dynamic obstacle to a safety distance.

[0013] The planned path determination module is configured to determine a safety loss matched with each dynamic obstacle according to each safety factor, input the safety loss as a prior feature of each dynamic obstacle into a prediction model, and determine a planned path of the autonomous vehicle.

[0014] According to another aspect of the present application, an electronic device is provided, which comprises:

[0015] at least one processor; and

[0016] a memory connected to the at least one processor in communication; wherein

[0017] The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the autonomous vehicle path planning method according to any one of the embodiments of the present application.

[0018] According to another aspect of the present application, a computer readable storage medium is provided, which stores computer instructions for enabling a processor to execute the autonomous vehicle path planning method according to any one of the embodiments of the present application.

[0019] The technical scheme of the embodiments of the present application acquires perception information of the vehicle, determines a safety factor between the vehicle and at least one dynamic obstacle according to the perception information, determines a safety loss matched with each dynamic obstacle according to each safety factor, inputs the safety loss as a prior feature of each dynamic obstacle into a prediction model, and determines a planned path of the autonomous vehicle. This scheme solves the problems of low accuracy and high time cost caused by the prediction model directly extracting abstract features from the perception information, and can improve the prediction accuracy and model generation efficiency of the prediction model while enhancing the explainability of the prediction model.

[0020] It should be understood that the contents described in this part are not intended to identify key or important features of the embodiments of the present application, nor are they used to limit the scope of the present application. Other features of the present application will become apparent from the following description. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiments description. Obviously, the drawings in the following description only show some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without any creative effort.

[0022] Figure 1 is a flow chart of an automatic driving vehicle path planning method according to an embodiment of the present application;

[0023] Figure 2 is a structural schematic diagram of a prediction model according to an embodiment of the present application;

[0024] Figure 3 is a flow chart of an automatic driving vehicle path planning method according to an embodiment of the present application;

[0025] Figure 4 is a candidate region division schematic diagram in a simple driving scene according to an embodiment of the present application;

[0026] Figure 5 is a candidate region division schematic diagram in a complex driving scene according to an embodiment of the present application;

[0027] Figure 6 is a structural schematic diagram of an automatic driving vehicle path planning device according to an embodiment of the present application;

[0028] Figure 7 is a structural schematic diagram of an electronic device for implementing an automatic driving vehicle path planning method according to an embodiment of the present application. DETAILED DESCRIPTION

[0029] In order to make the technical personnel in the art better understand the present application scheme, the following will combine the drawings in the embodiments of the present application, and the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without any creative effort should belong to the scope of protection of the present application.

[0030] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and in the above drawings are used to distinguish similar objects, and do not necessarily have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the application described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, device, product or equipment including a series of steps or units does not have to be limited to only those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or equipment. The acquisition, storage, use, processing and other data in the technical solutions of the present application comply with the relevant provisions of national laws and regulations.

[0031] Embodiment one

[0032] Figure 1 A flowchart of an automatic driving vehicle path planning method is provided for the first embodiment of the present application. The present embodiment can be applied to the path planning scenario of an automatic driving vehicle. The method can be executed by an automatic driving vehicle path planning device, which can be realized in the form of hardware and / or software. The device can be configured in an electronic device.

[0033] As shown in Figure 1 , the method comprises:

[0034] S110, acquiring perception information of the vehicle; wherein the perception information comprises speed information of the vehicle, speed information of dynamic obstacles, distance information between the vehicle and the dynamic obstacles, and response time.

[0035] The present solution can be executed by an automatic driving vehicle controller. The controller can acquire multi-dimensional perception information through sensors such as radars and cameras deployed on the vehicle. The perception information can include the speed of the vehicle, the speed of dynamic obstacles, the distance between the vehicle and the dynamic obstacles, and the response time, etc. The controller can identify the types of each target according to the driving scene images collected by the camera, such as static obstacles and dynamic obstacles. Further, the controller can also achieve fine identification of each target, such as fine identification of static targets, which can include green belts and buildings, and fine identification of dynamic obstacles, which can include vehicles, pedestrians and bicycles, etc. According to the radar data, the controller can determine the distance information between the vehicle and each dynamic obstacle, and the speed information of the dynamic obstacles. It is easy to understand that the controller can determine the speed information of the vehicle according to the speed sensor of the vehicle, and can acquire the acceleration of the vehicle through an inertial sensor such as IMU. The controller can also record the response time of the vehicle to the control decision.

[0036] S120, determine a safety coefficient of the ego vehicle and the at least one dynamic obstacle according to the perception information; wherein the safety coefficient is a ratio of a current distance between the ego vehicle and the dynamic obstacle and a safety distance.

[0037] According to the perception information of the ego vehicle, the controller can calculate the safety coefficient of the ego vehicle and each dynamic obstacle in the perception range, wherein the safety coefficient can be used to evaluate the degree of danger of the dynamic obstacle to the ego vehicle. Specifically, the safety coefficient can be represented by the ratio of the current distance between the ego vehicle and the dynamic obstacle and the safety distance, which can be calculated based on the speed of the ego vehicle, the speed of the dynamic obstacle and the response time of the ego vehicle to driving decisions.

[0038] S130, determine a safety loss matched with each dynamic obstacle according to each safety coefficient, and input the safety loss of each dynamic obstacle as prior features of each dynamic obstacle into a prediction model to determine a planning path of the autonomous vehicle.

[0039] After obtaining the safety coefficient corresponding to each dynamic obstacle, the controller can normalize each safety coefficient to obtain a safety loss. The safety loss matched with each dynamic obstacle is input into the prediction model as prior features of each dynamic obstacle, and according to the output result of the prediction model, the controller can determine the planning path of the autonomous vehicle.

[0040] The prediction model can use the motion features of the ego vehicle, the navigation map features and the obstacle features as the basis to plan the path of the ego vehicle and output prediction information of the motion of the ego vehicle. The motion features of the ego vehicle can include the position, driving direction and motion speed of the ego vehicle. The navigation map features can include the road features of the driving environment of the ego vehicle, such as road boundary lines, reference lines and other features. The obstacle features can include the prior features of the dynamic obstacle, and can also include the identification, position, motion direction and motion speed of the obstacle. The prediction model can be constructed based on a deep learning algorithm, and the prediction information can include the position, orientation angle, speed, acceleration, curvature, acceleration and time interval of the ego vehicle. According to the prediction information, the controller can generate a planning path of the autonomous vehicle to ensure safe driving of the ego vehicle.

[0041] Specifically, the prediction model can be obtained by pre-training the training sample data, and the training sample data can include multiple groups to achieve good training effect of the prediction model. Each group of training sample data can include feature data and label data. Similar to the application process of the prediction model, the feature data can include the motion features of the ego vehicle, the navigation map features and the features of at least one obstacle. The label data can be the motion information of the ego vehicle matched with the feature data.

[0042] Figure 2 This is a schematic diagram of the structure of a prediction model provided according to an embodiment of the present invention. Figure 2 As shown, the prediction model may include an input branching structure, a multi-head self-attention structure, a multilayer perceptron structure, and a long short-term memory structure. The input branching structure can be used to extract features from vehicle motion features, obstacle features, and navigation map features. The multi-head self-attention structure can be used to extract relationship features between various input features, thereby improving the accuracy and efficiency of the prediction model. The multilayer perceptron structure can compress features. The long short-term memory structure can extract temporal features between multiple frames of input data.

[0043] In a preferred embodiment, the Long Short-Term Memory (LSTM) structure can use the current frame and four historical frames as input data to extract temporal features between the input data, thereby achieving good prediction results.

[0044] In a feasible scheme, the input branch structure includes a first input branch, a second input branch, and a third input branch; the first input branch, the second input branch, and the third input branch are respectively used to extract features from the vehicle's motion features, obstacle features, and navigation map features; the input branch structure, the multi-head self-attention structure, the multilayer perceptron structure, and the long short-term memory structure are sequentially connected.

[0045] As is easily understood, each input branch in the input branch structure is used to extract features from the vehicle's motion features, obstacle features, and navigation map features, respectively. Each input branch can include at least one fully connected layer, and the structures of the input branches can be the same or different. The first input branch, the second input branch, and the third input branch can extract features from the vehicle's motion features, obstacle features, and navigation map features, respectively, to obtain the first feature extraction result, the second feature extraction result, and the third feature extraction result.

[0046] The controller can input the first, second, and third feature extraction results into a multi-head self-attention structure according to a preset input method to extract the relationship features between the input features. After obtaining the output features with relationship features from the multi-head self-attention structure, the controller can use a multilayer perceptron mechanism to further extract deeper features from the output features, and input the extracted deep features into a long short-term memory structure to extract the temporal features between deep features across multiple frames.

[0047] The prediction model structure of this scheme can realize multi-dimensional feature extraction of vehicle motion characteristics, obstacle characteristics, and navigation map characteristics, which is beneficial to improving the robustness and accuracy of the model.

[0048] Based on the above scheme, optionally, the multi-head self-attention structure can include three inputs; wherein the first input and the second input are both determined based on the first feature extraction result of the first input branch, the second feature extraction result of the second input branch, and the third feature extraction result of the third input branch; and the third input is determined based on the first feature extraction result of the first input branch.

[0049] Specifically, the controller can fuse the first feature extraction result, the second feature extraction result, and the third feature extraction result according to a preset fusion manner, and take the fused feature as the first input or the second input. The first input and the second input can be the same or different. The controller can directly splice the first feature extraction result, the second feature extraction result, and the third feature extraction result according to a certain arrangement order to obtain the fused feature. For example, A, B, and C represent the first feature extraction result, the second feature extraction result, and the third feature extraction result respectively, and the fused feature can be represented as [A, B, C].

[0050] The controller can also sequentially select part of the first feature extraction result, the second feature extraction result, and the third feature extraction result, and alternately splice different feature extraction results to generate the fused feature. For example, the fused feature can be represented as [A1, B1, C1, A2, B2, C2], wherein A1 represents the first part of the first feature extraction result, A2 represents the second part of the first feature extraction result, and B1, C1, B2, and C2 are the same.

[0051] It can be understood that the controller can also take the first feature extraction result as the third input to determine the output feature having the relationship between the vehicle motion feature and the navigation map feature and the obstacle feature according to the first input, the second input, and the third input. The calculation formula of the output feature can be represented as:

[0052]

[0053] wherein K represents the first input, V represents the second input, Q represents the third input, d k represents the dimension of the first input, softmax represents the normalization exponential function, and Attention(Q, K, V) represents the output feature of the multi-head attention mechanism.

[0054] The prediction model with the above multi-head self-attention structure can improve the prediction efficiency and accuracy.

[0055] The scheme is not directly inputting the perception information of the vehicle into the prediction model to extract the abstract features of the obstacles and make the driving decision judgment, but according to the perception information, the prior features of the obstacles are determined, the prior features are taken as the basis for the driving decision, and are input into the prediction model to obtain the driving decision of the vehicle. Therefore, the scheme can improve the decision accuracy and decision efficiency of the prediction model.

[0056] The technical scheme determines the safety factors of the vehicle and at least one obstacle according to the perception information, determines the safety loss matched with each obstacle according to each safety factor, and takes each safety loss as the prior feature of each obstacle to input the prediction model to determine the planning path of the autonomous vehicle. The scheme solves the problems of low accuracy and high time cost caused by the direct abstract feature extraction of the perception information by the prediction model, and can improve the prediction accuracy and model generation efficiency of the prediction model while enhancing the explainability of the prediction model.

[0057] Embodiment two

[0058] Figure 3 A flowchart of an autonomous vehicle path planning method is provided for embodiment two of the present application, and the embodiment is refined based on the above-mentioned embodiment. For dynamic obstacles in the obstacles, embodiment two further refines the calculation of the safety factors, as shown in the formula (1), the method comprises the steps of: Figure 3

[0059] S210, acquiring perception information of the vehicle; wherein the perception information comprises speed information of the vehicle, speed information of the dynamic obstacle, distance information between the vehicle and the dynamic obstacle, and response time.

[0060] In the scheme, the speed information can include lateral speed, longitudinal speed, lateral acceleration and longitudinal acceleration, and the distance information can include lateral distance and longitudinal distance information.

[0061] S220, determining the safety distance between the vehicle and each dynamic obstacle according to the speed information of the vehicle, the speed information of the dynamic obstacle and the response time.

[0062] ​It can be understood that, due to the driving direction of the dynamic obstacle being different in the driving scene, the controller can analyze the host vehicle and each dynamic obstacle from both the lateral and longitudinal directions. According to the speed of the host vehicle, the acceleration of the host vehicle, the speed of the dynamic obstacle, the acceleration of the dynamic obstacle, and the response time and other information, the controller can calculate the lateral safety distance of the host vehicle and each dynamic obstacle. According to the lateral safety distance, the controller can calculate the lateral safety factor matched by each dynamic obstacle. According to the speed of the host vehicle, the acceleration of the host vehicle, the speed of the dynamic obstacle, the acceleration of the dynamic obstacle, and the response time and other information, the controller can also calculate the longitudinal safety distance of the host vehicle and each dynamic obstacle. According to the longitudinal safety distance, the controller can calculate the longitudinal safety factor matched by each dynamic obstacle.

[0063] In one possible implementation, the safety distance includes a longitudinal distance.

[0064] Correspondingly, the determination of the safety distance of the host vehicle and each dynamic obstacle according to the speed information of the host vehicle, the speed information of the dynamic obstacle, and the response time includes:

[0065] According to the speed information of the host vehicle and the speed information of the dynamic obstacle, the relative motion direction of the host vehicle and the dynamic obstacle is determined.

[0066] According to the relative motion direction, the speed information of the host vehicle, the speed information of the dynamic obstacle, and the response time, the longitudinal safety distance of the host vehicle and each dynamic obstacle is determined.

[0067] It can be easily understood that the speed information of the host vehicle can indicate the motion direction of the host vehicle, and the speed information of the dynamic obstacle can indicate the motion direction of the dynamic obstacle. For the longitudinal safety distance, the controller can determine the relative motion direction of the host vehicle and the dynamic obstacle according to the running direction of the host vehicle and the motion direction of the dynamic obstacle. Based on the lateral and longitudinal directions of the host vehicle, the speed of the dynamic obstacle is decomposed, and the relative motion direction of the host vehicle and the dynamic obstacle is determined according to the decomposed speed of the dynamic obstacle in the longitudinal direction of the host vehicle. The controller can set different longitudinal safety distance calculation methods according to the relative motion direction of the host vehicle and the dynamic obstacle.

[0068] The present scheme determines the relative motion direction of the host vehicle and the dynamic obstacle through the motion direction indicated by the speed information, and then calculates the longitudinal safety distance of the host vehicle and each dynamic obstacle according to the different relative motion directions, using the speed information of the host vehicle, the speed information of the dynamic obstacle, and the response time. This scheme realizes accurate longitudinal safety distance calculation, which is conducive to ensuring the accuracy of safety factor calculation.

[0069] On the basis of the above scheme, optionally, the relative motion direction comprises same direction and opposite direction; the speed information comprises speed, lateral speed information and longitudinal speed information; wherein the longitudinal speed information comprises longitudinal speed, longitudinal deceleration and longitudinal acceleration; the lateral speed information comprises lateral speed;

[0070] The determining of the longitudinal safety distance between the host vehicle and each dynamic obstacle according to the relative motion direction, the speed information of the host vehicle, the speed information of the dynamic obstacle and the response time comprises:

[0071] If the relative motion direction between the host vehicle and the dynamic obstacle is same direction, the longitudinal safety distance between the host vehicle and the dynamic obstacle is determined according to the longitudinal speed of the preceding traffic target, the longitudinal speed of the following traffic target, the maximum longitudinal deceleration, the maximum longitudinal acceleration, the minimum longitudinal deceleration and the response time.

[0072] If the relative motion direction between the host vehicle and the dynamic obstacle is opposite direction, the longitudinal safety distance between the host vehicle and the dynamic obstacle is determined according to the longitudinal speed of the host vehicle, the longitudinal speed of the dynamic obstacle, the minimum longitudinal deceleration, the maximum longitudinal acceleration, the minimum longitudinal deceleration of the host vehicle and the response time.

[0073] If the relative motion direction between the host vehicle and the dynamic obstacle is same direction, the controller can calculate the longitudinal safety distance between the host vehicle and the dynamic obstacle according to the longitudinal speed of the preceding traffic target, the longitudinal speed of the following traffic target, the maximum longitudinal deceleration, the maximum longitudinal acceleration, the minimum longitudinal deceleration and the response time. The preceding traffic target and the following traffic target are different, and the preceding traffic target can be the host vehicle or the dynamic obstacle, and similarly, the following traffic target can be the host vehicle or the dynamic obstacle. The maximum longitudinal deceleration, the maximum longitudinal acceleration and the minimum longitudinal deceleration are determined based on the statistical results of traffic target attribute parameters. The traffic target attribute parameters can include the category, model, size, load range, acceleration range and deceleration range of the traffic target. The controller can obtain the attribute parameters of the traffic targets existing in the traffic environment, and statistically analyze the parameters of each category of traffic targets to obtain the maximum longitudinal deceleration, the maximum longitudinal acceleration and the minimum longitudinal deceleration of each type of traffic target. The controller can determine the maximum longitudinal deceleration, the maximum longitudinal acceleration and the minimum longitudinal deceleration of the dynamic obstacle according to the maximum longitudinal deceleration, the maximum longitudinal acceleration and the minimum longitudinal deceleration corresponding to the category of the dynamic obstacle.

[0074] In one specific scheme, the controller can determine a maximum displacement of the following traffic object within a response time according to a longitudinal speed of the following traffic object, a maximum longitudinal acceleration, and the response time. The controller can determine a maximum braking distance between the leading traffic object and the following traffic object within the response time according to the longitudinal speed of the following traffic object, a longitudinal speed of the leading traffic object, the maximum longitudinal acceleration, a minimum longitudinal deceleration, a maximum longitudinal deceleration, and the response time. The maximum displacement of the following traffic object and the maximum braking distance are added as the safety distance. Specifically, when the relative movement direction of the host vehicle and the dynamic obstacle is the same direction, the calculation formula of the longitudinal safety distance can be represented as:

[0075]

[0076] where v r represents the longitudinal speed of the following traffic object, v f represents the longitudinal speed of the leading traffic object, t represents the response time, a lon,max,accel represents the maximum longitudinal acceleration, a lon,min,brake represents the minimum longitudinal deceleration, and a lon,max,brake represents the maximum longitudinal deceleration.

[0077] If the relative movement direction of the host vehicle and the dynamic obstacle is the opposite direction, the controller can determine the longitudinal safety distance of the host vehicle and the dynamic obstacle according to the longitudinal speed of the host vehicle, the longitudinal speed of the dynamic obstacle, the minimum longitudinal deceleration, the maximum longitudinal acceleration, the minimum longitudinal deceleration of the host vehicle, and the response time.

[0078] Taking the host vehicle driving in the forward direction and the dynamic obstacle driving in the reverse direction as an example, the controller can determine the maximum speed of the host vehicle within the response time according to the longitudinal speed of the host vehicle, the maximum longitudinal acceleration, and the response time. Similarly, the controller can determine the maximum speed of the dynamic obstacle within the response time according to the longitudinal speed of the dynamic obstacle, the maximum longitudinal acceleration, and the response time. The controller can calculate the maximum displacement of the host vehicle according to the longitudinal speed of the host vehicle, the maximum speed of the host vehicle within the response time, the minimum longitudinal deceleration of the host vehicle, and the response time. The controller can calculate the maximum displacement of the dynamic obstacle according to the longitudinal speed of the dynamic obstacle, the maximum speed of the dynamic obstacle within the response time, the minimum longitudinal deceleration, and the response time. The maximum displacement of the host vehicle and the maximum displacement of the dynamic obstacle are added as the safety distance. When the host vehicle drives in the reverse direction and the dynamic obstacle drives in the forward direction, the relative movement direction of the host vehicle and the dynamic obstacle is the opposite direction, and the calculation formula of the longitudinal safety distance can be represented as:

[0079]

[0080] v 1lon,t = v 1lon+ta lon,max,accel ;

[0081] v 2lon,t =|v 2lon |+ta lon,max,accel ;

[0082] Among them, v 1lon v represents the speed of a traffic target traveling in the forward direction. 2lon Let t represent the speed of the traffic target traveling in the opposite direction, and a represent the response time. lon,max,accel a represents the maximum longitudinal acceleration. lon,min,brake a represents the minimum longitudinal deceleration. lon,min,brake,correct This indicates the minimum longitudinal deceleration of the vehicle.

[0083] In another feasible solution, the safety distance also includes a lateral safety distance; the lateral velocity information also includes lateral acceleration and lateral deceleration;

[0084] Accordingly, determining the safe distance between the vehicle and each dynamic obstacle based on the vehicle's speed information, the speed information of the dynamic obstacle, and the response time includes:

[0085] Based on the vehicle's lateral velocity, the lateral velocity of the dynamic obstacle, the maximum lateral acceleration, the minimum lateral deceleration, and the response time, determine the safe lateral distance between the vehicle and each dynamic obstacle.

[0086] For lateral safety distance, the controller can determine the lateral safety distance between the vehicle and each dynamic obstacle based on the relative position of the vehicle and the obstacle. The controller can use the lateral displacement of two traffic targets approaching each other with maximum lateral acceleration and braking with minimum lateral deceleration until they come to a stop as the lateral safety distance.

[0087] In a specific scheme, the formula for calculating the lateral safety distance can be expressed as:

[0088]

[0089] v 1lat,t =v 1lat +ta lat,max,accel ;

[0090] v 2lat,t =v 2lat -ta lat,max,accel ;

[0091] Where μ represents the fault tolerance space, v 1lat v represents the lateral velocity of the traffic target located on the left. 2lat The lateral velocity of the traffic target on the right is represented by t, and the response time is represented by a. lat,max,accelrepresents the maximum lateral acceleration, a lat,min,brake represents the minimum lateral deceleration.

[0092] The present scheme calculates the safety distance from the lateral and longitudinal angles respectively, which is conducive to the accurate calculation of the safety coefficient and maximizes the safety of the vehicle driving. In some embodiments, μ is 0.2m, and the fault tolerance is increased on the basis of the calculation of the safety distance to ensure driving safety.

[0093] S230, according to the safety distance and the distance information between the vehicle and the dynamic obstacle, determining the safety coefficient of the vehicle and each dynamic obstacle.

[0094] It is easy to understand that the safety coefficient can include a lateral safety coefficient and a longitudinal safety coefficient. The calculation formula of the lateral safety coefficient and the longitudinal safety coefficient of the vehicle and the dynamic obstacle can be represented as:

[0095]

[0096]

[0097] wherein, lon_dis represents the current longitudinal distance between the vehicle and the dynamic obstacle, d min represents the longitudinal safety distance, and lon_safe_coeff represents the longitudinal safety coefficient; lat_dis represents the current lateral distance between the vehicle and the dynamic obstacle, d min,lat represents the lateral safety distance, and lat_safe_coeff represents the lateral safety coefficient.

[0098] S240, according to each safety coefficient, determining the safety loss matched with each dynamic obstacle, taking each safety loss as the prior feature of each dynamic obstacle, inputting the prediction model, and determining the planning path of the autonomous vehicle.

[0099] In the present scheme, optionally, the safety loss matched with each dynamic obstacle is determined according to each safety coefficient, including:

[0100] If the safety coefficient is less than a preset coefficient threshold, the safety loss matched with the dynamic obstacle is determined based on the inverted Gaussian model.

[0101] According to the calculation method of the safety coefficient in S230, the greater the current distance relative to the safety distance, the greater the safety coefficient, and the lower the risk level of the dynamic obstacle. If the safety coefficient is greater than or equal to a preset coefficient threshold, it means that the dynamic obstacle corresponding to the safety coefficient is not dangerous or has a low risk level for the vehicle. The controller can directly set the safety loss of the dynamic obstacle without danger or low risk level to 0, without the need for calculation to calculate the safety loss.

[0102] If the safety coefficient is less than a preset coefficient threshold, the controller can constrain the safety coefficient matched with each dynamic obstacle to between (0, 1) according to the inverted Gaussian model, so as to accelerate the convergence speed of the prediction model in the training process and improve the driving decision accuracy of the prediction model.

[0103] On the basis of the above scheme, optionally, the safety coefficient includes a longitudinal safety coefficient and a lateral safety coefficient.

[0104] Correspondingly, the safety loss matched with the dynamic obstacle is determined based on the inverted Gaussian model, including:

[0105] Based on the inverted Gaussian model, the longitudinal safety loss is determined according to the longitudinal safety coefficient, and the lateral safety loss is determined according to the lateral safety coefficient.

[0106] The safety loss matched with the dynamic obstacle is determined according to the longitudinal safety loss, the lateral safety loss, and a predetermined weight coefficient, wherein the weight coefficient is determined based on a safety loss change rate, and the safety loss change rate is determined based on the distance and relative speed between the vehicle and the dynamic obstacle.

[0107] Specifically, the calculation formula of the longitudinal safety loss and the lateral safety loss can be represented as:

[0108] lon_safe_cost=-1×[1-gaussian(K1-lon_safe_coeff)];

[0109] lat_safe_cost=-1×[1-gaussian(K2-lat_safe_coeff)];

[0110] Wherein, lon_safe_cost represents the longitudinal safety loss, lon_safe_coeff represents the longitudinal safety coefficient, lat_safe_cost represents the lateral safety loss, lat_safe_coeff represents the lateral safety coefficient, gaussian represents Gaussian distribution, K1 and K2 represent preset coefficient thresholds, and K1 and K2 can be the same or different.

[0111] The controller can determine the overall safety loss of the running target according to the lateral safety loss, the longitudinal safety loss, and the weight coefficient. Specifically, the calculation formula of the safety loss can be represented as:

[0112] safe_cost=w_lon_safe×lon_safe_cost+w_lat_safe×lat_safe_cost;

[0113] wherein w_lon_safe represents a weight coefficient of the longitudinal safety loss, and w_lat_safe represents a weight coefficient of the lateral safety loss.

[0114] The weight coefficients can be obtained according to historical safety loss data, for example, the weight coefficient of the lateral safety loss can be set to 0.54, and the weight coefficient of the longitudinal safety loss can be set to 0.46. The weight coefficients can also be determined based on a safety loss change rate. The controller can determine the safety loss change rate according to the current distance between the host vehicle and the dynamic obstacle and the current speed. The safety loss change rate can include a lateral safety loss change rate and a longitudinal safety loss change rate. The lateral safety loss change rate can be determined according to a ratio of the current lateral speed to the current lateral distance between the host vehicle and the dynamic obstacle, and the longitudinal safety loss change rate can be determined according to a ratio of the current longitudinal speed to the current longitudinal distance between the host vehicle and the dynamic obstacle.

[0115] The calculation formula of the longitudinal safety loss change rate and the lateral safety loss change rate can be represented as:

[0116]

[0117]

[0118] wherein lon_speed represents the relative speed of the host vehicle and the dynamic obstacle in the longitudinal direction, and lat_speed represents the relative speed of the host vehicle and the dynamic obstacle in the lateral direction.

[0119] The scheme can set matching weight coefficients for the lateral safety loss and the longitudinal safety loss to calculate the overall safety loss of the dynamic obstacle, which is conducive to accurate evaluation of the safety of the dynamic obstacle.

[0120] In order to achieve more reliable path planning, the controller can also input the safety loss and the safety loss rate as prior features of each dynamic obstacle into a prediction model to determine the planned path of the autonomous vehicle.

[0121] The vehicle motion feature, the navigation map feature, and the obstacle feature of each obstacle are input into the pre-trained prediction model, and the controller can determine the prediction information of the vehicle. The prediction information can include position, orientation angle, speed, acceleration, curvature, acceleration, and time interval. According to the prediction information, the vehicle control device can determine the planned path of the vehicle. At the same time, according to the prediction information output by the prediction model and the current motion information, the vehicle control device can generate the driving decision of the vehicle in the current decision cycle. For example, if the prediction model outputs a vehicle speed of 10 m / s and the current speed of the vehicle is 8 m / s, the vehicle control device can generate an acceleration instruction to control the vehicle power device to increase the speed. According to the driving decision, the vehicle control device can control the vehicle to drive according to the predicted planned path, ensuring the safety and reliability of the vehicle driving.

[0122] In complex driving scenes such as traffic jams, there can be a large number of obstacles within the perception range of the vehicle. Inputting the obstacle features of all obstacles into the prediction model greatly reduces the prediction efficiency and requires sufficient hardware resources. There can also be obstacles within the perception range of the vehicle that have no impact on the driving decision of the vehicle, such as obstacles that do not pose a collision risk to the vehicle. Therefore, the vehicle control device can screen the obstacles within the perception range of the vehicle and select obstacles that have guiding significance for the driving decision of the vehicle as the input of the prediction model.

[0123] Therefore, in a preferred scheme, the controller can determine the weight of the matching of at least one candidate region divided based on the position of the vehicle according to the motion state of the vehicle, and determine the ranking result of the obstacles according to the weight of each candidate region, the prior feature of each obstacle, and the position of each dynamic obstacle. According to the ranking result, the controller can determine at least one dangerous obstacle from the obstacles, and input the vehicle motion feature, the navigation map feature, and the feature of the at least one dangerous obstacle into the pre-trained prediction model to determine the prediction information of the vehicle. The controller can generate a planned path of the autonomous vehicle according to the prediction information of the vehicle.

[0124] The controller can divide one or more candidate regions based on the position of the vehicle. The controller can set the area size of each candidate region according to the driving scene, and set multiple groups of weights for the candidate regions according to the motion state of the vehicle. The motion state can include straight driving, right turning, left turning, and U-turning. Each motion state can correspond to one or more groups of candidate region weights, such as weight group A for straight driving.

[0125] The controller can determine the candidate region to which each obstacle belongs according to the position of each obstacle. According to the prior feature of each obstacle and the weight of the candidate region to which it belongs, the risk assessment value of each obstacle is calculated. The risk assessment values of the obstacles are sorted, and the vehicle control device can obtain the ranking result of the obstacles.

[0126] The controller can select a preset number of obstacles with higher risk assessment values as dangerous obstacles according to the ranking result. For example, the 64 obstacles with the highest risk assessment values in the ranking result are determined as dangerous obstacles. The vehicle control device can also select obstacles with risk assessment values greater than a preset assessment threshold as dangerous obstacles according to the ranking result. For example, obstacles with risk assessment values greater than 0.8 in the ranking result are determined as dangerous obstacles.

[0127] The vehicle control device can take the vehicle motion feature, the navigation map feature, and the dangerous obstacle feature as inputs of the prediction model, and perform operation information prediction of the vehicle to improve the prediction efficiency of the prediction model.

[0128] In a specific example, the determination process of the dangerous target can be as follows:

[0129] (1) Determine at least one candidate region according to the vehicle position.

[0130] After obtaining the prior features of the obstacles, the controller can divide the vehicle perception range into candidate regions according to the vehicle position. The region division manner can be different according to different driving scenarios. Figure 4 is a schematic diagram of candidate region division in a simple driving scenario according to an embodiment of the present application. As shown in Figure 4 in a narrow one-way driving scenario, the controller can divide the vehicle perception range into three candidate regions, candidate region ①, candidate region ②, and candidate region ③, according to the current position of the vehicle. The controller can also divide the vehicle perception range into five candidate regions, candidate region ①, candidate region ②, candidate region ③, candidate region ④, and candidate region ⑤, according to the current position of the vehicle.

[0131] Figure 5 is a schematic diagram of candidate region division in a complex driving scenario according to an embodiment of the present application. In a complex driving scenario such as traffic congestion, the vehicle control device can take the candidate region ⑤ where the vehicle is located as the center region, and divide the vehicle perception range into nine candidate regions as shown in Figure 5 The vehicle control device can set the area size of each candidate region according to the driving scenario.

[0132] It should be noted that the above candidate region division according to the complexity of the driving scenario is only one of the candidate region division manners, and the controller can also divide the candidate regions according to the lane where the vehicle is located, the driving section of the vehicle, and other factors. The present embodiment does not limit the candidate region division manner. The area sizes of the candidate regions can be the same or different, and the shapes of the candidate regions can be regular or irregular, and the present embodiment does not limit the number, size, and shape of the candidate regions.

[0133] (2) According to the motion state of the vehicle, the weight of each candidate region is determined according to a preset weight distribution principle.

[0134] The motion state of the autonomous vehicle can include straight driving, U-turn, left turn, right turn, and reverse driving. The controller can set the weight of each candidate region matching the motion state according to the influence degree of each motion state on each candidate region. For example, as shown in the candidate region division manner Figure 5 , the candidate regions affected by the vehicle in the straight driving state can include ③, ⑤, ⑥, and ⑨, and the influence degree of the candidate regions ③, ⑤, ⑥, and ⑨ can be ranked as ⑤, ⑥, ③, and ⑨. Therefore, the weights of the candidate regions ①-⑨ matching the straight driving state can be 1, 1, 1.5, 1, 2, 1.8, 1, 1, and 1.5, respectively. The controller can also increase the weight of the target candidate region according to the order of the influence of the motion state of the vehicle on each candidate region. Still taking the candidate region division manner Figure 5 as an example, the vehicle in the left turn state affects the candidate regions ⑤, ⑥, ③, ②, and ① in turn. Therefore, the weights of the candidate regions ①-⑨ matching the left turn state can be 1.2, 1.4, 1.6, 1, 2, 1.8, 1, 1, and 1, respectively.

[0135] (3) According to the positions of the obstacles, the candidate regions to which the obstacles belong are determined.

[0136] The controller can compare the positions of the obstacles with the ranges of the candidate regions to determine the candidate regions to which the obstacles belong.

[0137] (4) According to the weights of the candidate regions, the candidate regions to which the obstacles belong, and the safety losses of the obstacles, the ranking results of the obstacles are determined.

[0138] In one possible implementation, the determination of the ranking results of the obstacles according to the weights of the candidate regions, the candidate regions to which the obstacles belong, and the safety losses of the obstacles includes:

[0139] The weights of the candidate regions to which the obstacles belong are determined, and the weighted safety losses of the obstacles are determined according to the weights of the candidate regions to which the obstacles belong and the safety losses of the obstacles.

[0140] The ranking results of the obstacles are determined according to the weighted safety losses of the obstacles.

[0141] The controller can determine the weight of each obstacle safety loss matching according to the weight of each candidate region and the candidate region to which each obstacle belongs. The controller can obtain the weighted safety loss of each obstacle by multiplying the safety loss of each obstacle and the matching weight. The controller can output the sorting result of the obstacles by sorting the weighted safety loss of each obstacle.

[0142] The scheme can determine the weighted safety loss of each obstacle, which is conducive to realizing the reliability of the hazard degree evaluation of the obstacles.

[0143] (5) According to the sorting result, a preset number of obstacles are determined as dangerous obstacles from the obstacles.

[0144] The controller can determine the hazard degree of each obstacle according to the weighted safety loss in the sorting result of the obstacles, and select a preset number of obstacles with relatively high hazard degrees as dangerous obstacles from the obstacles.

[0145] The above scheme selects a certain number of obstacles as dangerous obstacles from the obstacles, which is conducive to improving the safety and reliability of the driving decision while ensuring the timeliness of the driving decision. The technical scheme determines the safety coefficient of the host vehicle and at least one obstacle according to the perception information of the host vehicle, determines the weighted safety loss matching each obstacle according to each safety coefficient and the weight of each candidate region, and inputs each safety loss as a prior feature of each obstacle into a prediction model to determine the planning path of the autonomous vehicle. The scheme solves the problems of low accuracy and high time cost caused by directly extracting abstract features from the perception information by the prediction model, and can improve the prediction accuracy and model generation efficiency of the prediction model while enhancing the explainability of the prediction model.

[0146] Embodiment Three

[0147] Figure 6 A structural schematic diagram of an automatic driving vehicle path planning device provided for Embodiment Three of the present application is shown in FIG. 3. As shown in FIG. 3, the device comprises: Figure 6

[0148] A perception information acquisition module 310 is configured to acquire the perception information of the host vehicle. The perception information comprises the speed information of the host vehicle, the speed information of the dynamic obstacle, the distance information between the host vehicle and the dynamic obstacle, and the response time.

[0149] A safety coefficient determination module 320 is configured to determine the safety coefficient of the host vehicle and at least one dynamic obstacle according to the perception information. The safety coefficient is the ratio of the current distance between the host vehicle and the dynamic obstacle to the safety distance.

[0150] ​The planning path determination module 330 is configured to determine a safety loss matched with each dynamic obstacle according to each safety coefficient, and input the safety loss as a prior feature of each dynamic obstacle into a prediction model to determine a planning path of the autonomous vehicle.

[0151] In the scheme, optionally, the safety coefficient determination module 320 comprises:

[0152] The safety distance determination unit is configured to determine a safety distance between the host vehicle and each dynamic obstacle according to the speed information of the host vehicle, the speed information of the dynamic obstacle, and the response time.

[0153] The safety coefficient determination unit is configured to determine a safety coefficient between the host vehicle and each dynamic obstacle according to the safety distance and distance information between the host vehicle and the dynamic obstacle.

[0154] In one possible implementation, the safety distance comprises a longitudinal distance.

[0155] Correspondingly, the safety distance determination unit comprises:

[0156] The direction determination sub-unit is configured to determine a relative motion direction between the host vehicle and the dynamic obstacle according to the speed information of the host vehicle and the speed information of the dynamic obstacle.

[0157] The distance determination sub-unit is configured to determine a longitudinal safety distance between the host vehicle and each dynamic obstacle according to the relative motion direction, the speed information of the host vehicle, the speed information of the dynamic obstacle, and the response time.

[0158] In the above scheme, the relative motion direction comprises a same direction and a different direction; the speed information comprises speed, lateral speed information, and longitudinal speed information; the longitudinal speed information comprises longitudinal speed, longitudinal deceleration, and longitudinal acceleration; and the lateral speed information comprises lateral speed.

[0159] The distance determination sub-unit is specifically configured to:

[0160] If the relative motion direction between the host vehicle and the dynamic obstacle is the same direction, the distance determination sub-unit is configured to determine the longitudinal safety distance between the host vehicle and the dynamic obstacle according to the longitudinal speed of the preceding traffic target, the longitudinal speed of the following traffic target, the maximum longitudinal deceleration, the maximum longitudinal acceleration, the minimum longitudinal deceleration, and the response time; wherein the maximum longitudinal deceleration, the maximum longitudinal acceleration, and the minimum longitudinal deceleration are determined based on a statistical result of traffic target attribute parameters.

[0161] If the relative movement direction between the vehicle and the dynamic obstacle is opposite, the longitudinal safety distance between the vehicle and the dynamic obstacle is determined according to the longitudinal speed of the vehicle, the longitudinal speed of the dynamic obstacle, the minimum longitudinal deceleration, the maximum longitudinal acceleration, the minimum longitudinal deceleration of the vehicle and the response time.

[0162] In another possible implementation, the safety distance further includes a lateral safety distance; and the lateral speed information further includes a lateral acceleration and a lateral deceleration.

[0163] Correspondingly, the safety distance determination unit is specifically configured to:

[0164] The lateral safety distance between the vehicle and each dynamic obstacle is determined according to the lateral speed of the vehicle, the lateral speed of the dynamic obstacle, the maximum lateral acceleration, the minimum lateral deceleration and the response time.

[0165] In a preferred implementation, the path planning module 330 includes:

[0166] The safety loss determination unit is configured to determine the safety loss matched with the dynamic obstacle based on an inverted Gaussian model if the safety coefficient is less than a preset coefficient threshold.

[0167] On the basis of the above implementation, the safety coefficient includes a longitudinal safety coefficient and a lateral safety coefficient.

[0168] Correspondingly, the safety loss determination unit is specifically configured to:

[0169] The longitudinal safety loss is determined according to the longitudinal safety coefficient based on the inverted Gaussian model, and the lateral safety loss is determined according to the lateral safety coefficient.

[0170] The safety loss matched with the dynamic obstacle is determined according to the longitudinal safety loss, the lateral safety loss and a predetermined weight coefficient; the weight coefficient is determined based on a safety loss change rate, and the safety loss change rate is determined based on distance information between the vehicle and the dynamic obstacle and the speed of the dynamic obstacle.

[0171] The automatic driving vehicle path planning device provided in the embodiments can execute the automatic driving vehicle path planning method provided in any of the embodiments, and has the corresponding function modules and beneficial effects of the execution method.

[0172] Embodiment Four

[0173] Figure 7A structural diagram of an electronic device 410 that can be used to implement embodiments of the present application is shown. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular telephones, smartphones, wearable devices (e.g., headsets, glasses, watches, etc.), and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present application described and / or claimed in this document.

[0174] As shown in Figure 7 The electronic device 410 includes at least one processor 411, and a memory, such as a read-only memory (ROM) 412, a random access memory (RAM) 413, etc., connected to the at least one processor 411, where the memory stores computer programs executable by the at least one processor. The processor 411 can perform various appropriate actions and processes according to the computer programs stored in the read-only memory (ROM) 412 or loaded into the random access memory (RAM) 413 from the storage unit 418. In the RAM 413, various programs and data required for the operation of the electronic device 410 can also be stored. The processor 411, the ROM 412, and the RAM 413 are connected to each other through a bus 414. An input / output (I / O) interface 415 is also connected to the bus 414.

[0175] Various components in the electronic device 410 are connected to the I / O interface 415, including an input unit 416, such as a keyboard, a mouse, etc., an output unit 417, such as various types of displays, a speaker, etc., a storage unit 418, such as a magnetic disk, an optical disk, etc., and a communication unit 419, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 419 allows the electronic device 410 to exchange information / data with other devices through a computer network, such as the Internet, and / or various telecommunication networks.

[0176] The processor 411 can be various general and / or special purpose processing components with processing and computing capabilities. Some examples of the processor 411 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 411 performs various methods and processes described above, such as the autonomous vehicle path planning method.

[0177] In some embodiments, the autonomous vehicle path planning method can be implemented as a computer program tangibly embodied in a computer readable storage medium, e.g., storage unit 418. In some embodiments, parts or all of the computer program can be loaded and / or installed onto electronic device 410 via, e.g., ROM 412 and / or communication unit 419. When the computer program is loaded onto RAM 413 and executed by processor 411, one or more steps of the autonomous vehicle path planning method described above can be performed. Alternatively, in other embodiments, processor 411 can be configured to perform the autonomous vehicle path planning method by other means, e.g., with the aid of firmware.

[0178] Various implementations of the systems and techniques described above can be realized in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a programmable logic device (PLD), a computer hardware, firmware, software, and / or combinations thereof. These various implementations can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0179] Computer programs used to implement the methods of the present application can be written in any combination of one or more programming languages. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program, when executed by the processor of the machine, implements the functions / acts specified in the flowcharts and / or block diagrams. The computer program can be executed entirely on a machine, partially on a machine, partially on a machine as a stand-alone software package, partially on a machine and partially on a remote machine or entirely on a remote machine or server.

[0180] In the context of the present application, a computer-readable storage medium can be a tangible medium that can contain or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. A computer-readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium can be a machine-readable signal medium. More specific examples of a machine-readable storage medium will include one or more lines of a program of instructions in a transitory signal, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0181] To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0182] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), blockchain network, and the Internet.

[0183] The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service.

[0184] It should be understood that the various forms of flow shown above can be used to reorder, add or delete steps. For example, each step described in the present application can be executed in parallel, sequentially or in a different order, as long as the desired results of the technical solutions of the present application can be achieved, which is not limited herein.

[0185] The above detailed description does not constitute a limitation on the scope of protection of the present application. Those skilled in the art should understand that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements and improvements made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A path planning method for autonomous vehicles, characterized in that, The method includes: Acquire the vehicle's perception information; wherein, the perception information includes the vehicle's speed information, the speed information of dynamic obstacles, the distance information between the vehicle and dynamic obstacles, and the response time; Based on the perceived information, a safety factor is determined between the vehicle and at least one dynamic obstacle; wherein the safety factor is the ratio of the current distance between the vehicle and the dynamic obstacle to a safe distance; Based on each safety factor, the safety loss matching each dynamic obstacle is determined, and each safety loss is used as a priori feature of each dynamic obstacle, which is then input into the prediction model to determine the planned path of the autonomous vehicle. The determination of the safety loss matching each dynamic obstacle based on each safety factor includes: If the safety factor is less than a preset threshold, then based on the inverted Gaussian model, the safety loss matching the dynamic obstacle is determined, including: the safety factor includes a longitudinal safety factor and a lateral safety factor; Based on the inverted Gaussian model, the longitudinal safety loss is determined according to the longitudinal safety factor, and the lateral safety loss is determined according to the lateral safety factor. Based on the longitudinal safety loss, the lateral safety loss, and a predetermined weighting coefficient, a safety loss matching the dynamic obstacle is determined; wherein the weighting coefficient is determined based on the safety loss change rate, and the safety loss change rate is determined based on the distance and relative speed between the vehicle and the dynamic obstacle.

2. The method according to claim 1, characterized in that, Determining the safety factor of the vehicle relative to at least one dynamic obstacle based on the perceived information includes: Based on the vehicle's speed information, the speed information of dynamic obstacles, and the response time, determine the safe distance between the vehicle and each dynamic obstacle; Based on the aforementioned safe distance and the distance information between the vehicle and dynamic obstacles, the safety factor between the vehicle and each dynamic obstacle is determined.

3. The method according to claim 2, characterized in that, The safety distance includes the longitudinal distance; Accordingly, determining the safe distance between the vehicle and each dynamic obstacle based on the vehicle's speed information, the speed information of the dynamic obstacle, and the response time includes: Based on the speed information of the vehicle and the speed information of the dynamic obstacle, determine the relative motion direction between the vehicle and the dynamic obstacle; Based on the relative motion direction, the vehicle's speed information, the speed information of the dynamic obstacles, and the response time, the longitudinal safe distance between the vehicle and each dynamic obstacle is determined.

4. The method according to claim 3, characterized in that, The relative motion directions include both in the same direction and opposite directions; the velocity information includes velocity, lateral velocity information, and longitudinal velocity information; wherein, the longitudinal velocity information includes longitudinal velocity, longitudinal deceleration, and longitudinal acceleration; the lateral velocity information includes lateral velocity; The step of determining the longitudinal safe distance between the vehicle and each dynamic obstacle based on the relative motion direction, the vehicle's speed information, the speed information of the dynamic obstacle, and the response time includes: If the relative motion direction between the vehicle and the dynamic obstacle is the same, the longitudinal safe distance between the vehicle and the dynamic obstacle is determined based on the longitudinal velocity of the preceding traffic target, the longitudinal velocity of the following traffic target, the maximum longitudinal deceleration, the maximum longitudinal acceleration, the minimum longitudinal deceleration, and the response time; wherein the maximum longitudinal deceleration, the maximum longitudinal acceleration, and the minimum longitudinal deceleration are determined based on the statistical results of the traffic target attribute parameters. If the relative motion directions of the vehicle and the dynamic obstacle are opposite, the longitudinal safe distance between the vehicle and the dynamic obstacle is determined based on the longitudinal speed of the vehicle, the longitudinal speed of the dynamic obstacle, the minimum longitudinal deceleration, the maximum longitudinal acceleration, the minimum longitudinal deceleration of the vehicle, and the response time.

5. The method according to claim 4, characterized in that, The safety distance also includes a lateral safety distance; the lateral velocity information also includes lateral acceleration and lateral deceleration. Accordingly, determining the safe distance between the vehicle and each dynamic obstacle based on the vehicle's speed information, the speed information of the dynamic obstacle, and the response time includes: Based on the vehicle's lateral velocity, the lateral velocity of the dynamic obstacle, the maximum lateral acceleration, the minimum lateral deceleration, and the response time, determine the safe lateral distance between the vehicle and each dynamic obstacle.

6. A path planning device for an autonomous vehicle, characterized in that, The device includes: The perception information acquisition module is used to acquire the perception information of the vehicle; wherein, the perception information includes the vehicle's speed information, the speed information of dynamic obstacles, the distance information between the vehicle and the dynamic obstacles, and the response time; A safety factor determination module is used to determine the safety factor between the vehicle and at least one dynamic obstacle based on the perceived information; wherein the safety factor is the ratio of the current distance between the vehicle and the dynamic obstacle to a safe distance; The path planning and determination module is used to determine the safety loss matching each dynamic obstacle based on each safety factor, so that each safety loss is used as a priori feature of each dynamic obstacle and input into the prediction model to determine the planned path of the autonomous vehicle. The path planning and determination module includes: A safety loss determination unit is used to determine the safety loss matching the dynamic obstacle based on an inverted Gaussian model if the safety factor is less than a preset coefficient threshold; the safety factor includes a longitudinal safety factor and a lateral safety factor. The safety loss determination unit is specifically used to determine the longitudinal safety loss based on the longitudinal safety coefficient and the lateral safety loss based on the lateral safety coefficient, according to the inverted Gaussian model; and to determine the safety loss matching the dynamic obstacle based on the longitudinal safety loss, the lateral safety loss and the predetermined weighting coefficients; wherein the weighting coefficients are determined based on the safety loss change rate, and the safety loss change rate is determined based on the distance and relative speed between the vehicle and the dynamic obstacle.

7. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the autonomous vehicle path planning method according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the autonomous vehicle path planning method according to any one of claims 1-5.

Images