A path planning method, device, equipment, and medium for unmanned vehicles.

By constructing historical trajectory sequences and calculating selection probabilities, and combining gravity and repulsion to control the movement of unmanned vehicles, the problem of unreasonable path planning was solved, and more reasonable and safer path planning was achieved.

CN116989813BActive Publication Date: 2026-11-17SHENZHEN RES INST CENT SOUTH UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
CN202310965532.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-08-02
Publication Date
2026-11-17
Estimated Expiration
2043-08-02

AI Technical Summary

Technical Problem

Existing autonomous vehicle path planning methods are prone to getting stuck in local optima and unreachable objectives in complex environments, failing to effectively consider the mutual influence between vehicles and actual driving scenarios, resulting in unreasonable path planning.

Method used

By constructing a historical trajectory sequence, randomly selecting candidate trajectory points, calculating the selection probabilities of direction, distance, and speed, and combining the gravitational force of the target trajectory point and the repulsive force of obstacles, the traction force is obtained to control the unmanned vehicle to move to the target trajectory point.

Benefits of technology

It improves the rationality and safety of path planning, enabling unmanned vehicles to effectively avoid obstacles and achieve more rational and safer movement trajectories.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN116989813B_ABST
    Figure CN116989813B_ABST
Patent Text Reader

Abstract

This application relates to the field of intelligent transportation technology and provides a path planning method, apparatus, device, and medium for unmanned vehicles. The path planning method for unmanned vehicles includes: constructing a historical trajectory sequence based on the trajectory points of the unmanned vehicle at T time points; randomly selecting candidate trajectory points from N time points after the T-th time point based on the trajectory point of the unmanned vehicle at the current time point; obtaining the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point; calculating the comprehensive selection probability of the candidate trajectory points and selecting the target trajectory point of the unmanned vehicle based on the comprehensive selection probability; during the unmanned vehicle's movement towards the target trajectory point, obtaining the traction force of the unmanned vehicle based on the attraction of the target trajectory point to the unmanned vehicle and the repulsive force of obstacles to the unmanned vehicle, and using the traction force to control the unmanned vehicle to move to the target trajectory point. The path planning method provided in this application can improve the rationality of path planning for unmanned vehicles.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of intelligent transportation technology, and in particular to a path planning method, device, equipment and medium for unmanned vehicles. Background Technology

[0002] Over the past few years, we have witnessed unprecedented growth in autonomous vehicles and their application across various sectors, such as logistics, ride-sharing, and sanitation. A long-term goal for autonomous vehicle applications is to build an intelligent system that can perform various tasks independently without human intervention. In this work, we seek to develop a technology that allows autonomous vehicles to automatically plan their movement trajectory from any initial point in large-scale, complex environments (e.g., congested urban areas) until they reach their destination.

[0003] Common path planning schemes include artificial potential fields, fast exploratory random trees, approximate obstacle avoidance algorithms, and fuzzy inference. Among these, the artificial potential field method has the simplest principle and the ability to acquire environmental information in real time. Therefore, it has a significant advantage in complex environments. However, this scheme also has certain limitations, easily getting trapped in local optima and unreachable objectives. Modern autonomous vehicle path planning algorithms must be able to handle various driving scenarios to conform to real-world conditions. Path planning directly affects the feasibility and safety of vehicle trajectory. In real-world driving scenarios, vehicles interact and influence each other, but existing path planning methods do not take this into account. Therefore, there is a problem of unreasonable path planning for autonomous vehicles. Summary of the Invention

[0004] This application provides a path planning method, apparatus, device, and medium for unmanned vehicles, which can solve the problem of unreasonable path planning for unmanned vehicles.

[0005] In a first aspect, embodiments of this application provide a path planning method for unmanned vehicles, the path planning method comprising:

[0006] A historical trajectory sequence is constructed based on the trajectory points of the unmanned vehicle at time T; the Tth time is the current time.

[0007] Based on the trajectory points of the unmanned vehicle at the current moment, candidate trajectory points are randomly selected for N moments after the Tth moment.

[0008] Based on the historical trajectory sequence, obtain the direction selection probability, distance selection probability, and speed selection probability for each candidate trajectory point; the direction selection probability is used to represent the probability that the candidate trajectory point is selected based on its offset angle; the distance selection probability is used to represent the probability that the candidate trajectory point is selected based on its distance from other candidate trajectory points; the speed selection probability is used to represent the probability that the candidate trajectory point is selected based on its speed at the candidate trajectory point.

[0009] Based on the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point, the comprehensive selection probability of each candidate trajectory point is calculated, and the target trajectory point that the unmanned vehicle will pass through from the trajectory point at the current moment to the destination is selected based on the calculated comprehensive selection probability.

[0010] During the process of the unmanned vehicle moving towards the target trajectory point, the traction force of the unmanned vehicle is obtained based on the gravitational force of the target trajectory point on the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle, and the unmanned vehicle is controlled to move to the target trajectory point based on the traction force.

[0011] Optionally, the historical trajectory sequence TS is:

[0012] TS = {LP1,...,LP} t ,...,LP T}

[0013] Where LP1 represents the trajectory point of the autonomous vehicle at the first moment, LP t Let LP represent the trajectory point of the unmanned vehicle at time t. T Let T represent the trajectory point of the unmanned vehicle at time T, where t = 1, 2, ..., T, and T represents the current time.

[0014] Optionally, based on the historical trajectory sequence, obtain the direction selection probability, distance selection probability, and velocity selection probability for each candidate trajectory point, including:

[0015] For each candidate trajectory point, perform the following steps;

[0016] The offset angle of the unmanned vehicle at the candidate trajectory point is calculated to obtain the direction selection probability of the candidate trajectory point;

[0017] The distance selection probability of a candidate trajectory point is obtained based on the historical trajectory sequence and the distance between the candidate trajectory point and other candidate trajectory points.

[0018] The speed selection probability of the candidate trajectory point is obtained based on the speed of the unmanned vehicle at the candidate trajectory point.

[0019] Optionally, the offset angle of the unmanned vehicle at the candidate trajectory point is calculated to obtain the direction selection probability of the candidate trajectory point, including:

[0020] Through the formula:

[0021]

[0022] Calculate the difference between the offset angle of the j-th candidate trajectory point at time T+n and the offset angle of the j-th candidate trajectory point at time (T+n)-1.

[0023] in, This represents the offset angle of the j-th candidate trajectory point at time T+n. Let represent the offset angle of the j-th candidate trajectory point at time (T+n)-1. If n = T+1, then... Let represent the offset angle of the trajectory point at time T, n = 1, 2, ..., N, where N represents the Nth time after the current time T, j = 1, 2, ..., Q, where Q represents the total number of candidate trajectory points at each time.

[0024] Through the formula:

[0025]

[0026] Calculate the direction selection probability of the j-th candidate trajectory point at time T+n.

[0027] Where τ is a preset value;

[0028] Based on the historical trajectory sequence and the distance between candidate trajectory points and other candidate trajectory points, the distance selection probability of candidate trajectory points is obtained, including:

[0029] Through the formula:

[0030]

[0031] Calculate the relative distance selection probability of the j-th candidate trajectory point at time T+n.

[0032] in, This represents the distance between the j-th candidate trajectory point at time T+n and the j-th candidate trajectory point at time (T+n)-1. This represents the j-th candidate trajectory point at time T+n. Let represent the j-th candidate trajectory point at time (T+n)-1, where n = 1, 2, ..., N, N represents the Nth time after the current time T, j = 1, 2, ..., Q, Q represents the total number of candidate trajectory points at the same time, and σ, e, and μ are all constants;

[0033] Through the formula:

[0034]

[0035] Calculate the probability of selecting the j-th candidate trajectory point by the movement distance at time T+n.

[0036] Where, di(LP) T-1 ,LP T ) represents the distance between the trajectory point at time T and the trajectory point at time T-1, LP T Let LP represent the trajectory point at time T. T-1 This represents the trajectory point at time T-1;

[0037] Through the formula:

[0038]

[0039] Calculate the distance selection probability of the j-th candidate trajectory point at time T+n.

[0040] Based on the speed of the unmanned vehicle at the candidate trajectory points, the speed selection probability of the candidate trajectory points is obtained, including:

[0041] Through the formula:

[0042]

[0043] Calculate the velocity of the unmanned vehicle at the j-th candidate trajectory point at time T+n.

[0044] Among them, t T+n This represents the value at the (T+n)th time point, t. (T+n)-1 This represents the value at the (T+n)-1th time point;

[0045] Through the formula:

[0046]

[0047] Calculate the velocity selection probability of the j-th candidate trajectory point at time T+n.

[0048] in, LP represents the average velocity of the historical trajectory sequence, LP1 represents the trajectory point at time 1, LP T This represents the trajectory point at time T.

[0049] Optionally, based on the direction selection probability, distance selection probability, and velocity selection probability of each candidate trajectory point, a comprehensive selection probability is calculated, including:

[0050] Through the formula:

[0051]

[0052] Calculate the overall selection probability of the j-th candidate trajectory point at time T+n.

[0053] in, This represents the j-th candidate trajectory point at time (T+n)-1. This represents the j-th candidate trajectory point at time T+n. Let represent the direction selection probability of the j-th candidate trajectory point at time T+n. Let represent the distance selection probability of the j-th candidate trajectory point at time T+n. This represents the velocity selection probability of the j-th candidate trajectory point at time T+n.

[0054] Optionally, the combined selection probability f of all candidate trajectory points at time T+n. T+n for:

[0055]

[0056] in, This represents the first candidate trajectory point at time (T+n)-1. This represents the first candidate trajectory point at time T+n. This represents the overall selection probability of the first candidate trajectory point at time T+n. This represents the overall selection probability of the Qth candidate trajectory point at time T+n;

[0057] Based on the calculated comprehensive selection probability, the target trajectory points that the unmanned vehicle will pass through from the current trajectory point to the destination are selected, including:

[0058] The combined selection probability f of all candidate trajectory points at time T+n T+n The candidate trajectory point with the highest comprehensive selection probability is taken as the target trajectory point at time T+n.

[0059] Optionally, during the movement of the unmanned vehicle towards the target trajectory point, the traction force of the unmanned vehicle is obtained based on the gravitational force of the target trajectory point on the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle, including:

[0060] Through the formula:

[0061]

[0062] Calculate the unmanned vehicle on the T+n m The benefits of action strategies at each moment Action strategy is a combination of the environment state and the action state selected in that environment state;

[0063] in, This indicates that the unmanned vehicle is in the T+n stage. m Action strategy at each moment, s T+m Represents the T+nth m The environmental state at any given moment This indicates that the unmanned vehicle is in the T+n stage. m The action state at each moment, n m =n1,n2,...,n M n M Let T represent the M-th time interval between the (n-1)-th time interval and the n-th time interval, where n = 1, 2, ..., N, and N represents the N-th time interval after the current time T. This indicates that the unmanned vehicle is in the T+n stage. m The distance gain of the action strategy at each moment This indicates that the unmanned vehicle is in the T+n stage. m The repulsive benefit of the action strategy at each moment This indicates that the unmanned vehicle is in the T+n stage. m The time benefit of the action strategy at each moment:

[0064]

[0065]

[0066]

[0067] Among them, E O This indicates environmental information for autonomous vehicles. Represents the T+nth m The distance reward function value at each time step. Represents the T+nth m The repulsive reward function value at each moment. Represents the T+nth m Time reward function value at each moment:

[0068]

[0069]

[0070]

[0071] Where, n i =n1,n2,...,n M , It means that at the time of T+n m The distance between an unmanned vehicle and an obstacle at any given moment. Let q represent the position of the unmanned vehicle at time T+m, γ represent the discount rate, γ∈(0,1), and q obs Indicates the location of the obstacle, random(0,10) -3 ) represents from (0, 10 -3 Randomly select a value from the given information, where d is a constant. The repulsive force of the obstacle's repulsive field on the autonomous vehicle:

[0072]

[0073] Where ε and δ both represent gain coefficients, This represents the differential operation, and ρ0 represents the effective radius of the repulsive field of the obstacle.

[0074] Adjust the unmanned vehicle at time T+n m The action state at time T+n is adjusted by incrementing the count by 1, and the autonomous vehicle's action state at time T+n is calculated. m Benefits of the adjusted action strategy at each moment

[0075] Determine the unmanned vehicle at time T+n m Benefits of the adjusted action strategy at each moment Is it less than the number of unmanned vehicles at time T+n? m The benefits of action strategies at each moment

[0076] If so, determine whether the number of adjustments has reached the preset number of adjustments. If the preset number of adjustments has been reached, then adjust the unmanned vehicle on the (T+n)th adjustment. m Benefits of the adjusted action strategy at each moment As an unmanned vehicle at T+n m The benefit of the optimal action strategy at each moment If the preset number of adjustments is not reached, the unmanned vehicle will be moved to the next position on the T+nth day. m The benefit of the adjusted action strategy at time T+n is used as the autonomous vehicle's benefit at time T+n. m The benefits of action strategies at each moment And return to adjust the unmanned vehicle at time T+n m The action state of the unmanned vehicle at time T+n is calculated. m Benefits of the adjusted action strategy at each moment Steps;

[0077] Otherwise, determine whether the preset number of adjustments has been reached. If the preset number of adjustments has been reached, then adjust the unmanned vehicle on the (T+n)th adjustment. m The benefits of action strategies at each moment As an unmanned vehicle at T+n m The benefit of the optimal action strategy at each moment If the preset number of adjustments is not reached, return to the point where the unmanned vehicle is adjusted on the (T+n)th adjustment cycle. m The action state of the unmanned vehicle at time T+n is calculated. m Benefits of the adjusted action strategy at each moment Steps;

[0078] Based on the unmanned vehicle at the T+n m The benefit of the optimal action strategy at each moment The unmanned vehicle is obtained at time T+n m The optimal action strategy at time is

[0079] in, Represents the adjusted T+n-th digit. m The environmental state at any given moment This indicates that the adjusted unmanned vehicle is in the state T+n. m The action state at any given moment Represents the T+nth m The environment state corresponding to the optimal action strategy at each time point. Represents the T+nth m The action state corresponding to the optimal action strategy at each time point;

[0080] Through the formula:

[0081]

[0082] The calculation yields the unmanned vehicle at time T+n. m The combined force at that moment

[0083] in, It means that at the time of T+n m The gravitational force of the gravitational field of the nth target trajectory point at time n on the unmanned vehicle:

[0084]

[0085] in, It means that at the time of T+n m The distance q between the nth target trajectory point at time n and the unmanned vehicle. n This indicates the position of the nth target trajectory point;

[0086] The unmanned vehicle at time T+n m The resultant force experienced by the unmanned vehicle at time T+n is used as the force at time T+n. m The traction at that moment.

[0087] Secondly, embodiments of this application provide a path planning device for an unmanned vehicle, comprising:

[0088] The module constructs a historical trajectory sequence based on the trajectory points of the unmanned vehicle at time T; the Tth time is the current time.

[0089] The random selection module randomly selects candidate trajectory points for N times after the Tth time, based on the trajectory points of the unmanned vehicle at the current time.

[0090] The acquisition module obtains the direction selection probability, distance selection probability, and speed selection probability for each candidate trajectory point based on the historical trajectory sequence. The direction selection probability represents the probability that the candidate trajectory point is selected based on its offset angle. The distance selection probability represents the probability that the candidate trajectory point is selected based on its distance from other candidate trajectory points. The speed selection probability represents the probability that the candidate trajectory point is selected based on its speed at the candidate trajectory point.

[0091] The selection module calculates the comprehensive selection probability of each candidate trajectory point based on the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point, and selects the target trajectory points that the unmanned vehicle will pass through from the trajectory point at the current moment to the destination based on the calculated comprehensive selection probability.

[0092] The mobile module, during the process of the unmanned vehicle moving towards the target trajectory point, obtains the traction force of the unmanned vehicle based on the gravitational force of the target trajectory point on the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle, and controls the unmanned vehicle to move to the target trajectory point based on the traction force.

[0093] Thirdly, embodiments of this application provide a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the path planning method for the unmanned vehicle described above.

[0094] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned path planning method for unmanned vehicles.

[0095] The above-mentioned solution in this application has the following beneficial effects:

[0096] In the embodiments of this application, a historical trajectory sequence is constructed based on the trajectory points of the unmanned vehicle at T time points. Then, based on the trajectory point of the unmanned vehicle at the current time point, candidate trajectory points are randomly selected from N time points after the Tth time point. Then, based on the historical trajectory sequence, the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point are obtained. Then, based on the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point, the comprehensive selection probability of each candidate trajectory point is calculated. Based on the calculated comprehensive selection probability, the target trajectory point that the unmanned vehicle will pass through from the trajectory point at the current time point to the destination is selected. Finally, during the process of the unmanned vehicle moving towards the target trajectory point, the traction force of the unmanned vehicle is obtained based on the attraction of the target trajectory point to the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle. The unmanned vehicle is then controlled to move to the target trajectory point based on the traction force. Specifically, the system selects the target trajectory point that the unmanned vehicle will take from its current trajectory point to its destination based on the calculated comprehensive selection probability. This takes into account factors such as distance, offset angle, and speed, which improves the rationality of the target trajectory point. The system obtains the traction force of the unmanned vehicle based on the attraction of the target trajectory point to the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle. The system then controls the unmanned vehicle to move to the target trajectory point based on the traction force, allowing the unmanned vehicle to move according to the actual situation and effectively avoid obstacles during movement, thereby improving the rationality and safety of the movement trajectory.

[0097] Other beneficial effects of this application will be described in detail in the following detailed description section. Attached Figure Description

[0098] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0099] Figure 1 A flowchart of a path planning method for an unmanned vehicle provided in an embodiment of this application;

[0100] Figure 2 A schematic diagram of candidate trajectory points provided in an embodiment of this application;

[0101] Figure 3 Force analysis diagram of an unmanned vehicle provided in an embodiment of this application;

[0102] Figure 4 A force analysis diagram of an unmanned vehicle under defective conditions provided in an embodiment of this application;

[0103] Figure 5 A structural diagram of a path planning system for an unmanned vehicle provided in an embodiment of this application;

[0104] Figure 6 A schematic diagram of the first path planning for an unmanned vehicle provided in an embodiment of this application;

[0105] Figure 7 A schematic diagram of the second path planning for an unmanned vehicle provided in an embodiment of this application;

[0106] Figure 8 A schematic diagram of third path planning for an unmanned vehicle provided in an embodiment of this application;

[0107] Figure 9 A schematic diagram of the fourth path planning for an unmanned vehicle provided in an embodiment of this application;

[0108] Figure 10 A schematic diagram of the structure of a path planning device for an unmanned vehicle provided in an embodiment of this application;

[0109] Figure 11 This is a schematic diagram of the structure of a terminal device provided in an embodiment of this application. Detailed Implementation

[0110] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0111] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0112] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0113] To address the problem of unreasonable path planning in current autonomous vehicles, this application provides a path planning method for autonomous vehicles.

[0114] The path planning method for unmanned vehicles provided in this application is illustrated below.

[0115] like Figure 1 As shown, the path planning method for unmanned vehicles provided in this application includes the following steps:

[0116] Step 11: Construct a historical trajectory sequence based on the trajectory points of the unmanned vehicle at time T.

[0117] The Tth time point mentioned above is the current time point, and the T-1 times point before the Tth time point are all historical times.

[0118] In some embodiments of this application, the constructed historical trajectory sequence TS is:

[0119] TS = {LP1,...,LP} t ,...,LP T}

[0120] Where LP1 represents the trajectory point of the autonomous vehicle at the first moment, LP t Let LP represent the trajectory point of the unmanned vehicle at time t. T Let T represent the trajectory point of the unmanned vehicle at time T, where t = 1, 2, ..., T, and T represents the current time.

[0121] It is worth mentioning that the historical trajectory sequence can represent the historical movement trajectory of the unmanned vehicle, including information about the trajectory points of the unmanned vehicle at T time points, such as the location and time of the trajectory points.

[0122] Step 12: Based on the trajectory points of the unmanned vehicle at the current moment, randomly select candidate trajectory points for N moments after the Tth moment.

[0123] In some embodiments of this application, at each time point after the current time, multiple candidate trajectory points can be selected by random sampling.

[0124] For example, for a trajectory point at the current time T, multiple candidate trajectory points are selected at time T+1 by random sampling. For each candidate trajectory point at time T+1, multiple candidate trajectory points are selected at time T+2 by random sampling. Similarly, all candidate trajectory points at N times after time T are selected. By treating all candidate trajectory points as nodes and the connections between candidate trajectory points as edges, a directed graph with N layers can be constructed.

[0125] Step 13: Based on the historical trajectory sequence, obtain the direction selection probability, distance selection probability, and velocity selection probability for each candidate trajectory point.

[0126] Specifically, for each candidate trajectory point, the following steps can be performed: calculate the offset angle of the unmanned vehicle at the candidate trajectory point to obtain the direction selection probability of the candidate trajectory point; obtain the distance selection probability of the candidate trajectory point based on the historical trajectory sequence and the distance between the candidate trajectory point and other candidate trajectory points; obtain the speed selection probability of the candidate trajectory point based on the speed of the unmanned vehicle at the candidate trajectory point.

[0127] Among them, the direction selection probability is used to represent the probability that a candidate trajectory point is selected based on its offset angle; the distance selection probability is used to represent the probability that a candidate trajectory point is selected based on its distance from other candidate trajectory points; and the speed selection probability is used to represent the probability that a candidate trajectory point is selected based on its speed at the candidate trajectory point.

[0128] It should be noted that the aforementioned offset angle is the angle between the direction of the unmanned vehicle at the corresponding candidate trajectory point and the direction of the unmanned vehicle at the previous trajectory point.

[0129] It is worth mentioning that the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point are obtained by considering multiple factors of the candidate trajectory point and taking into account the historical trajectory sequence of the unmanned vehicle during the calculation. The results are based on reality and are reasonable.

[0130] Step 14: Calculate the comprehensive selection probability of each candidate trajectory point based on the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point, and select the target trajectory point that the unmanned vehicle will pass through from the trajectory point at the current moment to the destination based on the calculated comprehensive selection probability.

[0131] The above-mentioned comprehensive selection probability is used to represent the probability that the corresponding candidate trajectory point is selected.

[0132] In the embodiments of this application, for the same moment, the comprehensive selection probability of each candidate trajectory point is calculated, and the candidate trajectory point with the largest comprehensive selection probability is selected from all the comprehensive selection probabilities. The candidate trajectory point corresponding to this comprehensive selection probability is the target trajectory point at this moment.

[0133] It should be noted that the above destination is the last target trajectory point in the unmanned vehicle path planning, that is, the target trajectory point at time T+N.

[0134] For example, in some embodiments of this application, the above steps can be used to select multiple candidate trajectory points at each time after the current time T to obtain the target trajectory point at each time after the current time T. For a directed graph composed of all candidate trajectory points, all nodes in the Nth layer of the directed graph are end nodes. For each end node, the candidate trajectory points corresponding to the end node and all its predecessor nodes constitute a candidate trajectory sequence of N candidate trajectory points. The above steps are used to calculate the comprehensive selection probability of each candidate trajectory point in the candidate trajectory sequence, and these comprehensive selection probabilities are added together to obtain the selection probability of the candidate trajectory sequence. The selection probabilities of all candidate trajectory sequences are calculated, and the candidate trajectory sequence with the highest selection probability is taken as the target candidate trajectory sequence. All candidate trajectory points in the target candidate trajectory sequence are taken as target trajectory points.

[0135] It is worth mentioning that by comprehensively selecting target candidate trajectory points based on probability, it is possible to select target trajectory points that are both realistic and reasonable.

[0136] Step 15: During the process of the unmanned vehicle moving towards the target trajectory point, the traction force of the unmanned vehicle is obtained based on the gravitational force of the target trajectory point on the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle, and the unmanned vehicle is controlled to move to the target trajectory point based on the traction force.

[0137] It should be noted that the target trajectory point can be used as the center of the gravitational field, and the obstacle can be used as the center of the repulsive field to construct the gravitational field and the repulsive field respectively. The gravitational field and the repulsive field will generate attraction and repulsion on the unmanned vehicle respectively.

[0138] In some embodiments of this application, the optimal action strategy of the unmanned vehicle can be obtained by using a reward function based on the attraction of the target trajectory point to the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle. The traction force of the unmanned vehicle is obtained based on the optimal action strategy, and the unmanned vehicle is controlled to move to the target trajectory point based on the traction force.

[0139] The optimal movement strategy described above refers to the direction and speed at which the unmanned vehicle can reach the target trajectory point as quickly as possible while staying away from obstacles. These obstacles are entities such as other vehicles and roadblocks that obstruct the movement of the unmanned vehicle.

[0140] It is worth mentioning that the method of selecting the target trajectory point for the unmanned vehicle from the current trajectory point to the destination based on the calculated comprehensive selection probability takes into account factors such as distance, offset angle, and speed, which can improve the rationality of the target trajectory point. Based on the attraction of the target trajectory point to the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle, the traction force of the unmanned vehicle is obtained, and the unmanned vehicle is controlled to move to the target trajectory point according to the traction force. This allows the unmanned vehicle to move according to the actual situation, effectively avoid obstacles during operation, and improve the rationality and safety of the movement trajectory.

[0141] The specific steps of step 13 described above will be illustrated below with reference to specific embodiments.

[0142] In some embodiments of this application, the specific implementation process of step 13 above includes the following steps:

[0143] Step 13.1: Calculate the offset angle of the unmanned vehicle at the candidate trajectory point to obtain the direction selection probability of the candidate trajectory point.

[0144] Specifically, through the formula:

[0145]

[0146] Calculate the difference between the offset angle of the j-th candidate trajectory point at time T+n and the offset angle of the j-th candidate trajectory point at time (T+n)-1.

[0147] in, This represents the offset angle of the j-th candidate trajectory point at time T+n. Let represent the offset angle of the j-th candidate trajectory point at time (T+n)-1. If n = T+1, then... Let represent the offset angle of the trajectory point at time T, n = 1, 2, ..., N, where N represents the Nth time after the current time T, j = 1, 2, ..., Q, where Q represents the total number of candidate trajectory points at each time.

[0148] Through the formula:

[0149]

[0150] Calculate the direction selection probability of the j-th candidate trajectory point at time T+n.

[0151] Where τ is a preset value.

[0152] It is worth mentioning that autonomous vehicles have directionality and purpose when they actually move. By using the direction selection probability, candidate trajectory points with deviation angles that do not meet expectations can be eliminated, ensuring that the direction of movement of the autonomous vehicle is always consistent with the direction of the destination.

[0153] Step 13.2: Based on the historical trajectory sequence and the distance between candidate trajectory points and other candidate trajectory points, obtain the distance selection probability of candidate trajectory points.

[0154] Specifically, through the formula:

[0155]

[0156] Calculate the relative distance selection probability of the j-th candidate trajectory point at time T+n.

[0157] in, This represents the distance between the j-th candidate trajectory point at time T+n and the j-th candidate trajectory point at time (T+n)-1. This represents the j-th candidate trajectory point at time T+n. Let represent the j-th candidate trajectory point at time (T+n)-1, where n = 1, 2, ..., N, N represents the Nth time after the current time T, j = 1, 2, ..., Q, Q represents the total number of candidate trajectory points at the same time, and σ, e, and μ are all constants;

[0158] Through the formula:

[0159]

[0160] Calculate the probability of selecting the j-th candidate trajectory point by the movement distance at time T+n.

[0161] Where, di(LP) T-1 ,LP T ) represents the distance between the trajectory point at time T and the trajectory point at time T-1, LP T Let LP represent the trajectory point at time T. T-1 This represents the trajectory point at time T-1;

[0162] Through the formula:

[0163]

[0164] Calculate the distance selection probability of the j-th candidate trajectory point at time T+n.

[0165] It is worth mentioning that the distance selection probability takes into account the historical travel distance of the autonomous vehicle and the distance between the candidate trajectory point and the previous candidate trajectory point. If the distance between the candidate trajectory point and the previous candidate trajectory point is smaller than the distance between other candidate trajectory points and the previous candidate trajectory point, and the distance between the candidate trajectory point and the previous candidate trajectory point and the distance between other candidate trajectory points are numerically close to the historical travel distance of the autonomous vehicle, then the distance selection probability is large.

[0166] Step 13.3: Based on the speed of the unmanned vehicle at the candidate trajectory point, obtain the speed selection probability of the candidate trajectory point.

[0167] Specifically, through the formula:

[0168]

[0169] Calculate the velocity of the unmanned vehicle at the j-th candidate trajectory point at time T+n.

[0170] Among them, t T+n t(T) represents the value at the (T+n)th time point. +n)-1 This represents the value at the (T+n)-1th time point.

[0171] Through the formula:

[0172]

[0173] Calculate the velocity selection probability of the j-th candidate trajectory point at time T+n.

[0174] in, LP represents the average velocity of the historical trajectory sequence, LP1 represents the trajectory point at time 1, LP T This represents the trajectory point at time T.

[0175] It is worth mentioning that the calculation of the speed selection probability takes into account the speed of the unmanned vehicle at the candidate trajectory point and the average speed of the unmanned vehicle's historical trajectory sequence. Based on the principle that the speed of the unmanned vehicle is relatively stable within a certain range, if the speed of the unmanned vehicle at the candidate trajectory point is close to the average speed of the unmanned vehicle's historical trajectory sequence, then the speed selection probability of the candidate trajectory point is large.

[0176] The above steps will be explained by way of example with a specific example below.

[0177] like Figure 2 As shown, for the current trajectory point LP at the current time t (i.e., the current time T mentioned above) tBy randomly selecting points, six candidate trajectory points are obtained at time t+1. These six candidate trajectory points are compared with the current trajectory point LP. t The relationships between them are defined as selection structures. Step (a) is executed, considering the relative distance selection probabilities of the six candidate trajectory sequences, and then selecting the current trajectory point LP. t The five candidate trajectory points that are relatively close to each other and the current trajectory point LP t The relationships between them are defined as an enhanced structure, preserving the relationship with the current trajectory point LP. t The relationship between candidate trajectory points is used to strengthen the structure, and points that are related to the current trajectory point LP are removed. t The relationship between the candidate trajectory points of the selection structure Then, transform all reinforcement structures into selection structures and execute step (b). Consider the direction selection probabilities of the remaining 5 candidate trajectory points, and match the candidate trajectory points with non-zero direction selection probabilities with the current trajectory point LP. t The relationships between them are defined as an enhanced structure, removing the relationship with the current trajectory point LP. t The relationship between the candidate trajectory points of the selection structure and Then, transform all reinforcement structures into selection structures and execute step (c). Consider the movement distance selection probability of the remaining 3 candidate trajectory points, and match the candidate trajectory point with the relatively high movement distance selection probability with the current trajectory point LP. t The relationships between them are defined as an enhanced structure, removing the relationship with the current trajectory point LP. t The relationship between the candidate trajectory points of the selection structure Two remaining candidate trajectory points and In the diagram, arrows indicate the direction of movement of the unmanned vehicle, dots represent historical trajectory points of the unmanned vehicle, diamond dots represent current trajectory points, square dots represent candidate trajectory points, thick arrows indicate the purpose of expansion (i.e., the direction of the steps), short dashed lines represent selection structures, and long dashed lines represent reinforcement structures.

[0178] The specific steps of step 14 described above will be illustrated below with reference to specific embodiments.

[0179] In some embodiments of this application, the specific implementation process of step 14 above includes the following steps:

[0180] Step 14.1: Calculate the comprehensive selection probability of each candidate trajectory point based on the direction selection probability, distance selection probability, and velocity selection probability of each candidate trajectory point.

[0181] Specifically, through the formula:

[0182]

[0183] Calculate the overall selection probability of the j-th candidate trajectory point at time T+n.

[0184] in, This represents the j-th candidate trajectory point at time (T+n)-1. This represents the j-th candidate trajectory point at time T+n. Let represent the direction selection probability of the j-th candidate trajectory point at time T+n. Let represent the distance selection probability of the j-th candidate trajectory point at time T+n. This represents the velocity selection probability of the j-th candidate trajectory point at time T+n.

[0185] Step 14.2: Calculate the overall selection probability of all candidate trajectory points.

[0186] Specifically, the combined selection probability f of all candidate trajectory points at time T+n. T+n for:

[0187]

[0188] in, This represents the first candidate trajectory point at time (T+n)-1. This represents the first candidate trajectory point at time T+n. This represents the overall selection probability of the first candidate trajectory point at time T+n. This represents the overall selection probability of the Qth candidate trajectory point at time T+n;

[0189] Step 14.3: Select the target trajectory points that the unmanned vehicle will pass through from the current trajectory point to the destination based on the calculated comprehensive selection probability.

[0190] The combined selection probability f of all candidate trajectory points at time T+n T+n The candidate trajectory point with the highest comprehensive selection probability is taken as the target trajectory point at time T+n.

[0191] It should be noted that the above steps are performed for N time points after the current time T, resulting in a total of N target trajectory points. These N target trajectory points are all the target trajectory points that the unmanned vehicle passes through from the trajectory point at the current time to the destination.

[0192] It is worth mentioning that by comprehensively selecting target candidate trajectory points based on probability, it is possible to select target trajectory points that are both realistic and reasonable.

[0193] The specific steps of step 15 described above will be illustrated below with reference to specific embodiments.

[0194] In some embodiments of this application, the specific implementation process of step 15 above includes the following steps:

[0195] Step 15.1: Based on the gravitational force of the target trajectory point on the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle on the unmanned vehicle, the optimal action strategy of the unmanned vehicle can be obtained using the reward function.

[0196] Specifically, through the formula:

[0197]

[0198] Calculate the unmanned vehicle on the T+n m The benefits of action strategies at each moment Action strategy is a combination of the environment state and the action state selected in that environment state;

[0199] in, This indicates that the unmanned vehicle is in the T+n stage. m Action strategy at each moment, s T+m Represents the T+nth m The environmental state at any given moment This indicates that the unmanned vehicle is in the T+n stage. m The action state at each moment, n m =n1,n2,...,n M n M Let T represent the M-th time interval between the (n-1)-th time interval and the n-th time interval, where n = 1, 2, ..., N, and N represents the N-th time interval after the current time T. This indicates that the unmanned vehicle is in the T+n stage. m The distance gain of the action strategy at each moment This indicates that the unmanned vehicle is in the T+n stage. m The repulsive benefit of the action strategy at each moment This indicates that the unmanned vehicle is in the T+n stage. m The time benefit of the action strategy at each moment:

[0200]

[0201]

[0202]

[0203] Among them, E O This indicates environmental information for autonomous vehicles. Represents the T+nth m The distance reward function value at each time step. Represents the T+nth mThe repulsive reward function value at each moment. Represents the T+nth m Time reward function value at each moment:

[0204]

[0205]

[0206]

[0207] Where, n i =n1,n2,...,n M , It means that at the time of T+n m The distance between an unmanned vehicle and an obstacle at any given moment. Let q represent the position of the unmanned vehicle at time T+m, γ represent the discount rate, γ∈(0,1), and q obs Indicates the location of the obstacle, random(0,10) -3 ) represents from (0, 10 -3 Randomly select a value from the given information, where d is a constant. The repulsive force of the obstacle's repulsive field on the autonomous vehicle:

[0208]

[0209] Where ε and δ both represent gain coefficients, This represents the differential operation, and ρ0 represents the effective radius of the repulsive field of the obstacle.

[0210] Adjust the unmanned vehicle at time T+n m The action state at time T+n is adjusted by incrementing the count by 1, and the autonomous vehicle's action state at time T+n is calculated. m Benefits of the adjusted action strategy at each moment

[0211] Determine the unmanned vehicle at time T+n m Benefits of changing action strategies at any given moment Is it less than the number of unmanned vehicles at time T+n? m The benefits of action strategies at each moment

[0212] If so, determine whether the number of adjustments has reached the preset number of adjustments. If the preset number of adjustments has been reached, then adjust the unmanned vehicle on the (T+n)th adjustment. m Benefits of the adjusted action strategy at each moment As an unmanned vehicle at T+n m The benefit of the optimal action strategy at each moment If the preset number of adjustments is not reached, the unmanned vehicle will be moved to the next position on the T+nth day.m The benefit of the adjusted action strategy at time T+n is used as the autonomous vehicle's benefit at time T+n. m The benefits of action strategies at each moment And return to adjust the unmanned vehicle at time T+n m The action state of the unmanned vehicle at time T+n is calculated. m Benefits of the adjusted action strategy at each moment Steps;

[0213] Otherwise, determine whether the preset number of adjustments has been reached. If the preset number of adjustments has been reached, then adjust the unmanned vehicle on the (T+n)th adjustment. m The benefits of action strategies at each moment As an unmanned vehicle at T+n m The benefit of the optimal action strategy at each moment If the preset number of adjustments is not reached, return to the point where the unmanned vehicle is adjusted on the (T+n)th adjustment cycle. m The action state of the unmanned vehicle at time T+n is calculated. m Benefits of the adjusted action strategy at each moment Steps;

[0214] Based on the unmanned vehicle at the T+n m The benefit of the optimal action strategy at each moment The unmanned vehicle is obtained at time T+n m The optimal action strategy at time is

[0215] in, Represents the adjusted T+n-th digit. m The environmental state at any given moment This indicates that the adjusted unmanned vehicle is in the state T+n. m The action state at any given moment Represents the T+nth m The environment state corresponding to the optimal action strategy at each time point. Represents the T+nth m The action state corresponding to the optimal action strategy at time point .

[0216] It should be noted that the aforementioned environmental states represent the physical information such as the location of the unmanned vehicle, its corresponding destination location, obstacle locations, and the length and width of the road where the unmanned vehicle is located. The action states represent the unmanned vehicle's direction and speed of movement, while the environmental information represents the traffic environment of the area where the unmanned vehicle is located, such as the positions of all vehicles and road conditions. The unmanned vehicle can adjust its action states by adjusting its speed and direction of movement. When the unmanned vehicle adjusts its action states, the predicted position is based on the changes in the action states, and the corresponding environmental states will also change. Each time the unmanned vehicle adjusts its action states, the adjusted action states are different from all previous adjusted action states. For the time interval between the time of the nth target trajectory point and the time of the (n-1)th target trajectory point, this time interval is divided into multiple time intervals. For example, if the time of the first target trajectory point is second 1, and the time of the second target trajectory point is second 2, the time interval from second 1 to second 2 can be divided into three time intervals: second 1, second 1.5, and second 2.

[0217] For example, for the T+nth m At any given moment, the preset number of adjustments is 5. The benefit of the adjusted action strategy after the 4th adjustment is... like Greater than or equal to The benefit of not updating the action strategy. like Less than The benefit of updating the action strategy will be As The autonomous vehicle adjusts its action strategy again, and the adjustment count is incremented by 1. Since the adjustment count has been reached, a decision is made... Is it less than If so, The corresponding action strategy is the optimal action strategy. Otherwise, The corresponding action strategy is the optimal action strategy.

[0218] It is worth mentioning that by constructing a repulsive field for the obstacle and an attractive field for the target trajectory point, the distances between the unmanned vehicle and the obstacle and the target trajectory point can be reasonably calculated, achieving the result of avoiding the obstacle and approaching the target trajectory point. The purpose of adjusting the action state as described above is to minimize the benefit of the optimal action strategy. A smaller benefit indicates a smaller repulsive force from the obstacle on the unmanned vehicle and a larger distance between the unmanned vehicle and the obstacle.

[0219] Step 15.2: Obtain the traction force of the unmanned vehicle based on the optimal action strategy.

[0220] Through the formula:

[0221]

[0222] The calculation yields the unmanned vehicle at time T+n. m The combined force at that moment

[0223] in, It means that at the time of T+n m The gravitational force of the gravitational field of the nth target trajectory point at time n on the unmanned vehicle:

[0224]

[0225] in, It means that at the time of T+n m The distance q between the nth target trajectory point at time n and the unmanned vehicle. n This indicates the position of the nth target trajectory point;

[0226] The unmanned vehicle at time T+n m The resultant force experienced by the unmanned vehicle at time T+n is used as the force at time T+n. m The traction at that moment.

[0227] Step 15.3: Control the unmanned vehicle to move to the target trajectory point based on the traction force.

[0228] Specifically, traction determines the direction and speed of movement of the unmanned vehicle, which then moves toward the target trajectory point based on the direction and speed of movement.

[0229] It is worth mentioning that by introducing various reward functions such as time, repulsion, and distance to adjust and optimize the action strategy of unmanned vehicles, the unmanned vehicles can reach their destination while avoiding obstacles, thereby improving the safety and rationality of the unmanned vehicle's path.

[0230] The above steps will be explained by way of example with a specific example below.

[0231] The force analysis of an unmanned vehicle provided in one embodiment of this application is as follows: Figure 3 As shown, the unmanned vehicle in the figure is subjected to repulsive forces from obstacle 1 and obstacle 2, and simultaneously to an attractive force from the destination (i.e., the target trajectory point mentioned above). The repulsive and attractive forces act on the unmanned vehicle at the same time, causing it to experience a resultant force to the right. The star-shaped mark in the figure represents the destination.

[0232] Based on the above force analysis, the two defect scenarios are as follows: Figure 4As shown, in the first defective case, the repulsive forces of obstacle 1 and obstacle 2 on the unmanned vehicle are exactly equal in magnitude and opposite in direction to the gravitational force of the endpoint (i.e., the target trajectory point mentioned above) on the unmanned vehicle. In this case, the net force on the unmanned vehicle is 0, and it falls into a local optimum (i.e., cannot move). In the second defective case, the unmanned vehicle is subjected to the repulsive force of the obstacle and the gravitational force of the endpoint. The net force generated by the two is not in the same direction as the endpoint, and the unmanned vehicle cannot reach the endpoint. The polygon in the figure represents the unmanned vehicle.

[0233] It is easy to see that directly calculating the resultant force on an unmanned vehicle based on repulsive and attractive forces results in a resultant force on the vehicle that is only related to its position, which has limitations.

[0234] One embodiment of this application provides a path planning system for unmanned vehicles, such as... Figure 5 As shown in the figure, the target trajectory points are obtained through the direction selection, distance selection, and speed selection functions in the target trajectory sequence generation module. The target trajectory points form a target trajectory sequence. Based on the target trajectory sequence, the path refinement module obtains the gravitational and repulsive forces experienced by the unmanned vehicle when moving towards the target trajectory points in the target trajectory sequence. The gravitational and repulsive forces are then processed through reinforcement learning to obtain the final path of the unmanned vehicle.

[0235] The path of the unmanned vehicle obtained through the path planning system described above is as follows: Figure 6 As shown.

[0236] In the diagram, the target vehicle (i.e. the unmanned vehicle mentioned above) starts from the starting point and, when it approaches the first static obstacle, adjusts its motion state and obtains the optimal motion strategy to avoid the obstacle. Similarly, when the target vehicle approaches other obstacles, it adjusts its motion state and obtains the optimal motion strategy to avoid the obstacles. Figure 6 The square represents the target vehicle. The solid line starting from the target vehicle represents the target vehicle's trajectory. The triangle is the first target trajectory point, the star is the second target trajectory point, the rhombus is the third target trajectory point, the pentagon is the fourth target trajectory point, the circle is a static obstacle, the dotted line is the center line of the road, and the two parallel solid lines are the boundary lines of the road. Each grid is 2m wide and 50m long.

[0237] The path of another autonomous vehicle obtained through the path planning system of the aforementioned autonomous vehicle is as follows: Figure 7 As shown.

[0238] In the diagram, the target vehicle (i.e. the unmanned vehicle mentioned above) starts from the starting point. There are dynamic obstacles 1 and 2 traveling parallel to it, as well as static obstacles nearby. When the target vehicle approaches the static obstacle, it adjusts its action state and obtains the optimal action strategy to avoid the dynamic obstacle while avoiding the static obstacle. Figure 7The square represents the target vehicle. The solid line starting from the target vehicle represents its trajectory. The triangle represents the first target trajectory point, the star represents the second target trajectory point, and the pentagon represents the third target trajectory point. The circle represents a static obstacle. The circle with a cross represents dynamic obstacle 1. The dashed line starting from the circle of dynamic obstacle 1 represents its trajectory. The triangle with a cross represents the first trajectory point of dynamic obstacle 1, the star with a cross represents the second trajectory point, and the rhombus with a cross represents the third trajectory point. The circle with a cross represents dynamic obstacle 2. The solid line starting from dynamic obstacle 2 represents its trajectory. The triangle with a cross represents the first trajectory point of dynamic obstacle 2. The segmented dashed lines represent the center line of the road, and the two parallel solid lines represent the boundary lines of the road. Each grid is 2m wide and 50m long.

[0239] The path of another autonomous vehicle obtained through the path planning system of the aforementioned autonomous vehicle is as follows: Figure 8 As shown.

[0240] In the diagram, the target vehicle (i.e., the aforementioned unmanned vehicle) starts from the starting point. There are dynamic obstacles and static obstacles traveling parallel to it in its vicinity. The road traveled by the target vehicle is curved. The target vehicle avoids the obstacles by adjusting its action state and obtaining the optimal action strategy. When the road is not curved, it travels parallel to the dynamic obstacles at a certain distance. When the road is curved, by adjusting its action state and obtaining the optimal action strategy, the target vehicle increases the distance between itself and the dynamic obstacles. After that, the target vehicle's travel path remains parallel to the boundary of the curved road. Figure 8 The square represents the target vehicle. The solid line starting from the target vehicle represents its trajectory. The triangle is the first target trajectory point, the star is the second, the rhombus is the third, and the pentagon is the fourth. The circle represents a static obstacle, and the circle with a cross represents a dynamic obstacle. The dashed line starting from the dynamic obstacle represents its trajectory. The triangle with a cross is the first trajectory point of the dynamic obstacle, the star with a cross is the second, the rhombus with a cross is the third, the pentagon with a cross is the fourth, and the arrow with a cross is the fifth. The segmented dashed lines are the center lines of the road, and the two parallel solid lines are the boundary lines of the road. Each grid is 4m wide and 50m long.

[0241] The path of another autonomous vehicle obtained through the path planning system of the aforementioned autonomous vehicle is as follows: Figure 9 As shown.

[0242] In the diagram, the target vehicle (i.e., the aforementioned unmanned vehicle) starts from the starting point. Near it, there is a dynamic obstacle 1 traveling in the same direction as it, a dynamic obstacle 2 in front of it, and a static obstacle. The target vehicle adjusts its action state and obtains the optimal action strategy to avoid the dynamic obstacle 1 while avoiding the static obstacle. When it gets close to the dynamic obstacle 1 and the dynamic obstacle 2, it adjusts its action state and obtains the optimal action strategy to avoid the dynamic obstacle 1 and the dynamic obstacle 2. Finally, it travels in front of the dynamic obstacle 1 at a speed greater than that of the dynamic obstacle 1. Figure 9 The square represents the target vehicle. The solid line starting from the target vehicle represents its trajectory. The triangle represents the first target trajectory point, the star represents the second, the rhombus represents the third, the pentagon represents the fourth, and the arrow represents the fifth. The circle represents a static obstacle. The circle with an X represents dynamic obstacle 1. The solid line starting from dynamic obstacle 1 represents its trajectory. The triangle with an X represents the first trajectory point of dynamic obstacle 1, the star with an X represents the second, the rhombus with an X represents the third, and so on. The pentagon is the fourth trajectory point, the arrow with a cross is the fifth trajectory point, the circle with a cross represents dynamic obstacle 2, the dashed line starting from dynamic obstacle 2 is the movement trajectory of dynamic obstacle 2, the triangle with a cross is the first trajectory point of dynamic obstacle 2, the star with a cross is the second trajectory point, the rhombus with a cross is the third trajectory point, the pentagon with a cross is the fourth trajectory point, the arrow with a cross is the fifth trajectory point, the segmented dashed lines are the center line of the road, the two parallel solid lines are the boundary lines of the road, the width of each grid is 4m, and the length of each grid is 50m.

[0243] in, Figure 6 , Figure 7 , Figure 8 , Figure 9 In this context, the length of the road is used as the longitudinal axis and the width as the lateral axis to represent the longitudinal and lateral positions of each point on the road.

[0244] It can be seen that the path planning method for unmanned vehicles provided in this application can improve the rationality of path planning, and at the same time, it can effectively avoid obstacles and drive towards the destination during the driving process, thereby improving the safety of unmanned vehicle driving.

[0245] The path planning device for unmanned vehicles provided in this application will be described exemplarily below.

[0246] like Figure 10 As shown, this application embodiment provides a path planning device for an unmanned vehicle. The path planning device 1000 for the unmanned vehicle includes:

[0247] Module 1001 constructs a historical trajectory sequence based on the trajectory points of the unmanned vehicle at time T; the Tth time is the current time.

[0248] The random selection module 1002 randomly selects candidate trajectory points for N times after the Tth time based on the trajectory points of the unmanned vehicle at the current time.

[0249] The acquisition module 1003 acquires the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point based on the historical trajectory sequence. The direction selection probability represents the probability that the candidate trajectory point is selected based on its offset angle. The distance selection probability represents the probability that the candidate trajectory point is selected based on its distance from other candidate trajectory points. The speed selection probability represents the probability that the candidate trajectory point is selected based on its speed at the candidate trajectory point.

[0250] The selection module 1004 calculates the comprehensive selection probability of each candidate trajectory point based on the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point, and selects the target trajectory point that the unmanned vehicle will pass through from the trajectory point at the current moment to the destination based on the calculated comprehensive selection probability.

[0251] The mobile module 1005, during the process of the unmanned vehicle moving towards the target trajectory point, obtains the traction force of the unmanned vehicle based on the gravitational force of the target trajectory point on the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle, and controls the unmanned vehicle to move to the target trajectory point based on the traction force.

[0252] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application. For details on their specific functions and technical effects, please refer to the method embodiments section, and they will not be repeated here.

[0253] like Figure 11 As shown, an embodiment of this application provides a terminal device, wherein the terminal device D10 of this embodiment includes: at least one processor D100 ( Figure 11 The diagram shows only one processor, a memory D101, and a computer program D102 stored in the memory D101 and executable on the at least one processor D100, wherein the processor D100 executes the computer program D102 to implement the steps in any of the above method embodiments.

[0254] Specifically, when the processor D100 executes the computer program D102, it constructs a historical trajectory sequence based on the trajectory points of the unmanned vehicle at T time points. Then, based on the trajectory point of the unmanned vehicle at the current time point, it randomly selects candidate trajectory points at N time points after the Tth time point. Then, based on the historical trajectory sequence, it obtains the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point. Then, based on the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point, it calculates the comprehensive selection probability of each candidate trajectory point. Based on the calculated comprehensive selection probability, it selects the target trajectory point that the unmanned vehicle will pass through from the trajectory point at the current time point to the destination. Finally, during the process of the unmanned vehicle moving towards the target trajectory point, it obtains the traction force of the unmanned vehicle based on the attraction of the target trajectory point to the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle, and controls the unmanned vehicle to move to the target trajectory point based on the traction force.

[0255] The processor D100 can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0256] In some embodiments, the memory D101 may be an internal storage unit of the terminal device D10, such as a hard disk or memory of the terminal device D10. In other embodiments, the memory D101 may be an external storage device of the terminal device D10, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the terminal device D10. Furthermore, the memory D101 may include both internal and external storage units of the terminal device D10. The memory D101 is used to store the operating system, applications, bootloader, data, and other programs, such as the program code of the computer program. The memory D101 can also be used to temporarily store data that has been output or will be output.

[0257] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps described in the various method embodiments above.

[0258] This application provides a computer program product that, when run on a terminal device, enables the terminal device to implement the steps described in the various method embodiments above.

[0259] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of this application can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to the path planning method apparatus / terminal device of an unmanned vehicle, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.

[0260] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0261] The above description is the preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principles described in this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A path planning method for unmanned vehicles, characterized in that, include: According to unmanned vehicles The trajectory points at each moment construct a historical trajectory sequence; The current moment is represented by the current moment. Based on the trajectory points of the unmanned vehicle at the current moment, the first... After that moment Candidate trajectory points at each moment; Based on the historical trajectory sequence, the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point are obtained; the direction selection probability represents the probability that the candidate trajectory point is selected based on its offset angle; the distance selection probability represents the probability that the candidate trajectory point is selected based on its distance from other candidate trajectory points; and the speed selection probability represents the probability that the candidate trajectory point is selected based on the speed of the unmanned vehicle at the candidate trajectory point. Based on the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point, the comprehensive selection probability of each candidate trajectory point is calculated, and the target trajectory point that the unmanned vehicle will pass through from the trajectory point at the current moment to the destination is selected based on the calculated comprehensive selection probability. During the process of the unmanned vehicle moving toward the target trajectory point, the traction force of the unmanned vehicle is obtained based on the gravitational force of the target trajectory point on the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle on the unmanned vehicle, and the unmanned vehicle is controlled to move to the target trajectory point based on the traction force. Through the formula: The calculation yielded the unmanned vehicle at the [number]th [year]. The combined force at that moment ; Indicates that the unmanned vehicle is in the first The benefit of the optimal action strategy at each moment. Indicates the first The environment state corresponding to the optimal action strategy at each time point. Indicates the first The action state corresponding to the optimal action strategy at time t. This represents the repulsive force exerted by the obstacle's repulsive field on the unmanned vehicle. Indicates the first At the [time]th moment The gravitational force of the gravitational field of each target trajectory point on the unmanned vehicle; The unmanned vehicle in the first The resultant force experienced by the unmanned vehicle at time n is the force at time n. The traction at that moment.

2. The path planning method according to claim 1, characterized in that, The historical trajectory sequence for: in, This represents the trajectory point of the unmanned vehicle at the first moment. Indicates that the unmanned vehicle is in the first The trajectory point at each moment, Indicates that the unmanned vehicle is in the first The trajectory point at each moment, , Indicates the current moment.

3. The path planning method according to claim 1, characterized in that, The step of obtaining the direction selection probability, distance selection probability, and velocity selection probability of each candidate trajectory point based on the historical trajectory sequence includes: For each of the candidate trajectory points, perform the following steps: The offset angle of the unmanned vehicle at the candidate trajectory point is calculated to obtain the direction selection probability of the candidate trajectory point; The distance selection probability of the candidate trajectory point is obtained based on the historical trajectory sequence and the distance between the candidate trajectory point and other candidate trajectory points; The speed selection probability of the candidate trajectory point is obtained based on the speed of the unmanned vehicle at the candidate trajectory point.

4. The path planning method according to claim 3, characterized in that, The calculation of the offset angle of the unmanned vehicle at the candidate trajectory point to obtain the direction selection probability of the candidate trajectory point includes: Through the formula: Calculate the first The first moment The offset angle of the candidate trajectory point and the... The first moment The difference between the offset angles of the candidate trajectory points ; in, Indicates the first The moment of the first The offset angle of each candidate trajectory point Indicates the first The moment of the first The offset angle of each candidate trajectory point, if ,but Indicates the first The offset angle of the trajectory point at each moment. , Indicates the current time After that, the At that moment, , This represents the total number of candidate trajectory points at each time step; Through the formula: Calculate the first The moment of the first The probability of direction selection for each candidate trajectory point ; in, This is the default value; The step of obtaining the distance selection probability of the candidate trajectory point based on the historical trajectory sequence and the distance between the candidate trajectory point and other candidate trajectory points includes: Through the formula: Calculate the first The moment of the first The relative distance selection probability of candidate trajectory points ; in, Indicates the first The moment of the first The candidate trajectory point and the first The first moment The distance between candidate trajectory points Indicates the first The moment of the first Candidate trajectory points, Indicates the first The moment of the first Candidate trajectory points, , Indicates the current time After that, the At that moment, , This represents the total number of candidate trajectory points at the same time. , and Both represent constants; Through the formula: Calculate the first The moment of the first Probability of selecting the moving distance of each candidate trajectory point ; in, Indicates the first The trajectory point at time n and at time n The distance between trajectory points at each moment. Indicates the first The trajectory point at each moment, Indicates the first The trajectory points at each moment; Through the formula: Calculate the first The moment of the first The probability of selecting a candidate trajectory point based on its distance ; The step of obtaining the speed selection probability of the candidate trajectory point based on the speed of the unmanned vehicle at the candidate trajectory point includes: Through the formula: Calculate the autonomous vehicle in the first The moment of the first The velocity of each candidate trajectory point ; in, Indicates the first The value at each time point. Indicates the first The value at each time point; Through the formula: Calculate the first The moment of the first Velocity selection probability of candidate trajectory points ; in, This represents the average velocity of the historical trajectory sequence. This represents the trajectory point at time 1. Indicates the first The trajectory points at each moment.

5. The path planning method according to claim 1, characterized in that, The step of calculating the comprehensive selection probability of each candidate trajectory point based on the direction selection probability, distance selection probability, and velocity selection probability of each candidate trajectory point includes: Through the formula: Calculate the first The moment of the first The combined selection probability of each candidate trajectory point ; in, Indicates the first The moment of the first Candidate trajectory points, Indicates the first The moment of the first Candidate trajectory points, Indicates the first The moment of the first The probability of direction selection for each candidate trajectory point. Indicates the first The moment of the first The distance selection probability of each candidate trajectory point Indicates the first The moment of the first The velocity selection probability of each candidate trajectory point.

6. The path planning method according to claim 5, characterized in that, In the The combined selection probability of all candidate trajectory points at time 1 for: in, Indicates the first The first candidate trajectory point at time 1. Indicates the first The first candidate trajectory point at time 1. Indicates the first The overall selection probability of the first candidate trajectory point at each time step. Indicates the first The moment of the first The combined selection probability of each candidate trajectory point; The step of selecting the target trajectory points that the unmanned vehicle will pass through from its current trajectory point to its destination based on the calculated comprehensive selection probability includes: In the The combined selection probability of all candidate trajectory points at time 1 The candidate trajectory point corresponding to the highest comprehensive selection probability in the median value is used as the first... The target trajectory point at any given time.

7. The path planning method according to claim 1, characterized in that, During the process of the unmanned vehicle moving towards the target trajectory point, the traction force of the unmanned vehicle is obtained based on the gravitational force of the target trajectory point on the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle, including: Through the formula: Calculate the unmanned vehicle in the first... The benefits of action strategies at each moment The action strategy is a combination of the environmental state and the action state selected under that environmental state. in, Indicates that the unmanned vehicle is in the first Action strategy at each moment Indicates the first The environmental state at any given moment Indicates that the unmanned vehicle is in the first The action state at any given moment , Indicates the first The moment and the first Between the nth and nth moments At that moment, , Indicates the current time After that, the At that moment, Indicates that the unmanned vehicle is in the first The distance gain of the action strategy at each moment Indicates that the unmanned vehicle is in the first The repulsive benefit of the action strategy at each moment Indicates that the unmanned vehicle is in the first The time benefit of the action strategy at each moment: in, This indicates the environmental information of the unmanned vehicle. Indicates the first The distance reward function value at each time step. Indicates the first The repulsive reward function value at each moment. Indicates the first Time reward function value at each moment: in, , Indicates the first The distance between the unmanned vehicle and the obstacle at a given moment. Indicates the first The location of the unmanned vehicle at that moment. Indicates the discount rate. , Indicates the location of the obstacle. Indicates from Randomly select a value from the list. It is a constant; in, and Both represent gain coefficients. express Differential operations, The effective radius of the repulsive field of the obstacle is indicated; Adjust the unmanned vehicle in the first The action state at time is adjusted, the number of adjustments is incremented by 1, and the autonomous vehicle's action state at time is calculated. Benefits of the adjusted action strategy at each moment ; Determine the unmanned vehicle in the first... Benefits of the adjusted action strategy at each moment Is it smaller than the unmanned vehicle in the first... The benefits of action strategies at each moment ; If so, then determine whether the number of adjustments has reached a preset number of adjustments. If the preset number of adjustments has been reached, then the unmanned vehicle will be adjusted to the next preset number of adjustments. Benefits of the adjusted action strategy at each moment As the unmanned vehicle in The benefit of the optimal action strategy at each moment If the preset number of adjustments is not reached, then the unmanned vehicle will be moved to the next position. The benefit of the adjusted action strategy at time t is taken as the benefit of the unmanned vehicle at time t. The benefits of action strategies at each moment And return to adjust the unmanned vehicle in the first The action state of the unmanned vehicle at time n is calculated, and the action state of the unmanned vehicle at time n is calculated. Benefits of the adjusted action strategy at each moment Steps; Otherwise, determine whether the preset number of adjustments has been reached. If the preset number of adjustments has been reached, then the unmanned vehicle will be moved to the next position. The benefits of action strategies at each moment As the unmanned vehicle in The benefit of the optimal action strategy at each moment If the preset number of adjustments is not reached, return to the adjustment stage for the unmanned vehicle in the [number]th [stage]. The action state of the unmanned vehicle at time n is calculated, and the action state of the unmanned vehicle at time n is calculated. Benefits of the adjusted action strategy at each moment Steps; Based on the unmanned vehicle in the The benefit of the optimal action strategy at each moment The unmanned vehicle was obtained in the first... The optimal action strategy at time is ; in, Indicates the adjusted number The environmental state at any given moment The adjusted unmanned vehicle is in the first... The action state at any given moment; in, Indicates the first The first moment mentioned The distance between each target trajectory point and the unmanned vehicle Indicates the first The location of each target trajectory point.

8. A path planning device for an unmanned vehicle, characterized in that, include: Modules are built based on the autonomous vehicle's... The trajectory points at each moment construct a historical trajectory sequence; The current moment is represented by the current moment. The random selection module, based on the trajectory points of the unmanned vehicle at the current moment, randomly selects the first... After that moment Candidate trajectory points at each moment; The acquisition module acquires the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point based on the historical trajectory sequence. The direction selection probability represents the probability that the candidate trajectory point is selected based on its offset angle. The distance selection probability represents the probability that the candidate trajectory point is selected based on its distance from other candidate trajectory points. The speed selection probability represents the probability that the candidate trajectory point is selected based on the speed of the unmanned vehicle at the candidate trajectory point. The selection module calculates the comprehensive selection probability of each candidate trajectory point based on the direction selection probability, distance selection probability, and speed selection probability of each candidate trajectory point, and selects the target trajectory point that the unmanned vehicle will pass through from the trajectory point at the current moment to the destination based on the calculated comprehensive selection probability. The mobile module, during the process of the unmanned vehicle moving towards the target trajectory point, obtains the traction force of the unmanned vehicle based on the gravitational force of the target trajectory point on the unmanned vehicle and the repulsive force of the obstacles around the unmanned vehicle on the unmanned vehicle, and controls the unmanned vehicle to move to the target trajectory point based on the traction force. Through the formula: The calculation yielded the unmanned vehicle at the [number]th [year]. The combined force at that moment ; Indicates that the unmanned vehicle is in the first The benefit of the optimal action strategy at each moment. Indicates the first The environment state corresponding to the optimal action strategy at each time point. Indicates the first The action state corresponding to the optimal action strategy at time t. This represents the repulsive force exerted by the obstacle's repulsive field on the unmanned vehicle. Indicates the first At the [time]th moment The gravitational force of the gravitational field of each target trajectory point on the unmanned vehicle; The unmanned vehicle in the first The resultant force experienced by the unmanned vehicle at time n is the force at time n. The traction at that moment.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the path planning method for the unmanned vehicle as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the path planning method for the unmanned vehicle as described in any one of claims 1 to 7.

Citation Information

Patent Citations

  • Target obstacle vehicle trajectory prediction method based on Bayesian network

    CN111857134A

  • Local path planning method based on artificial potential field method and reinforcement learning

    CN114859905A