Trajectory planning method and system for vehicles to perceive blind spots when turning at non-signalized intersections

Abstract

This invention relates to the field of autonomous driving technology and discloses a trajectory planning method and system for vehicles to perceive blind spots when turning at non-signaled intersections. The trajectory planning method for vehicles to perceive blind spots when turning at non-signaled intersections includes the following steps: S1. Establishing a kinematic model of the vehicle trajectory; S2. Constructing an objective function J; S3. Establishing constraints on the data in step S2 and using the constraints to optimize the objective function J, obtaining an optimized overall optimization model; S4. Based on the overall optimization model, solving for the optimized trajectory using a nonlinear optimization solver. This invention can improve the traffic efficiency and safety of autonomous vehicles.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to a trajectory planning method and system for perceiving blind spots when a vehicle turns at a non-signal-controlled intersection. Background Technology

[0002] When autonomous vehicles pass through unsignalized intersections, they need to plan a safe and feasible driving trajectory based on information such as the surrounding environment, road maps, and vehicle status. However, parked vehicles or buildings at intersections may create blind spots in environmental perception, preventing autonomous vehicles from acquiring complete environmental information.

[0003] Existing autonomous vehicle trajectory planning methods typically assume that the vehicle can acquire all environmental perception information, but this assumption is unreasonable. Due to the potential presence of vehicles or pedestrians in perception blind spots, the lack of assessment of these risks often leads autonomous vehicles to take conservative actions such as slowing down or stopping when approaching intersections, thereby affecting traffic efficiency and even causing the vehicle to "freeze."

[0004] Therefore, there is an urgent need for a new type of trajectory planning method and system for vehicles to perceive blind spots when turning at non-signal-controlled intersections to solve the above problems. Summary of the Invention

[0005] (a) Technical problems to be solved

[0006] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a trajectory planning method and system for vehicles to perceive blind spots when turning at non-signal-controlled intersections, so as to solve the problems of the prior art.

[0007] (II) Technical Solution

[0008] To achieve the above objectives, a trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection includes the following steps:

[0009] S1, through the formula:

[0010] Establish a kinematic model of the vehicle trajectory; among which, Let be the derivative of the vehicle's position x-coordinate y. Let be the derivative of the vehicle's position x-coordinate y. vehicle heading angle The derivative, Let be the derivative of the vehicle's steering angle δ. Let x be the derivative of the velocity; x be the x-coordinate of the vehicle's position; and y be the y-coordinate of the vehicle's position. δ is the heading angle, v is the speed, a is the acceleration, and L is the distance between the front and rear wheels of the vehicle.

[0011] S2, through the formula: Construct the objective function J;

[0012] Where, N p For prediction in the time domain, Q is its weight matrix, R is its weight matrix, and J is... obs,i J is the obstacle avoidance function. visibility,i U is the visibility function. i To control the input sequence, η(t+i|t) is the predicted state variable, η ref (t+i|t) is the reference state variable;

[0013] S3. Establish kinematic model constraints and use the constraints to optimize the objective function J to obtain the optimized overall optimization model.

[0014] S4. Based on the overall optimization model, the optimized trajectory is solved using a nonlinear optimization solver.

[0015] Preferably, the constraints in step S3 include one or more of the following steps:

[0016] S31, using the formula: 0≤v k+i|k ≤v max (1) Establish velocity constraints; where v max v represents the maximum speed the vehicle can reach, k represents time k, and v k+i|k Let k be the predicted velocity value for the i-th future step at time k;

[0017] S32, through formula: a min ≤a k+i|k ≤a max (2) Establish acceleration constraints; where a min a is the maximum deceleration that the vehicle can achieve. max a is the maximum acceleration that the vehicle can achieve. k+i|k Let be the predicted acceleration value for the i-th future step at time k;

[0018] S33, through the formula: δ min ≤δ k+i|k ≤δ max (3) Establish steering angle constraints; where δ min δ is the minimum steering angle that the vehicle can achieve. max δ is the maximum steering angle that the vehicle can achieve. k+i|k Let be the predicted steering angle for the i-th future step at time k;

[0019] S34, through the formula: x min ≤x k+i|k ≤x max(4) Establish boundary constraints; where x min x represents the minimum lateral movement distance from the vehicle reference line to the left and right road boundaries. max x represents the maximum lateral movement distance of the vehicle from the reference line to the left and right road boundaries. k+i|k Let x be the predicted x-coordinate of the position at time k for the i-th future step.

[0020] S35, via formula: r k+1 =P(Z) t+1 |c t+1 If r ≤ ε(5), establish risk constraints; where ε is the risk threshold and r k+1 Z is the risk value calculated at time t+1. t+1 Z ∈{0,1} represents whether a risky vehicle is present in the oncoming lane at the blind spot intersection at time t+1. t+1 =1 indicates that the vehicle appears at time t+1, Z t+1 =0 indicates that no vehicle appears at time t+1; C represents the observation result, c t +1 ∈{0,1} represents the corresponding observation probability value, c t+1 =0 indicates that no vehicle was observed at time t+1, c t+1 =1 indicates that the vehicle was observed at time t+1; P(Z) t+1 |c t+1 ) represents the posterior probability at time t+1.

[0021] Preferably, by formula: Establish an overall optimization model.

[0022] Preferably, J in step S2 obs,i The calculation formula is:

[0023]

[0024] Where i represents time i, W obs For the weighting coefficients, (x) obs-i ,y obs-i (x) represents the position coordinates of the visible obstacle relative to the autonomous vehicle's body coordinate system at time i. i ,y i ) represents the centroid coordinates of the autonomous vehicle at time i, and ζ is used to prevent positive numbers with a denominator of 0. i Let be the speed of the vehicle at time i.

[0025] Preferably, J in step S2 visibility,i The calculation formula is:

[0026]

[0027] Among them, Wvis θ is the weighting coefficient. i Let Δx be the occlusion angle at time i. i and Δy i Let be the distance between the vehicle's position coordinates at time i and the blind spot coordinates of the slower point.

[0028] Preferably, the blind spot is the vertex of the perception blind zone caused by the obstruction of roadside buildings, and the coordinates of the blind spot are the geographic coordinates provided by the map positioning system of the vertex.

[0029] Preferably, in step S35, P(Z) t+1 |c t+1 The formula for calculating ) is:

[0030]

[0031] Among them, P(Z) t+1 |c t+1 Let P(Z) be the posterior probability at time t+1. t+1 =1) is the prior probability that the at-risk vehicle will appear at time t+1, P(c t+1 |Z t+1 =0) is the probability of observing a vehicle at time t+1 if no risky vehicle appears. P(c t+1 |Z t+1 =1) is the probability of observing a vehicle given that a risky vehicle appears at time t+1, and P(C=c) is the prior probability distribution of the observation result C.

[0032] Preferably, the kinematic model described in step S1 is a combination of the bicycle kinematic model and the MPC model.

[0033] Preferably, the method further includes step S36, establishing the calculation formula for the kinematic constraints as follows: s k+1 =f(s) k ,u k ); where s k+1 Let u be the state variable at time k+1. k Let s be the control quantity at time k. k Let f be the state variable at time k, and let f be the state transition function obtained from the state transition matrix.

[0034] 10. A trajectory planning system for perceiving blind spots when a vehicle turns at a non-signal-controlled intersection, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the trajectory planning method for perceiving blind spots when a vehicle turns at a non-signal-controlled intersection as described in any of the preceding claims.

[0035] (III) Beneficial Effects

[0036] This invention can improve the efficiency of passing through non-signal-controlled intersections by using reasonable trajectory planning and risk assessment to avoid autonomous vehicles from taking overly cautious actions due to blind spots, thereby preventing vehicles from getting stuck in a "frozen state" and improving driving efficiency. Attached Figure Description

[0037] Figure 1 This is a flowchart illustrating a trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection, according to the present invention.

[0038] Figure 2 This is a schematic diagram of the cubic Bézier curve generation method of the present invention;

[0039] Figure 3 This is a schematic diagram of the kinematic model of the present invention;

[0040] Figure 4 This is a schematic diagram of the optimized curve of the present invention;

[0041] Figure 5 This is a schematic diagram of risk assessment for a dynamic Bayesian network model.

[0042] Figure 6 This is a schematic diagram of a dynamic Bayesian network structure.

[0043] Figure 7 This is a schematic diagram of the planning system process. Detailed Implementation

[0044] To better explain and facilitate understanding of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0046] Furthermore, in this invention, descriptions involving "first," "second," etc., are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0047] To achieve the above objectives, please refer to Figures 1 to 7 A trajectory planning method for vehicles to perceive blind spots when turning at non-signaled intersections includes the following steps:

[0048] S1, through the formula:

[0049] Establish a kinematic model of the vehicle trajectory; among which, Let be the derivative of the vehicle's position x-coordinate y. Let be the derivative of the vehicle's position x-coordinate y. vehicle heading angle The derivative of the value describes the velocity at which the vehicle is changing direction. The derivative of the vehicle's steering angle δ describes the rate of change of the steering angle with time. Let x be the derivative of the velocity, x be the x-coordinate of the vehicle's position, and y be the y-coordinate of the vehicle's position. δ is the heading angle, δ is the steering angle, v is the velocity, a is the acceleration, and L is the distance between the front and rear wheels of the vehicle.

[0050] S2, construct the objective function J using the formula:

[0051]

[0052] Where, N p For prediction in the time domain, Q is its weight matrix, R is its weight matrix, and J is... obs,i J is the obstacle avoidance function. visibility,i U is the visibility function. i To control the input sequence, η(t+i|t) is the predicted state variable, η ref (t+i|t) is the reference state variable;

[0053] S3. Establish constraints on the data in step S2, and use the constraints to optimize the objective function J to obtain the optimized overall optimization model.

[0054] S4. Based on the overall optimization model, the optimized trajectory is solved using a nonlinear optimization solver.

[0055] In step S1, since the driving task of the autonomous vehicle passing through the non-line-controlled intersection is a low-speed scenario, the vehicle's speed, acceleration, and other parameters will not be too large, and the friction force is not saturated. Therefore, there is no need to consider the vehicle's friction force, tire slippage, and other dynamic parameters. The bicycle kinematic model simplifies vehicle dynamics and ignores the longitudinal and lateral dynamic coupling of the vehicle, abstracting the vehicle's kinematic characteristics into a bicycle model, making the control algorithm easier to implement and compute. At the same time, the bicycle kinematic model can accurately describe the vehicle's motion in a two-dimensional plane, making it suitable for trajectory planning and tracking problems, especially trajectory optimization problems for vehicles with low or medium speeds. Therefore, the bicycle kinematic model is selected in combination with the Model Predictive Control (MPC) model for trajectory optimization solutions.

[0056] Preferably, the kinematic model described in step S1 is a combination of the bicycle kinematic model and the MPC model; see details below. Figure 3 x k That is, the vehicle's current coordinates, y k That is, the vehicle's current coordinates, φ k This refers to the vehicle's current heading angle, v is the vehicle's speed, and L is the distance between the vehicle's front and rear wheels.

[0057] For details, please refer to [link / reference]. Figure 2 The specific approach to constructing the objective function J is as follows:

[0058] The path curve segments are generated using cubic Bézier curves. The principle is that the shape of the Bézier curve can be changed by selecting control points to meet the requirements of the autonomous vehicle's driving path.

[0059] A cubic Bézier curve requires four coordinate points for control, denoted as P0, P1, P2, and P3 respectively.

[0060] P0 (Starting Point): The location where the driverless vehicle is about to enter the intersection.

[0061] P3 (Destination): The point on the target lane that the driverless vehicle enters after completing the left turn.

[0062] P1 (First Control Point): The initial curve used to guide the left turn, located on the extension of P0, offset to the left by an appropriate distance.

[0063] P3 (Second Control Point): Used to determine the middle section of the left-turn curve, located on the extension of P3, offset to the left to ensure the curve ends smoothly in the target lane.

[0064] The coordinates of the four control points can be pre-calculated using parameters such as the intersection shape and lane coordinates derived from the high-definition map provided by the map positioning system.

[0065] The formula for calculating the parametric equation B(t) of a cubic Bézier curve is:

[0066] B(t)=P0(1-t) 3 +3P1(1-t) 2 t+3P2(1-t)t 2 +P3t 3 ,

[0067] Where B(t) represents the generated curve points, t is a parameter ranging from 0 to 1, and P i (i = 0, 1, 2, 3) are the coordinates of the control points.

[0068] By discretizing the parameter t—that is, selecting a series of t values ​​at equal intervals between 0 and 1—the corresponding curve points are calculated for each discrete t value using the parametric equation of a cubic Bézier curve. Compared to search-based methods such as A* and Dijkstra, the cubic Bézier curve-based method has advantages such as continuous and differentiable curvature, ease of tracking, and simple implementation (only a few control points are needed to generate a path). Those skilled in the art will not elaborate further here.

[0069] Since global planning does not consider dynamic obstacles, and intersections generally do not have static obstacles, to enable autonomous vehicles to pass through intersections as quickly as possible, the desired speed can be set to the maximum safe speed determined by the road's maximum speed limit and turning radius constraints. The desired speed V ref The calculation formula is:

[0070]

[0071] Among them, v limit Let r be the maximum speed limit on the road, g be the turning radius, and μ be the road surface friction coefficient.

[0072] Preferably, the obstacle avoidance function J in step S2 obs,i The calculation formula is:

[0073]

[0074] Where i represents time i, W obs For the weighting coefficients, (x) obs-i ,y obs-i (x) represents the position coordinates of the visible obstacle relative to the autonomous vehicle's body coordinate system at time i. i ,y i ) represents the centroid coordinates of the autonomous vehicle at time i, and ζ is used to prevent positive numbers with a denominator of 0. i Let be the speed of the vehicle at time i.

[0075] Before an autonomous vehicle enters an intersection, some lanes become blind spots due to obstruction by roadside buildings. However, the size of the obstructed area changes as the autonomous vehicle moves. The degree of obstruction is quantified by acquiring key points of the obstruction and calculating the obstruction angle. A visibility reward function is then designed based on the characteristics of changes in the obstruction angle.

[0076] Define the occlusion angle θ as the angle formed between the light emitted by the sensor of the autonomous vehicle and the blind spot O. Then, the occlusion angle θ at a certain moment i before the autonomous vehicle enters the intersection is... i The calculation formula is:

[0077]

[0078] Where, Δyi Δx i These are the position coordinates (x, y) of the autonomous vehicle at time i. i ,y i ) and the coordinates of blind spot B (x occlued ,y occlued The distance Δy i Δx i The calculation formula is:

[0079] Δx i =|x i -x occlued |,Δy i =|y i -y occlued |,

[0080] The occlusion angle exists until the autonomous vehicle enters the intersection, after which the occlusion disappears. As the autonomous vehicle approaches the intersection, the occlusion angle decreases. Similarly, the occlusion angle decreases as the autonomous vehicle moves away from buildings.

[0081] Based on these characteristics, a visibility function can be designed to guide autonomous vehicles to increase their visibility of blind spots by changing their position before entering an intersection, thereby acquiring more environmental perception information and achieving the effect of maximum information gain.

[0082] Preferably, J in step S2 visibility,i The calculation formula is:

[0083]

[0084] Among them, W vis These are the weighting coefficients.

[0085] The visibility function is defined as the ratio of the occlusion angle to the longitudinal coordinate distance. There are two main reasons for this: First, when the longitudinal distance between the autonomous vehicle and the blind spot O is too large, it will cause J... visibility The result approaches 0. At this point, the motion of the autonomous vehicle is not affected by the visibility function, and it will not prematurely change its lateral position, i.e., the formula is:

[0086]

[0087] Second, the closer the autonomous vehicle gets to the intersection and the closer its longitudinal distance to blind spot B, the more likely J will be to... visibility The result is approximately the same as Then, by increasing Δx, J can be made to... visibility Decrease to increase the visible area, i.e., the formula:

[0088] Preferably, the blind spot is the vertex of the perception blind zone caused by the obstruction of roadside buildings, and the coordinates of the blind spot are the geographical coordinates provided by the map positioning system of the vertex; when entering a non-signal-controlled intersection, the vehicle can adjust its driving path according to the location of the blind spot to minimize the potential collision risk and improve driving safety and efficiency.

[0089] The working principle of autonomous vehicle perception can be considered as sensors collecting information about their surroundings by emitting light in all directions. If an object exists within the sensor's working range, and the emitted light cannot penetrate the object, a blind spot will appear behind the object. For details based on this principle, please refer to [link / reference needed]. Figure 2 It can be seen that the top left corner of the building is the key point causing the perception blind spot. The coordinates of this point can be obtained through the high-definition map provided by the map positioning system, and denoted as blind point O(x). occlued ,y occlued ).

[0090] Furthermore, the constraints in step S3 include one or more of the following steps:

[0091] S31, using the formula: 0≤v k+i|k ≤v max (1) Establish velocity constraints; where v max v represents the maximum speed the vehicle can reach, k represents time k, and v k+i|k Let k be the predicted velocity value for the i-th future step at time k;

[0092] S32, through formula: a min ≤a k+i|k ≤a max (2) Establish acceleration constraints; where a min a is the maximum deceleration that the vehicle can achieve. max a is the maximum acceleration that the vehicle can achieve. k+i|k Let be the predicted acceleration value for the i-th future step at time k;

[0093] S33, through the formula: δ min ≤δ k+i|k ≤δ max (3) Establish steering angle constraints; where δ min δ is the minimum steering angle that the vehicle can achieve. max δ is the maximum steering angle that the vehicle can achieve. k+i|k Let be the predicted steering angle for the i-th future step at time k;

[0094] S34, through the formula: x min ≤x k+i|k ≤x max(4) Establish boundary constraints; where x min x represents the minimum lateral movement distance from the vehicle reference line to the left and right road boundaries. max x represents the maximum lateral movement distance of the vehicle from the reference line to the left and right road boundaries. k+i|k This represents the predicted horizontal coordinate of the position at time k for the i-th future step.

[0095] S35, via formula: r k+1 =P(Z) t+1 |c t+1 If r ≤ ε(5), establish risk constraints; where ε is the risk threshold and r k+1 Z is the risk value calculated at time t+1. t+1 Z ∈{0,1} represents whether a risky vehicle is present in the oncoming lane at the blind spot intersection at time t+1. t+1 =1 indicates that the vehicle appears at time t+1, Z t+1 =0 indicates that no vehicle appears at time t+1; C represents the observation result, c t +1 ∈{0,1} represents the corresponding observation probability value, c t+1 =0 indicates that no vehicle was observed at time t+1, c t+1 =1 indicates that the vehicle was observed at time t+1; P(Z) t+1 |c t+1 ) represents the posterior probability.

[0096] Dynamic Bayesian Networks (DBNs) can handle time-series data with multiple factors. By combining prior and posterior probabilities, they infer the system's state, making them highly suitable for decision-making and planning in complex environments for autonomous vehicles. Therefore, a DBN model is used to address risk assessment in non-line-controlled intersection scenarios. For details on the process and application of DBNs in the autonomous vehicle risk assessment presented in this paper, please refer to [link to relevant documentation]. Figure 5 .

[0097] Scene context information is used to determine prior probabilities, while observation data from the autonomous vehicle is used to update posterior probabilities. The autonomous vehicle samples data at fixed time intervals Δt and updates potential risk predictions in real time. Based on the predictions, the autonomous vehicle can adjust its speed to prevent reckless driving or excessive caution. When the autonomous vehicle is a certain distance from the intersection, and its field of vision gradually increases, a dynamic Bayesian network model can be used to predict whether any potential vehicles are entering the interaction zone from the blind spot, thereby calculating the risk of a collision with the autonomous vehicle. This prediction is achieved by considering the dependencies between nodes and updating the state using new observation data. The specific steps are as follows:

[0098] Step 1: Select Bayesian network nodes

[0099] 1. Vehicle speed node, traffic density node

[0100] Consider vehicle speed and traffic density as the node traffic density in a dynamic Bayesian network, denoted as node B1 and node B2. The scenario of passing through a non-line-controlled intersection involves short-distance traffic segments, where vehicles exhibit similar speed changes within the same road segment. Therefore, potential vehicles in lanes within the perception blind spot can be described using a uniform speed V. According to the Greenshields speed-traffic density model, there is a certain negative correlation between the potential vehicle speed V and the traffic density ρ. The formula for calculating vehicle speed V is:

[0101]

[0102] Where V represents the potential vehicle speed, V f ρ represents the maximum potential speed of vehicles when the traffic density is 0, and ρ represents the traffic density. j This indicates congestion density (the density when traffic flow is so high that movement becomes impossible).

[0103] 2. Traffic accident rate

[0104] For non-line-controlled intersections with blind spots, traffic accidents are prone to occur because traffic participants do not observe the traffic conditions sufficiently. The traffic accident rate D is calculated by using the actual monitoring data recorded by relevant departments at the intersection, and is denoted as node B3.

[0105] 3. Sensor information

[0106] When an autonomous vehicle passes through a non-line-controlled intersection with blind spots, the probability of a collision within those blind spots is also related to the real-time observation data of the autonomous vehicle. This real-time observation data is divided into visible area information and occluded area information, denoted as node B4 and node B5, respectively.

[0107] Step 2: Node network structure modeling

[0108] According to the principles of dynamic Bayesian networks, the coupling relationships between various information factors are described by directed edges between nodes. Using the scene context information variables and sensor information as child nodes of the dynamic Bayesian network, and the risk assessment result per unit time period t as the parent node (denoted as node A1), the specific structure of the established dynamic Bayesian network can be found in [link to network structure]. Figure 6 .

[0109] Step 3: Establish a conditional probability table

[0110] Prior probability refers to the probability that a node is in a certain state before considering any observed data. The sum of the prior probabilities of different possible states of each node is 1. In Bayesian networks, prior probabilities are usually determined based on historical data or expert knowledge. The conditional probability ρ of each prior node, learned from historical data and parameters, is shown in Table 1.

[0111] Table 1

[0112]

[0113] Assuming Z t Z ∈{0,1} represents whether a risky vehicle is present in the oncoming lane at time t of the blind spot intersection. t =1 indicates the presence of a vehicle, Z t =0 indicates no vehicle appears. P(Z) is calculated using scene context information. t =0) and P(Z) t If the result of (=1) is taken as the prior probability, then the result after Δt seconds is updated to P(Z). t+1 ), P(Z t+1 |Z t ) represents the vehicle state (i.e., Z) at a given current time t. t Given the value of Z, the probability that a risky vehicle will appear or not appear at this location at time t+1 reflects the probability of transitioning from the current state at time t to the next state at time t+1. t+1 |Z t The formula for calculating ) is:

[0114]

[0115] Where, k z The state transition coefficients of a dynamic Bayesian network can be assigned based on empirical data; ρ i Let be the prior probability of each node state; n is the number of node states.

[0116] Let C represent the observation result, c t+1 ∈{0,1} represents the corresponding observation probability value, c t+1 =0 indicates that no vehicle was observed, c t+1 =1 indicates that a vehicle was observed. Using Bayes' theorem, the observed data c... t+1 With the prior probability Z after the state transition t +1 Combining these factors, the calculated posterior probability P(c) t+1 |Z t+1 This refers to the probability of risk in the perception blind spot.

[0117] Preferably, in step S35, P(Z) t+1 |ct+1 The formula for calculating ) is:

[0118]

[0119] Wherein, P(c t+1 |Z t+1 P(Z) represents the probability of observing a specific outcome at time t+1 under different conditions (i.e., whether there is a risky vehicle); t+1 =1) is the prior probability that the at-risk vehicle will appear at time t+1, P(c t+1 |Z t+1 =0) is the probability of observing a vehicle at time t+1 if no risky vehicle appears. P(c t+1 |Z t+1 =1) is the probability of observing a vehicle given that a risky vehicle appears at time t+1, and P(C=c) is the prior probability distribution of the observation result C.

[0120] In step S35, P(Z) t+1 |c t+1 The posterior probability is the probability of a node being in a certain state before any observations are considered. The sum of the prior probabilities of each node's different possible states is 1. In Bayesian networks, prior probabilities are usually determined based on historical data or expert knowledge.

[0121] Dynamic Bayesian networks can handle time-series data from multiple factors and infer the state of the system by combining prior and posterior probabilities. They are well-suited for decision-making and planning of autonomous vehicles in complex environments. Therefore, dynamic Bayesian network models are used to address risk assessment in non-line-controlled intersection scenarios.

[0122] Consider vehicle speed and traffic density as the node traffic density in the dynamic Bayesian network, denoted as node B1 and node B2. The scenario of passing through a non-line-controlled intersection involves short-distance traffic segments, where vehicles exhibit similar speed changes within the same road segment. Therefore, potential vehicles in lanes within the perception blind spot can be described using a uniform speed V. The formula for calculating the potential vehicle speed V is:

[0123]

[0124] Among them, V f ρ represents the maximum potential speed of vehicles when the traffic density is 0, and ρ represents the traffic density. j This indicates congestion density (the density at which traffic flow is so high that vehicles cannot move).

[0125] Preferably, the method further includes step S36, establishing the calculation formula for the kinematic constraints as follows: s k+1 =f(s) k ,uk ); where s k+1 Let u be the state variable at time k+1. k Let s be the control quantity at time k. k Let f be the state variable at time k, and let f be the state transition function obtained from the state transition matrix.

[0126] Preferably, by formula: Establish an overall optimization model.

[0127] In step S4 above, the established optimization model is a nonlinear constrained optimization problem. This problem can be solved using an existing nonlinear optimization solver. By continuously optimizing and iteratively updating the motion trajectory, the optimized trajectory can be obtained. The trajectory optimization result is as follows: Figure 4 As shown, the trajectory contains position, speed, and angle information at a series of different time points. When this information is fed into the control system, it enables the autonomous vehicle to safely and efficiently pass through uncontrolled intersections with blind spots.

[0128] The present invention also provides a trajectory planning system for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection, comprising 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 steps of the above-described trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection.

[0129] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the computer.

[0130] The processor referred to can be a Central Processing Unit (CPU), or 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.

[0131] The memory can be an internal storage unit, such as a hard drive or RAM; it can also be an external storage device, such as an external hard drive, a Smart Media Card (SMC), a Secure Digital Card (SD), or a Flash Card. Furthermore, the memory may include both internal and external storage units. The memory is used to store the computer program and other programs and data required by the terminal device. The memory can also be used to temporarily store data that has been output or will be output.

[0132] It should be understood that the above description of specific embodiments of the present invention is only for illustrating the technical approach and features of the present invention, and is intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. However, the present invention is not limited to the specific embodiments described above. All changes or modifications made within the scope of the claims of the present invention should be covered within the protection scope of the present invention.

Claims

1. A trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection, characterized in that, Includes the following steps: S1, through the formula: Establish a kinematic model of the vehicle trajectory; among which, x-coordinate of vehicle position The derivative of x-coordinate of vehicle position The derivative, vehicle heading angle The derivative of Vehicle steering angle The derivative, The derivative of velocity, Let x be the x-coordinate of the vehicle's position. The vertical coordinate represents the vehicle's position. For heading angle, Steering angle, For speed, For acceleration, This is the distance between the front and rear wheels of the vehicle. S2, through the formula: Construct the objective function ; in, To predict the time domain, Its weight matrix, Its weight matrix, For obstacle avoidance function, For visibility functions, To control the input sequence, For the predicted state variables, For reference state variables; S3. Establish kinematic model constraints and optimize the objective function using the constraints. The optimized overall optimization model is obtained; S4. Based on the overall optimization model, the optimized trajectory is solved using a nonlinear optimization solver. The constraints in step S3 include one or more of the following steps: S31, through the formula: (1) Establish speed constraints; where, Let k be the maximum speed the vehicle can reach, and k represent time k. Let k be the predicted velocity value for the i-th future step at time k; S32, through the formula: (2) Establish acceleration constraints; among which, The maximum deceleration that the vehicle can achieve. The maximum acceleration that the vehicle can achieve. Let be the predicted acceleration value for the i-th future step at time k; S33, through the formula: (3) Establish steering angle constraints; where, The minimum steering angle that the vehicle can achieve. The maximum steering angle that the vehicle can achieve. Let be the predicted steering angle for the i-th future step at time k; S34, via formula: Establish boundary constraints; among them, This represents the minimum lateral movement distance from the vehicle reference line to the left and right road boundaries. This represents the maximum lateral movement distance of the vehicle from the reference line to the left and right road boundaries. This represents the predicted horizontal coordinate of the position at time k for the i-th future step. S35, via formula: Establish risk constraints; among them As a risk threshold, The risk value at time t+1. This indicates whether a risky vehicle is detected in the oncoming lane at the blind spot intersection at time t+1. This indicates that the vehicle appeared at time t+1. This indicates that no vehicle appears at time t+1; Indicates the observation results, This represents the corresponding observation probability value. This indicates that no vehicle was observed at time t+1. This indicates that the vehicle was observed at time t+1; Let be the posterior probability at time t+1.

2. The trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection as described in claim 1, characterized in that, Through the formula: Establish an overall optimization model.

3. The trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection as described in claim 1, characterized in that, In step S2 The calculation formula is: ， Where i represents time i. These are the weighting coefficients. It represents the position coordinates of the visible obstacle relative to the autonomous vehicle's body coordinate system at time i. It is the centroid coordinate of the autonomous vehicle at time i. To prevent positive numbers with a denominator of 0, Let be the speed of the vehicle at time i.

4. The trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection as described in claim 3, characterized in that, In step S2 The calculation formula is: ， in, These are the weighting coefficients. for The angle of occlusion at any moment, as well as For vehicles The distance between the position coordinates at a given moment and the blind spot coordinates at the slow point.

5. The trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection as described in claim 4, characterized in that, The blind spot is the vertex of the perception blind zone caused by the obstruction of roadside buildings, and the coordinates of the blind spot are the geographical coordinates provided by the map positioning system of the vertex.

6. The trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection as described in claim 1, characterized in that, In step S35 The calculation formula is: in, Let be the prior probability of the risky vehicle appearing at time t+1. Let be the probability of observing a vehicle at time t+1 if no risky vehicle is present. Let be the probability of observing a vehicle given that a risky vehicle is present at time t+1. Let C be the prior probability distribution of the observation result C.

7. The trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection as described in claim 1, characterized in that, The kinematic model mentioned in step S1 is a combination of the bicycle kinematic model and the MPC model.

8. The trajectory planning method for a vehicle to perceive blind spots when turning at a non-signal-controlled intersection as described in claim 1, characterized in that, It also includes step S36, which establishes the calculation formula for the kinematic constraints: ;in, for State quantity at any given time. for The amount of control at any given moment State quantity at any given time. This is the state transition function obtained from the state transition matrix.

9. A trajectory planning system for perceiving blind spots when a vehicle turns at a non-signal-controlled intersection, 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 steps of the trajectory planning method for perceiving blind spots when a vehicle turns at a non-signalized intersection as described in any one of claims 1 to 8.

Images