Vehicle maneuvering path following control method and system

Abstract

The present invention discloses a vehicle maneuvering path following control method and system. For emergency steering of the vehicle, a plurality of transition preview points are preset between main preview points. A first interval distance between the current position of the vehicle determined during the emergency steering and each acquired transition preview point, as well as a second interval distance between each transition preview point and a previous target main preview point, are input into a pre-trained neural network to obtain a transition level of each transition preview point selected as an intermediate transition node of a main preview following path. The transition preview point with the highest transition level is used as the intermediate transition node of the vehicle's current main preview following path. The vehicle travels along a local following path from the current vehicle position through the intermediate transition nodes determined above to the previous target main preview point, thereby effectively reducing the impact of sudden changes in path curvature.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic driving of intelligent vehicles, and more particularly to a vehicle maneuvering path following control method and system. Background Art

[0002] Intelligent vehicle technology mainly includes technologies such as intelligent vehicle maneuvering path tracking, lane keeping, and vehicle lane changing. The purpose of intelligent vehicle maneuvering path tracking control is to make the vehicle drive along the desired maneuvering path. In the existing technology, vehicle maneuvering path tracking is achieved through preview tracking technology. The preview tracking technology inputs the road path, sets a preview point in front of the vehicle to simulate the driver's driving line of sight, analyzes the relative state of the preview point and the path, and combines the current driving posture of the vehicle to output the steering wheel angle to achieve tracking. The patent application case with Chinese patent application number 202210643965.7 discloses a vehicle trajectory tracking control method. In this scheme, the current point of the vehicle and the preview point are obtained respectively. trajectory parameters, and determining the lateral control amount and vehicle posture control amount required for the vehicle to move from the current point to the preview point based on the trajectory parameters; then determining a vehicle posture control compensation amount based on the lateral control amount and the vehicle posture control amount; then controlling the steering wheel angle of the vehicle based on the lateral control amount, and applying braking force to the vehicle wheels based on the vehicle posture control compensation amount to control the vehicle to travel along the vehicle trajectory where the preview point is located. By applying braking force to the vehicle wheels, the deviation of the vehicle posture caused by controlling the steering wheel angle of the vehicle according to the lateral control amount is compensated. However, in the above scheme, when the vehicle handles an emergency turn, the path will produce a sudden change in curvature, affecting the stability of the vehicle. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a vehicle maneuvering path following control method and system, so as to reduce the influence of sudden change of path curvature when the vehicle makes an emergency turn during the vehicle maneuvering path tracking process.

[0004] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0005] In one aspect, the present invention provides a vehicle maneuvering path following control method, which comprises the following steps:

[0006] Obtaining a main preview following path that the vehicle needs to follow, wherein a plurality of transition preview points are set between two adjacent main preview points of the main preview following path;

[0007] Determining whether the vehicle needs to make an emergency turn while the vehicle is traveling along the main preview-following path;

[0008] When the judgment result is yes, all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed by are obtained;

[0009] Determine a first distance between the current position of the vehicle and each acquired transition preview point and a second distance between each transition preview point and a previous target main preview point;

[0010] Inputting the first interval distance between the determined vehicle current position and each acquired transition preview point and the second interval distance between each transition preview point and the previous target main preview point into a pre-trained neural network, obtaining a transition level at which each transition preview point is selected as an intermediate transition node of the main preview following path;

[0011] The transition preview point with the highest transition level is used as the intermediate transition node of the vehicle's current main preview following path. The vehicle travels along the local following path from the current vehicle position through the intermediate transition node determined above to the previous target main preview point.

[0012] Wherein, before obtaining all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed, the following steps are also included:

[0013] Determine whether the current vehicle speed exceeds the preset value;

[0014] If the judgment result is yes, then start to obtain all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed;

[0015] If the judgment result is no, the vehicle continues to follow the previous main preview path.

[0016] Wherein, after obtaining all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed, the following steps are also included:

[0017] Taking the vehicle's current position as the starting node, search for the transition preview point closest to the vehicle's position, and perform obstacle avoidance detection on the transition preview point. If the detection passes, it will be used as an optional transition preview point. If the detection fails, the transition preview point will be abandoned; continue searching for the next transition preview point until the transition preview point obtained is traversed.

[0018] Among them, the obstacle avoidance detection is to obtain the position of the obstacle in front of the vehicle through the radar, and determine whether the position of the transition preview point is within the safe range. If the judgment result is yes, the detection passes, otherwise, the detection fails.

[0019] Wherein, the neural network may be a back propagation neural network.

[0020] In another aspect, the present invention provides a vehicle maneuvering path following control system, comprising:

[0021] A main preview following path acquisition processing unit is used to acquire a main preview following path that the vehicle needs to follow, wherein a plurality of transition preview points are set between two adjacent main preview points of the main preview following path;

[0022] a judgment processing unit for judging whether the vehicle needs to make an emergency turn while the vehicle is traveling along the main preview-following path;

[0023] A transition preview point acquisition processing unit is used to acquire all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed when the judgment result is yes;

[0024] an interval distance determination processing unit, configured to determine a first interval distance between the vehicle's current position and each acquired transition preview point, and a second interval distance between each transition preview point and a previous target main preview point;

[0025] a transition level determination processing unit, configured to input a first distance between the determined current position of the vehicle and each acquired transition preview point, and a second distance between each transition preview point and a previous target main preview point, into a pre-trained neural network, and obtain a transition level at which each transition preview point is selected as an intermediate transition node of a main preview following path;

[0026] The control processing unit is used to use the transition preview point with the highest transition level as the intermediate transition node of the vehicle's current main preview following path, and the vehicle travels along the local following path from the current vehicle position through the intermediate transition node determined above to the previous target main preview point.

[0027] Additionally, it includes:

[0028] The vehicle speed judgment processing unit is used to judge whether the current vehicle speed exceeds the preset value; if the judgment result is yes, it starts to obtain all transition preview points between the vehicle's current position and the previous target main preview point passed by the vehicle; if the judgment result is no, it continues to drive according to the previous main preview following path.

[0029] Additionally, it includes:

[0030] The obstacle avoidance detection processing unit is used to search for the transition preview point closest to the vehicle position with the vehicle's current position as the starting node, and perform obstacle avoidance detection on the transition preview point. If the detection passes, it will be used as an optional transition preview point. If the detection fails, the transition preview point will be abandoned; the search for the next transition preview point will continue until the acquired transition preview points are traversed.

[0031] Among them, the obstacle avoidance detection processing unit obtains the position of the obstacle in front of the vehicle through radar, and determines whether the position of the transition preview point and the obstacle are within the safe range. If the judgment result is yes, the detection passes, otherwise, the detection fails.

[0032] Wherein, the neural network is a back propagation neural network.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] In the method and system of the present invention, for emergency steering of a vehicle, a number of transition preview points are preset between main preview points, a first interval distance between the current position of the vehicle determined when the vehicle makes an emergency steering and each acquired transition preview point, and a second interval distance between each transition preview point and a previous target main preview point are input into a pre-trained neural network, and a transition level of each transition preview point selected as an intermediate transition node of a main preview following path is obtained; the transition preview point with the highest transition level is used as the intermediate transition node of the current main preview following path of the vehicle, and the vehicle travels along a local following path from the current vehicle position through the intermediate transition nodes determined above to the previous target main preview point, which can effectively reduce the impact of sudden changes in path curvature. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flow chart of a specific embodiment of the vehicle maneuvering path following control method of the present invention;

[0036] Figure 2 Schematic diagram of the structure of a neural network trained in the vehicle maneuvering path following control method of the present invention;

[0037] Figure 3 This is a block diagram of the first specific embodiment of the vehicle maneuvering path following control system of the present invention;

[0038] Figure 4 This is a block diagram of a second specific embodiment of the vehicle maneuvering path following control system of the present invention;

[0039] Figure 5 It is a block diagram of the composition of the third specific embodiment of the vehicle maneuvering path following control system of the present invention. DETAILED DESCRIPTION

[0040] refer to Figure 1 , which is a flow chart of a specific embodiment of the vehicle maneuvering path following control method of the present invention. The method of this embodiment mainly includes the following steps:

[0041] In step S101, a main preview following path that the vehicle needs to follow is obtained. A number of transition preview points are set between two adjacent main preview points of the main preview following path. In this embodiment, for emergency vehicle steering, a number of transition preview points are set between two adjacent main preview points of the main preview following path. In the prior art, the preview points are usually set according to the preview distance. Therefore, the setting of the transition preview points can also be set according to the preview distance. However, as a preferred embodiment, different from the prior art, the transition preview points in this embodiment are not set according to the preview distance, but are set according to the vehicle driving hotspot between two adjacent main preview points. That is, in the area between two adjacent main preview points, several vehicle driving hot spots are selected as transition preview points. In specific implementation, the vehicle driving hot spots can be determined based on the vehicle driving data statistics in the most recent period, which will not be elaborated here. The vehicle driving hot spots are selected as transition preview points. Since they are vehicle driving hot spots, most vehicles have already been driving on the hot spot path, which means that there is a high probability that there will be no obstacles or the road conditions are good on the vehicle driving hot spot. The probability of subsequent vehicles making emergency turns and other emergencies is very small. Therefore, it is also beneficial to reduce the sudden change in the curvature of the subsequent vehicle driving path, so that the vehicle maintains stable driving and improves the safety of subsequent vehicles.

[0042] In step S102, the vehicle determines whether an emergency turn is required while traveling along the main preview-following path. Specifically, if an obstacle is detected on the path ahead of the vehicle, it can be determined that an emergency turn is required, or if other emergency conditions occur, it can be determined that the vehicle needs an emergency turn. No specific limitations are given here.

[0043] In step S103, if the result of the judgment is yes, all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed are obtained. It should be noted that, since when the vehicle speed is not high, even if an emergency turn is performed, a large sudden change in path curvature will not occur, and the vehicle can continue along the original following path without planning a new following path. Therefore, before obtaining all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed, this embodiment may further include:

[0044] Determine whether the current vehicle speed exceeds the preset value; if so, start acquiring all transition preview points between the vehicle's current position and the previous target main preview point the vehicle passed; if not, continue driving along the previous main preview path;

[0045] In addition, after obtaining all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed, the obtained transition preview points can be screened to discard unqualified transition preview points in advance, thereby improving the efficiency of system processing. As a preferred embodiment, the screening can be performed in the following manner, namely:

[0046] Taking the vehicle's current position as the starting node, search for the transition preview point closest to the vehicle's position, and perform obstacle avoidance detection on the transition preview point. If the detection passes, it will be used as an optional transition preview point. If the detection fails, the transition preview point will be abandoned; continue searching for the next transition preview point until the transition preview point obtained is traversed.

[0047] It should be noted that the obstacle avoidance detection described in the above embodiment can obtain the position of the obstacle in front of the vehicle through the radar, and determine whether the position of the transition preview point is within the safe range. If the judgment result is yes, the detection passes, otherwise, the detection fails, which will not be repeated here.

[0048] In step S104, the first interval distance between the current position of the vehicle and each acquired transition preview point and the second interval distance between each transition preview point and the previous target main preview point are determined. It should be noted that reducing the sudden change in path curvature can be achieved by smoothing the path, and the path can be smoothed by performing minimum optimization processing on the mutual distance relationship of each node that is about to be passed in the path. Therefore, in this embodiment, two distance parameters, namely the first interval distance between the current position of the vehicle and each acquired transition preview point and the second interval distance between each transition preview point and the previous target main preview point, are selected as input data of the neural network, and the distance minimum optimization processing is performed through the neural network, that is, the neural network determines through which transition preview point the vehicle deviates the least from the nearest main preview point. The smoother the path, the smaller the sudden change in path curvature and the higher the transition level.

[0049] In step S105, the first interval distance between the determined current position of the vehicle and each acquired transition preview point and the second interval distance between each transition preview point and the previous target main preview point are input into a pre-trained neural network to obtain the transition level of each transition preview point selected as the intermediate transition node of the main preview following path. In specific implementation, the transition level can be divided into five levels, such as best, good, medium, poor, and worst. In addition, in this embodiment, the transition level corresponds to the path smoothness as the judgment standard. The best path smoothness corresponds to the best transition level, and the good path smoothness corresponds to the good transition level. The medium, poor, and worst transition levels are obtained in this order, which will not be repeated here.

[0050] It should be noted that the transition level in this embodiment is determined by a pre-trained neural network. A neural network is a computational model composed of a large number of nodes (or neurons) that are directly interconnected. The neural network can output corresponding classifications based on input data. In this embodiment, the first interval distance and the second interval distance related to a transition preview point can be input through the neural network to output the transition level corresponding to the transition preview point. In order to implement the pre-trained neural network, as a preferred specific embodiment, taking the back propagation neural network as an example, the neural network in this embodiment adopts a three-layer network structure, including an input layer, a hidden layer and an output layer. The first interval distance between the current position of the vehicle and the transition preview point and the second interval distance between the transition preview point and the previous target main preview point are used as input data of the input layer, and the output layer outputs the transition level corresponding to the transition preview point. The specific network model of the neural network is as follows: Figure 2 As shown, when pre-training, establish Figure 2 After the neural network model is shown, it is necessary to train the neural network model through the training set and the validation set. First, 100 groups of observation data can be collected, 80 groups of which are used as training sets and 20 groups as validation sets. During training, the neural network optimizes the two input distance values ​​of the first interval distance between the transition preview point and the current position of the vehicle and the second interval distance between the transition preview point and the previous target main preview point related to each preview transition point in the 80 groups of training sets according to the minimum distance. That is, the neural network determines through which transition preview point the vehicle deviates least from the nearest main preview point. The smoother the path, the smaller the sudden change in path curvature, and the higher the transition level. According to the trained neural network, the transition level of each preview transition point can be determined according to the first interval distance and the second interval distance related to each preview transition point. No further details are given here.

[0051] Finally, in step S106, the transition preview point with the highest transition level is used as the intermediate transition node of the vehicle's current main preview following path, and the vehicle travels along the local following path from the current vehicle position through the intermediate transition node determined above to the previous target main preview point.

[0052] In the above embodiment, for emergency vehicle turns, since a number of transition preview points are preset between the main preview points, when determining the emergency vehicle turn, all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed are obtained, and the first interval distance between the current position of the vehicle and each of the obtained transition preview points and the second interval distance between each of the transition preview points and the previous target main preview point are used as inputs of the neural network to obtain the transition level of each transition preview point selected as the intermediate transition node of the main preview following path; the transition preview point with the highest transition level is used as the intermediate transition node of the current main preview following path of the vehicle, and the vehicle travels along the local following path from the current vehicle position through the above-determined intermediate transition nodes to the previous target main preview point, which can effectively reduce the impact of sudden changes in path curvature.

[0053] refer to Figure 3 This figure is a block diagram of the first specific embodiment of the vehicle maneuvering path following control system of the present invention. The system of this embodiment mainly includes: a main preview following path acquisition processing unit 101, a judgment processing unit 102, a transition preview point acquisition processing unit 103, a separation distance determination processing unit 104, a transition level determination processing unit 105 and a control processing unit 106, wherein

[0054] The main preview following path acquisition processing unit 101 is mainly used to obtain the main preview following path that the vehicle needs to follow. A number of transition preview points are set between two adjacent main preview points of the main preview following path. In this embodiment, for emergency vehicle steering, a number of transition preview points are set between two adjacent main preview points in the main preview following path. In the prior art, the preview points are usually set according to the preview distance. Therefore, the setting of the transition preview points can also be set according to the preview distance. However, as a preferred embodiment, different from the prior art, the transition preview points in this embodiment are not set according to the preview distance, but are set according to the vehicle driving hotspot between two adjacent main preview points, that is, in the area between two adjacent main preview points, a number of vehicle driving hotspots are selected as transition preview points. In specific implementation, the vehicle driving hotspot can be determined based on the vehicle driving data statistics in a recent period of time, which will not be repeated here.

[0055] The judgment processing unit 102 is mainly used to determine whether the vehicle needs to make an emergency turn while the vehicle is traveling along the main preview-following path. In specific implementation, if an obstacle is detected on the path ahead of the vehicle, the judgment processing unit 102 can determine that an emergency turn is required, or when other emergency situations occur, the judgment processing unit 102 can determine that an emergency turn is required. No specific limitation is made here.

[0056] The transition preview point acquisition processing unit 103 is mainly used to acquire all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed when the judgment result is yes.

[0057] It should be noted that when the vehicle speed is not high, even if an emergency turn is made, there will be no large sudden change in the path curvature. The vehicle can follow the original path without planning a new path. Therefore, Figure 4 As a preferred embodiment, the vehicle maneuvering path following control system of the present invention may further include: a vehicle speed judgment processing unit 107, which is used to judge whether the current vehicle speed exceeds a preset value; if the judgment result is yes, it starts to obtain all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed; if the judgment result is no, it continues to travel according to the previous main preview following path.

[0058] In addition, after obtaining all the transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed, the obtained transition preview points can be screened to discard unqualified transition preview points in advance to improve the efficiency of system processing. As a preferred embodiment, refer to Figure 5 The vehicle maneuvering path following control system of the present invention may further include: an obstacle avoidance detection processing unit 108, which is used to use the current position of the vehicle as the starting node, search for the transition preview point closest to the vehicle position, and perform obstacle avoidance detection on the transition preview point. If the detection passes, it is used as an optional transition preview point. If the detection fails, the transition preview point is abandoned; and the search for the next transition preview point continues until the acquired transition preview points are traversed.

[0059] It should be noted that the obstacle avoidance detection processing unit 108 can obtain the position of the obstacle in front of the vehicle through the radar and determine whether the position of the transition preview point is within the safe range. If the judgment result is yes, the detection passes, otherwise, the detection fails, which will not be repeated here.

[0060] The interval distance determination processing unit 104 is mainly used to determine the first interval distance between the current position of the vehicle and each acquired transition preview point, and the second interval distance between each transition preview point and the previous target main preview point. As mentioned above, reducing the sudden change in path curvature can be achieved by smoothing the path, and smoothing the path can be achieved by performing minimum optimization processing on the mutual distance relationship of each node to be passed in the path. Therefore, in this embodiment, two distance parameters, namely the first interval distance between the current position of the vehicle and each acquired transition preview point, and the second interval distance between each transition preview point and the previous target main preview point, are selected as input data of the neural network, and the distance minimum optimization processing is performed through the neural network, that is, the neural network determines through which transition preview point the vehicle deviates the least from the nearest main preview point. The smoother the path, the smaller the sudden change in path curvature, and the higher the transition level.

[0061] The transition level determination processing unit 105 is mainly used to input the first interval distance between the determined current position of the vehicle and each acquired transition preview point and the second interval distance between each transition preview point and the previous target main preview point into a pre-trained neural network, so as to obtain the transition level of each transition preview point selected as the intermediate transition node of the main preview following path. In specific implementation, the transition level can be divided into five levels, such as best, good, medium, poor, and worst. In addition, in this embodiment, the transition level corresponds to the path smoothness as the judgment standard. The best path smoothness corresponds to the best transition level, and the good path smoothness corresponds to the good transition level. The medium, poor, and worst transition levels are obtained in this order, which will not be repeated here.

[0062] It should be noted that in this embodiment, the transition level determination processing unit 105 judges the transition level through a pre-trained neural network. The neural network is a computing model composed of a large number of nodes (or neurons) that are directly connected to each other. The neural network can output the corresponding classification according to the input data. In this embodiment, the first interval distance and the second interval distance related to a transition preview point can be input through the neural network to output the transition level corresponding to the transition preview point. In order to realize the pre-training of the neural network, as a preferred specific embodiment, the neural network adopts a back propagation neural network as an example. The neural network in this embodiment adopts a three-layer network structure, including an input layer, a hidden layer and an output layer. The first interval distance between the current position of the vehicle and the transition preview point and the second interval distance between the transition preview point and the previous target main preview point are used as input data of the input layer, and the output layer outputs the transition level corresponding to the transition preview point. The specific network model of the neural network is as follows: Figure 2 As shown, no further details are given here.

[0063] The control processing unit 106 is mainly used to use the transition preview point with the highest transition level as the intermediate transition node of the vehicle's current main preview following path, and the vehicle travels along the local following path from the current vehicle position through the intermediate transition node determined above to the previous target main preview point.

[0064] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A vehicle maneuvering path following control method, characterized in that: The steps include: Obtain a main preview path that the vehicle needs to follow. A number of transition preview points are set between two adjacent main preview points on the main preview path. In the area between the two adjacent main preview points, a number of vehicle driving hotspots are selected as transition preview points. The vehicle driving hotspots are determined based on vehicle driving data statistics over a recent period of time. Determining whether the vehicle needs to make an emergency turn while the vehicle is traveling along the main preview-following path; When the judgment result is yes, all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed by are obtained; Determine a first distance between the current position of the vehicle and each acquired transition preview point and a second distance between each transition preview point and a previous target main preview point; Inputting the first interval distance between the determined vehicle current position and each acquired transition preview point and the second interval distance between each transition preview point and the previous target main preview point into a pre-trained neural network, obtaining a transition level at which each transition preview point is selected as an intermediate transition node of the main preview following path; The transition preview point with the highest transition level is used as the intermediate transition node of the vehicle's current main preview following path. The vehicle travels along the local following path from the current vehicle position through the intermediate transition node determined above to the previous target main preview point.

2. The method according to claim 1, characterized in that Before obtaining all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed through, it also includes: Determine whether the current vehicle speed exceeds the preset value; If the judgment result is yes, then start to obtain all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed; If the judgment result is no, the vehicle continues to follow the previous main preview path.

3. The method according to claim 1, characterized in that After obtaining all transition preview points between the vehicle's current position and the previous target main preview point the vehicle passed, the following steps are also included: Taking the vehicle's current position as the starting node, search for the transition preview point closest to the vehicle's position, and perform obstacle avoidance detection on the transition preview point. If the detection passes, it will be used as an optional transition preview point. If the detection fails, the transition preview point will be abandoned; continue searching for the next transition preview point until the transition preview point obtained is traversed.

4. The method according to claim 3, characterized in that The obstacle avoidance detection is to obtain the position of the obstacle in front of the vehicle through the radar, and determine whether the position of the transition preview point is within the safe range. If the judgment result is yes, the detection passes, otherwise, the detection fails.

5. The method according to claim 1, wherein The neural network is a back propagation neural network.

6. A vehicle maneuvering path following control system, characterized in that: include: A main preview following path acquisition processing unit is used to acquire a main preview following path that the vehicle needs to follow. A plurality of transition preview points are set between two adjacent main preview points of the main preview following path. In the area between the two adjacent main preview points, a plurality of vehicle driving hot spots are selected as transition preview points. The vehicle driving hot spots are determined based on the statistics of vehicle driving data in a recent period of time. a judgment processing unit for judging whether the vehicle needs to make an emergency turn while the vehicle is traveling along the main preview-following path; A transition preview point acquisition processing unit is used to acquire all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed when the judgment result is yes; an interval distance determination processing unit, configured to determine a first interval distance between the vehicle's current position and each acquired transition preview point, and a second interval distance between each transition preview point and a previous target main preview point; a transition level determination processing unit, configured to input a first distance between the determined current position of the vehicle and each acquired transition preview point, and a second distance between each transition preview point and a previous target main preview point, into a pre-trained neural network, and obtain a transition level at which each transition preview point is selected as an intermediate transition node of a main preview following path; The control processing unit is used to use the transition preview point with the highest transition level as the intermediate transition node of the vehicle's current main preview following path, and the vehicle travels along the local following path from the current vehicle position through the intermediate transition node determined above to the previous target main preview point.

7. The system according to claim 6, characterized in that Also includes: A vehicle speed determination processing unit, used to determine whether the current vehicle speed exceeds a preset value; If the judgment result is yes, then start acquiring all transition preview points between the current position of the vehicle and the previous target main preview point that the vehicle passed; if the judgment result is no, continue to drive according to the previous main preview following path.

8. The system according to claim 6, wherein: Also includes: The obstacle avoidance detection processing unit is used to search for the transition preview point closest to the vehicle position with the vehicle's current position as the starting node, and perform obstacle avoidance detection on the transition preview point. If the detection passes, it will be used as an optional transition preview point. If the detection fails, the transition preview point will be abandoned; the search for the next transition preview point will continue until the acquired transition preview points are traversed.

9. The system according to claim 8, characterized in that The obstacle avoidance detection processing unit obtains the position of the obstacle in front of the vehicle through radar and determines whether the position of the transition preview point is within the safe range. If the judgment result is yes, the detection passes, otherwise, the detection fails.

10. The system according to claim 6, wherein: The neural network is a back propagation neural network.

Images