An autonomous driving speed planning method and system

Abstract

The application discloses an automatic driving speed planning method and system, and belongs to the technical field of automatic driving.The application adopts a speed filtering method taking acceleration as a threshold to linearize obstacle constraints, and adopts a new linear programming method to linearize jerk constraints, so that the real-time performance of an algorithm is greatly improved; meanwhile, in the case that hard constraints cannot meet safe obstacle avoidance, the hard constraints are relaxed through a dynamic constraint frame strategy, so that a safe obstacle avoidance speed curve is generated.The application can improve the real-time performance, safety and comfort of automatic driving speed planning, and solves the problem of high calculation complexity of a speed planning algorithm, large occupied computing resources and poor real-time performance in the prior art.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to an autonomous driving speed planning method and system. Background Technology

[0002] The statements in this section merely refer to the background art related to this invention and do not necessarily constitute prior art.

[0003] In recent years, with the rapid development of computing technology and autonomous driving algorithms, the feasibility of autonomous vehicles has become increasingly clear. Research has gradually expanded from ensuring autonomous vehicles can safely reach their destinations to incorporating constraints such as comfort into motion planning. During daily commutes or journeys, people expect autonomous driving motion planning to provide a safe and comfortable riding experience.

[0004] In autonomous driving applications, speed planning is a crucial means of improving the comfort and safety of motion planning. This is especially true in urban or highway driving scenarios, where the impact of speed planning on safety and comfort is particularly pronounced. Regarding safety, in the dynamic and complex traffic environments of cities or highways, speed planning is indispensable for lane changing, overtaking, braking and following, and obstacle avoidance. Particularly when there is no space to adjust the path shape to avoid emergency collisions, adjusting the autonomous vehicle's speed to achieve safe obstacle avoidance is a more sensible decision. In terms of comfort, during commutes or journeys, when autonomous vehicles need to avoid obstacles (such as consecutive traffic light intersections or other road users crossing or changing lanes), the speed planning in current technologies often resembles the panicked behavior of a novice human driver, leading to a poor riding experience and potentially even motion sickness. Therefore, good speed planning must not only ensure driving safety but also meet the needs of passenger comfort.

[0005] For velocity planning algorithms to be applied in urban or highway scenarios, they also need to have high real-time performance in order to quickly plan the optimal trajectory in complex environments.

[0006] However, conventional search algorithms are difficult to apply in practice due to their high computational complexity. Most existing algorithms use hard constraints such as acceleration and Jerk's algorithm when solving velocity curves, meaning that the constraint indicators remain unchanged under the same driving scenario. In unknown and complex scenarios, this may cause the planner to fail to solve the problem, resulting in significant safety hazards.

[0007] Current research indicates that Jerk (jerk) is the most influential indicator on comfort in autonomous driving, and it is indispensable for the comfort planning of autonomous vehicles. However, Jerk constraints are non-convex and non-linear, requiring significant computational resources to solve. Introducing Jerk constraints into velocity planning weakens the real-time performance of the algorithm. Summary of the Invention

[0008] To address the shortcomings of existing technologies, this invention provides an autonomous driving speed planning method, system, electronic device, and computer-readable storage medium that obtains the optimal speed planning solution for a given path in the shortest time with minimal computation and logical operations.

[0009] In a first aspect, the present invention provides an autonomous driving speed planning method;

[0010] An autonomous driving speed planning method includes:

[0011] Based on the speed planning requirements of autonomous driving, obstacle constraints and Jerk constraints are introduced.

[0012] Obtain the driving parameters of the autonomous vehicle and the driving parameters of the obstacles. Based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacles, use deceleration as a threshold to convert the obstacle constraint into a maximum speed constraint.

[0013] With the goal of minimizing the approximate velocity, the first linear programming of the Jerk constraints is performed through the maximum velocity constraint, lateral acceleration constraint, and longitudinal acceleration constraint to obtain the approximate velocity curve; the approximate velocity curve is then substituted into the Jerk constraints to linearize them.

[0014] A second linear programming operation is performed on the linearized Jerk constraints to obtain the time-optimal velocity curve.

[0015] Furthermore, in emergency obstacle avoidance situations involving autonomous vehicles, this also includes:

[0016] Based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacles, a dynamic constraint box is constructed; the dynamic constraint box is used to relax the lateral acceleration constraints and the longitudinal acceleration constraints.

[0017] Furthermore, obtaining the minimum and maximum braking deceleration of the autonomous vehicle based on its driving parameters, obstacle driving parameters, and road information includes:

[0018] Based on the driving speed of the autonomous vehicle and the driving speed of the obstacle, the minimum braking deceleration of the autonomous vehicle to avoid the obstacle is obtained, with the distance between the autonomous vehicle and the obstacle on the road as the safe braking distance of the vehicle.

[0019] The maximum braking deceleration of the autonomous vehicle for obstacle avoidance is obtained based on the road surface adhesion coefficient.

[0020] Furthermore, the step of converting obstacle constraints into maximum speed constraints based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacles, using deceleration as a threshold, includes:

[0021] Based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacles, a safe braking distance function for the vehicle is constructed.

[0022] A first deceleration threshold and a second deceleration threshold are set, and the first deceleration threshold and the second deceleration threshold are substituted into the vehicle safe braking distance function to obtain the corresponding vehicle safe braking distance; wherein, the first deceleration threshold is less than the second deceleration threshold;

[0023] The maximum speed constraint of the autonomous vehicle is obtained based on the vehicle's safe braking distance, the autonomous vehicle's speed, and the obstacle's trajectory.

[0024] Preferably, the trajectory of the obstacle is represented as follows:

[0025] s out (t)=v obs (tt in )+s in ,t∈[t in ,t out ]

[0026] Among them, s in Let s be the arc length of the obstacle cut into the current path. out Let t be the arc length cut by the obstacle on the current path. in t is the time it takes for the obstacle to engage. out This is the cut-out time of the obstacle.

[0027] Furthermore, the second linear programming operation on the linearized Jerk constraints to obtain the time-optimal velocity curve specifically involves:

[0028] By applying maximum velocity constraints, lateral acceleration constraints, and longitudinal acceleration constraints, a second linear programming problem is solved on the linearized Jerk constraints to obtain the time-optimal velocity.

[0029] Furthermore, the Jerk constraint is expressed as

[0030]

[0031] Where j is the Jerk expression along path s, defined as the first derivative of acceleration α with respect to time t, β is the square of the first derivative of path s with respect to time t, α is the second derivative of path s with respect to time t, v is the driving speed of the autonomous vehicle, and i is a discrete point on path s.

[0032] Secondly, the present invention provides an autonomous driving speed planning system;

[0033] An autonomous driving speed planning system includes:

[0034] The obstacle speed limit filtering module is configured to: introduce obstacle constraints and Jerk constraints according to the autonomous driving speed planning requirements; obtain the driving parameters of the autonomous vehicle and the driving parameters of the obstacles; and convert the obstacle constraints into maximum speed constraints based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacles, with deceleration as the threshold.

[0035] The Jerk constraint linearization module is configured to: perform a first linear programming operation on the Jerk constraints with the goal of minimizing the approximate velocity, using maximum velocity constraints, lateral acceleration constraints, and longitudinal acceleration constraints to obtain an approximate velocity curve; substitute the approximate velocity curve into the Jerk constraints to linearize the Jerk constraints; and perform a second linear programming operation on the linearized Jerk constraints to obtain the time-optimal velocity curve.

[0036] Thirdly, the present invention provides an electronic device;

[0037] An electronic device includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, complete the steps of the above-described autonomous driving speed planning method.

[0038] Fourthly, the present invention provides a computer-readable storage medium;

[0039] A computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps of the above-described autonomous driving speed planning method.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] 1. The technical solution provided by this invention provides an efficient and real-time linear programming solution method for a determined path during the normal driving process of an autonomous vehicle, thereby quickly solving for the optimal time-speed curve on the path.

[0042] 2. The technical solution provided by this invention addresses the problem that hard constraints lack adaptability in dynamic scenarios. It proposes a dynamic constraint box strategy to soften and relax hard constraints, thereby resolving the decrease in vehicle maneuverability caused by hard constraints in speed planning during emergency obstacle avoidance scenarios for autonomous vehicles.

[0043] 3. The technical solution provided by this invention addresses the Jerk constraint non-convex and nonlinear problem by first using an acceleration and safety obstacle avoidance model to filter the obstacle speed limit, then using two linear programming operations to linearize the Jerk constraint, and finally obtaining the optimal time-velocity curve of the trajectory. This saves computational resources and improves the real-time performance of the autonomous driving speed planning method. Attached Figure Description

[0044] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0045] Figure 1 This is a flowchart provided for an embodiment of the present invention;

[0046] Figure 2 This is a schematic diagram of a vehicle changing lanes in a road scenario provided by an embodiment of the present invention;

[0047] Figure 3 This is a schematic diagram of the dynamic constraint frame for an autonomous vehicle provided in an embodiment of the present invention;

[0048] Figure 4 This is a schematic diagram of the obstacle avoidance speed filtering algorithm provided in an embodiment of the present invention. Detailed Implementation

[0049] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0050] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0051] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0052] Example 1

[0053] This invention provides a method for speed planning in autonomous driving. Next, in conjunction with... Figures 1-4 This embodiment provides a detailed description of an autonomous driving speed planning method.

[0054] The autonomous driving speed planning method includes the following steps:

[0055] S1. Based on the speed planning requirements of autonomous driving, obstacle constraints and Jerk constraints are introduced.

[0056] S2. Obtain the driving parameters of the autonomous vehicle and the obstacles. Based on the driving parameters of the autonomous vehicle and the obstacles, convert the obstacle constraints into maximum speed constraints with deceleration as the threshold. With the goal of minimizing the approximate speed, perform the first linear programming on the Jerk constraints through the maximum speed constraints, lateral acceleration constraints, and longitudinal acceleration constraints to obtain the approximate speed curve. Substitute the approximate speed curve into the Jerk constraints to linearize the Jerk constraints. Perform the second linear programming on the linearized Jerk constraints to obtain the time-optimal speed curve.

[0057] Specifically, to improve the comfort and safety of velocity planning, Jerk constraints and obstacle constraints are introduced, transforming the planning problem into a non-convex and nonlinear problem. Due to its non-convex and nonlinear characteristics, solving it requires significant computational resources, weakening the algorithm's real-time performance. Therefore, this embodiment utilizes obstacle velocity filtering and linear programming to approximate linearize the constraints, maintaining high accuracy while significantly improving the algorithm's computational efficiency. Linearization is completed in three steps:

[0058] The first step of this embodiment proposes a novel obstacle velocity limiting filtering algorithm using acceleration as a threshold, transforming obstacle avoidance constraints into maximum velocity constraints, thus completing the linearization process of obstacle avoidance constraints. The second step, based on the first step, involves a first linear programming operation based on the objective and constraints to obtain an approximate velocity curve. This approximate velocity curve is then substituted into Jerk constraints to linearize them. The third step, after adding the linearized Jerk constraints, involves a second linear programming operation to finally obtain the time-optimal velocity curve, completing the velocity programming solution process.

[0059] For example, it includes the following steps:

[0060] S201, Obstacle Speed ​​Limiting Filter

[0061] This embodiment introduces a new method for obstacle avoidance by modifying the maximum speed constraint using acceleration (which is deceleration during braking) as a threshold, as detailed below:

[0062] First, assume that the trajectory of the dynamic obstacle (i.e., the obstacle constraint) is a continuous linear function of its velocity as shown in (1):

[0063] s out (t)=v obs (tt in )+s in , t∈[t in ,t out (1)

[0064] Among them, s in s out t represents the arc lengths at which obstacles enter and exit the current path, respectively. in and t out These represent the entry and exit times of the obstacle, respectively. For example... Figure 4 As shown (dashed lines represent the trajectories of dynamic obstacles, and solid lines represent the trajectories of the autonomous vehicle under obstacle constraints), two deceleration thresholds α are defined. A and α B , so that α A ≤α B , will α A α B Substitute the values ​​into the vehicle's safe braking distance formula to calculate the corresponding safe threshold distance D. A and D B (The derivative of velocity with respect to time is acceleration, according to the deceleration threshold α) A and α B The corresponding speed can be calculated, and then the safe braking distance can be determined using the following formula: The formula for the safe braking distance of a vehicle is as follows:

[0065]

[0066] First, the autonomous vehicle maintains its original maximum speed until the distance to the dynamic obstacle becomes D. A When the distance to the obstacle is less than D A At that time, the maximum speed is in s A When decelerating α A The vehicle undergoes uniform deceleration. Let the current discrete point be k, and k+1 be the next discrete point. The speed adjustment during the pre-deceleration process of the autonomous vehicle is as follows:

[0067]

[0068] When the distance between the autonomous vehicle and the obstacle is less than D BAt that time, with a deceleration α B The autonomous vehicle adjusts its speed as follows to undergo uniform deceleration:

[0069]

[0070] Distance less than D B When the speed decreases to match the obstacle's speed, it maintains the same speed as the obstacle and follows it at the same speed, thus completing obstacle speed filtering. Once the obstacle disappears from the autonomous vehicle's path, the original maximum speed constraint is restored under the maximum acceleration constraint. The obstacle speed limiting filter completes the first step of constraint linearization.

[0071] The specific algorithm is shown in Algorithm 1.

[0072]

[0073]

[0074] Among them, v max The initial maximum velocity constraint, the maximum velocity constraint for N discrete trajectory points can be expressed as v max :=[v max,1 ,...,v max,N ] T s ego The distance s represents the distance along the path traveled by the autonomous vehicle. ego :=[s1,...,s N ] T , This represents the speed constraint output after obstacle speed limit filtering.

[0075] S202. After the obstacle velocity filtering process is completed, the maximum velocity constraint has been calculated, and the values ​​of the longitudinal acceleration constraint and the lateral acceleration constraint are preset. The following focuses on how to handle the Jerk constraint.

[0076] Jerk constraints are represented as follows:

[0077]

[0078] Where j is the Jerk expression along path s, defined as the first derivative of acceleration α with respect to time t, β is the square of the first derivative of path s with respect to time t, α is the second derivative of path s with respect to time t, v is the driving speed of the autonomous vehicle, and i is the discrete sampling point on path s.

[0079] As can be seen from the above analysis, the reason why the Jerk constraint becomes nonlinear is the bilinear term. This means that if one of the linear terms v can be fixed... iThis would make the equation linear. The basic idea of ​​the current research is to roughly estimate an optimized velocity profile and use it to replace v during optimization. i This makes the Jerk constraint linear.

[0080] The original Pseudo-Jerk method:

[0081]

[0082] This method involves omitting the linear term v. i This transforms the Jerk problem from solving for the first derivative of acceleration with respect to time to solving for the first derivative of acceleration with respect to arc length *s*, thus linearizing the Jerk constraint. However, the Pseudo-Jerk method differs significantly from the actual Jerk constraint value over a wide speed range (e.g., at high speeds, because the speed term is omitted, the Pseudo-Jerk value is lower than the actual Jerk value, and when the autonomous vehicle's speed is 0, the Pseudo-Jerk value is infinite).

[0083] Another Jerk linearization method involves using the maximum acceleration and Jerk's linearity to perform forward-backward Jerk filtering, which yields an estimated maximum velocity curve v. p To fix v i The forward-backward Jerk method essentially amplifies the current velocity v. i Therefore, in The constraints actually narrow the actual range of Jerk constraints and reduce the actual dynamic range of Jerk. This not only restricts the change in acceleration, but the hard constraints of Jerk can also lead to the inability to obtain a solution for linear programming in the critical case of Jerk.

[0084] Therefore, based on the original research, this embodiment proposes a new Jerk linearization method based on linear programming.

[0085] First, without considering the nonlinear Jerk constraints, a more accurate predicted velocity profile is obtained using the first linear programming approach. The result of the first linear programming approach is expressed as follows: The result As an approximate optimal solution, it replaces the bilinear term in the Jerk constraints. v in i The Jerk constraints are then linearized. The linearized Jerk constraints are then substituted into the problem, and a second linear programming solution is performed to obtain the final velocity curve.

[0086] The objective and constraints of the first linear programming problem are as follows:

[0087]

[0088] Subject to

[0089] β i+1 -β i =2α i (s i+1 -s i (8)

[0090]

[0091]

[0092]

[0093] Under the constraints of vehicle kinematics (8), maximum speed (9), longitudinal acceleration (10), and lateral acceleration (11), the function (7) is solved by linear programming.

[0094] Lateral acceleration constraints are introduced to prevent the vehicle from skidding during operation.

[0095] The problem presented in this embodiment can be proven by the formula to be convex. From (9) and (10), it can be seen that for the target, it is a negative power function, therefore, it is convex. For the decision variables β and α in velocity planning, since the derivative is a linear operator, the relationship between β and α is convex. For the velocity and acceleration constraint inequalities, the equality and inequality constraints of β and α are linear constraints, therefore, they are convex. For the lateral acceleration constraint... The variable is curvature (equivalent to a constant value), and the constraint is the square function of the absolute value of curvature, so it is convex. In the proposed optimization problem, since problem (7) is convex, and the equality constraints (8) are mutually mapped, and the inequality constraints (9)(10)(11) are convex, the optimization problem in this embodiment is convex.

[0096] S203. After obstacle speed limit filtering and linearization of Jerk constraints, the linearized Jerk constraints obtained after the first linear programming (as shown in formula (12)) are substituted into the function, and the function (7) is solved by the second linear programming under the constraints (8)(9)(10)(11).

[0097]

[0098] This is a linear programming problem that can be solved efficiently. After obstacle speed limit filtering and two linear programming iterations, the time-optimal speed v is finally obtained. i for:

[0099]

[0100] This method of approximating Jerk's algorithm with linear programming results is more accurate than the original coarse linearization method, making the solution of the programming algorithm more complete. The optimal velocity solution is obtained through the first linear programming iteration, therefore the optimization result will not exceed the speed and acceleration constraints. Since the cost function is designed to maximize the speed, the velocity profile obtained after the first linear programming iteration will not significantly deviate from the actual speed value. This approximation approach allows the Jerk constraints to maintain a higher level of accuracy. In the second linear programming iteration, more precise Jerk constraints are added, resulting in a final velocity profile that, while satisfying the constraints, has a more accurate optimal solution.

[0101] Example 2

[0102] When an autonomous vehicle is in an emergency obstacle avoidance situation, the hard constraints cannot be successfully solved for the constraints in Example 1. Therefore, the autonomous driving speed planning method provided in this example includes the following steps:

[0103] S1. Obtain the driving parameters of the autonomous vehicle, the driving parameters of the obstacles, and the road information. Based on the driving parameters of the autonomous vehicle, the driving parameters of the obstacles, and the road information, obtain the minimum braking deceleration and the maximum braking deceleration of the autonomous vehicle for obstacle avoidance. Based on the minimum braking deceleration and the maximum braking deceleration, construct a dynamic constraint box for autonomous driving speed planning.

[0104] When hard constraints cannot meet the requirements for safe obstacle avoidance, the comfort constraints are dynamically relaxed according to the vehicle's safe braking distance to achieve the effect of a dynamic constraint box and generate a safe speed profile to adapt to different types of dynamic environments and achieve safe obstacle avoidance.

[0105] For example, such as Figure 2 In the scenario shown, Ego is an autonomous vehicle, and other vehicles are traffic participants. When the prediction module predicts that a traffic participant is about to enter the autonomous vehicle's driving path, the autonomous vehicle needs to brake to avoid the obstacle. However, in a dynamic environment, the constraints in the hard constraints may cause the speed planner to fail to solve the problem and cannot obtain a safe obstacle avoidance solution.

[0106] When the autonomous vehicle travels at the planned speed v c0 During driving, the perception and prediction module detects a path with the prefix v at the current path S1. cf The vehicle traveling on the obstacle needs to brake to avoid it. The entire braking process is divided into a braking preparation phase, an initial braking phase, and a full braking phase.

[0107] First, the braking preparation phase is the time t1 from the issuance of the braking command to the operation of the brakes, also known as the braking delay time. During this time, the vehicle maintains its initial speed and travels at a constant speed. This section considers this time as the average time for the speed planner to perform one planning operation. Second, in the braking initiation phase, assuming the vehicle's braking deceleration is α1, under the constraint of maximum Jerk (j max The current acceleration changes to α1 after time t2, and this stage lasts for t2. Finally, the full braking stage is when the vehicle decelerates uniformly until it reaches the same speed as the vehicle in front (for static obstacles, the vehicle stops and its speed is 0), and this stage lasts for t3.

[0108] The safe braking distance of the vehicle is D1.

[0109]

[0110] Under hard constraints, when the planner fails to solve the problem, the minimum braking deceleration of the autonomous vehicle to avoid obstacles is calculated by taking the distance S1 of the autonomous vehicle on the path at the initial moment of braking as the safe braking distance of the vehicle:

[0111]

[0112] The relaxation range of the constraint box is not infinite. For example, the maximum braking deceleration of an autonomous vehicle is constrained by the road adhesion coefficient, and the formula for calculating the maximum value is:

[0113]

[0114] μ is the road surface adhesion coefficient, and μg is the maximum deceleration of the vehicle. The dynamic constraint box range can be obtained through (15) and (16).

[0115] Based on the initial lateral and longitudinal acceleration constraints and the road adhesion coefficient, the maximum safe acceleration range of the vehicle can be calculated as follows: Figure 2 The vehicle constraint box shown.

[0116] In the figure, the coordinate axis α L α N These are the vehicle's longitudinal acceleration and lateral acceleration, respectively. When hard constraints cannot meet the safety obstacle avoidance requirements, the planner can more fully utilize the autonomous vehicle's maneuverability to generate a safe speed curve by relaxing comfort constraints. By substituting the current autonomous vehicle speed and Jerk parameters into equation (15), the deceleration constraint box range required for safe obstacle avoidance can be calculated, as shown below. Figure 3The relaxed constraint box, shown by the dashed line, achieves the effect of an adaptive dynamic constraint box. This dynamic constraint box maximizes comfort while utilizing the autonomous vehicle's greater maneuverability to ensure safe driving. Ultimately, the algorithm dynamically and adaptively adjusts the autonomous vehicle's driving parameters for different traffic conditions, completing speed planning for safe obstacle avoidance.

[0117] S2. Based on the speed planning requirements of autonomous driving, obstacle constraints and Jerk constraints are introduced.

[0118] S3. Obtain the driving parameters of the autonomous vehicle and the obstacles. Based on the driving parameters of the autonomous vehicle and the obstacles, convert the obstacle constraints into maximum speed constraints with deceleration as the threshold. With the goal of minimizing the approximate speed, perform the first linear programming on the Jerk constraints through the maximum speed constraints, lateral acceleration constraints, and longitudinal acceleration constraints to obtain the approximate speed curve. Substitute the approximate speed curve into the Jerk constraints to linearize the Jerk constraints. Perform the second linear programming on the linearized Jerk constraints to obtain the time-optimal speed curve.

[0119] Unlike Embodiment 1, the values ​​of the lateral acceleration constraint and the longitudinal acceleration constraint in step S3 are based on the dynamic constraint box setting. The other contents in steps S2 and S3 are the same as in Embodiment 1, and will not be repeated here.

[0120] The method described in this embodiment can calculate the final velocity curve, acceleration curve, and Jerk curve within a planning time of tens of milliseconds, all of which are within the constraint range. Furthermore, it can relax hard constraints using a hard constraint dynamic relaxation algorithm under emergency vehicle conditions, thus planning a reasonable obstacle avoidance speed curve.

[0121] Experimental results data at different vehicle speeds in the same scenario

[0122]

[0123] Example 2

[0124] This embodiment discloses an autonomous driving speed planning system, including:

[0125] The obstacle speed limit filtering module is configured to: introduce obstacle constraints and Jerk constraints according to the autonomous driving speed planning requirements; obtain the driving parameters of the autonomous vehicle and the driving parameters of the obstacles; and convert the obstacle constraints into maximum speed constraints based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacles, with deceleration as the threshold.

[0126] The Jerk constraint linearization module is configured to: perform a first linear programming operation on the Jerk constraints with the objective of minimizing the approximate velocity, using maximum velocity constraints, lateral acceleration constraints, and longitudinal acceleration constraints to obtain an approximate velocity curve; substitute the approximate velocity curve into the Jerk constraints to linearize them; and perform a second linear programming operation on the linearized Jerk constraints to obtain the time-optimal velocity curve. It should be noted that the obstacle velocity limit filtering module and the Jerk constraint linearization module described above correspond to the steps in Embodiment 1. The examples and application scenarios implemented by these modules and their corresponding steps are the same, but they are not limited to the content disclosed in Embodiment 1. It should also be noted that these modules, as part of a system, can be executed in a computer system, such as a set of computer-executable instructions.

[0127] Example 3

[0128] Embodiment 3 of the present invention provides an electronic device, including a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When the computer instructions are executed by the processor, they complete the steps of the above-mentioned autonomous driving speed planning method.

[0129] Example 4

[0130] Embodiment 4 of the present invention provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps of the above-described autonomous driving speed planning method.

[0131] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0132] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0133] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment, whereby a series of operational steps are performed to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0134] The descriptions of each embodiment in the above embodiments have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0135] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An autonomous driving speed planning method, characterized in that, include: Based on the speed planning requirements of autonomous driving, obstacle constraints and Jerk constraints are introduced. The driving parameters of the autonomous vehicle and the obstacles are obtained. Based on these parameters, and using deceleration as a threshold, the obstacle constraints are converted into maximum speed constraints. The obstacle trajectory is represented as follows: in, Let arc length be the distance cut by obstacles on the current path. The arc length cut by the obstacle on the current path. The time it takes for the obstacle to engage. The cut-out time of the obstacle; Define two deceleration thresholds as well as ,make ,Will , Substitute the values ​​into the vehicle's safe braking distance formula to calculate the corresponding safe threshold distance. as well as The formula for safe braking distance of a vehicle is expressed as follows: in, Indicates the vehicle's safe braking distance; This represents the maximum Jerk constraint value; Indicates the time of the braking preparation phase; Indicates the start time of the braking phase; Indicates the duration of the full braking phase; Indicates the vehicle's braking deceleration; The autonomous vehicle maintains its original maximum speed until the distance to the dynamic obstacle becomes... When the distance to the obstacle is less than At that time, the maximum speed was deceleration Perform uniformly decelerated motion; define the current discrete point as... , For the next discrete point, the speed adjustment during the pre-deceleration process of the autonomous vehicle is as follows: When the distance between the autonomous vehicle and the obstacle is less than At that time, with deceleration The autonomous vehicle adjusts its speed as follows to undergo uniform deceleration: Distance less than When the speed decreases to the same speed as the obstacle, it maintains the same speed as the obstacle and follows it at the same speed, thus completing the obstacle speed filtering; when the obstacle disappears from the path of the autonomous vehicle, the original maximum speed constraint is restored under the maximum acceleration constraint. With the goal of minimizing the approximate velocity, the first linear programming of the Jerk constraints is performed through the maximum velocity constraint, lateral acceleration constraint, and longitudinal acceleration constraint to obtain the approximate velocity curve; The approximate velocity curve is then substituted into the Jerk constraint to linearize the Jerk constraint. A second linear programming operation is performed on the linearized Jerk constraints to obtain the time-optimal velocity curve; The Jerk constraint is expressed as in, For Jerk along the path The expression for acceleration is defined as follows: Regarding time The first derivative; For path Regarding time The square of the first derivative; For path Regarding time The second derivative; v denoted as the driving speed of the autonomous vehicle; i represents a discrete point on path s. First, without considering nonlinear Jerk constraints, a more accurate predicted velocity profile is obtained using the first linear programming approach. The result of the first linear programming approach is expressed as follows: , the results As an approximate optimal solution, it replaces the bilinear term in the Jerk constraints. In The linearization of the Jerk constraints is completed; the linearized Jerk constraints are then substituted into the problem, and a second linear programming solution is performed to obtain the final velocity curve. The objective of the first linear programming problem is as follows: The constraints for the first linear programming problem are as follows: Vehicle kinematic constraints: ; Maximum speed constraint: ; Longitudinal acceleration constraint: ; Lateral acceleration constraints: ; Under constraints, the function is solved using linear programming; for optimization problems, the above formula is proven to be convex. After obstacle speed limiting filtering and linearization of the Jerk constraints, the linearized Jerk constraints obtained from the first linear programming are substituted into the function, and a second linear programming problem is performed to solve the function under the constraints: in, This represents the speed-optimal solution obtained from the first linear programming iteration; After obstacle speed limiting filtering and two linear programming iterations, the time-optimal speed is finally obtained. for: ; The aforementioned autonomous driving speed planning method further includes, in the event of an emergency obstacle avoidance by the autonomous vehicle: Based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacles, a dynamic constraint box is constructed; the dynamic constraint box is used to relax the lateral acceleration constraints and the longitudinal acceleration constraints. The process of constructing a dynamic constraint box based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacles includes: Based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacle, obtain the minimum braking deceleration and maximum braking deceleration of the autonomous vehicle. Construct a dynamic constraint box based on the minimum and maximum braking deceleration of the autonomous vehicle; Among them, converting obstacle constraints into maximum velocity constraints using deceleration as a threshold includes: Based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacles, a safe braking distance function for the vehicle is constructed. A first deceleration threshold and a second deceleration threshold are set, and the first deceleration threshold and the second deceleration threshold are substituted into the vehicle safe braking distance function to obtain the corresponding vehicle safe braking distance; wherein, the first deceleration threshold is less than the second deceleration threshold; The maximum speed constraint of the autonomous vehicle is obtained based on the vehicle's safe braking distance, the autonomous vehicle's speed, and the obstacle's trajectory.

2. The autonomous driving speed planning method as described in claim 1, characterized in that, The specific steps for performing a second linear programming operation on the linearized Jerk constraints to obtain the time-optimal velocity curve are as follows: By applying maximum velocity constraints, lateral acceleration constraints, and longitudinal acceleration constraints, a second linear programming problem is solved on the linearized Jerk constraints to obtain the time-optimal velocity.

3. An autonomous driving speed planning system, employing an autonomous driving speed planning method as described in any one of claims 1-2, characterized in that, include: The obstacle speed limit filtering module is configured to: introduce obstacle constraints and Jerk constraints according to the autonomous driving speed planning requirements; obtain the driving parameters of the autonomous vehicle and the driving parameters of the obstacles; and convert the obstacle constraints into maximum speed constraints based on the driving parameters of the autonomous vehicle and the driving parameters of the obstacles, with deceleration as the threshold. The Jerk constraint linearization module is configured to: take the minimum approximate velocity as the objective, perform a first linear programming on the Jerk constraint through the maximum velocity constraint, lateral acceleration constraint, and longitudinal acceleration constraint to obtain an approximate velocity curve; substitute the approximate velocity curve into the Jerk constraint to linearize the Jerk constraint; and perform a second linear programming on the linearized Jerk constraint to obtain the time-optimal velocity curve.

4. An electronic device, characterized in that, It includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, complete the steps of the autonomous driving speed planning method according to any one of claims 1-2.

5. A computer-readable storage medium, characterized in that, Used to store computer instructions, which, when executed by a processor, complete the steps of an autonomous driving speed planning method as described in any one of claims 1-2.

Images