A method for intelligent vehicle trajectory planning that considers dynamic pedestrian interaction

Abstract

This invention discloses an intelligent vehicle trajectory planning method considering dynamic pedestrian interaction, comprising: Step 1, real-time acquisition of relevant road and pedestrian information; Step 2, establishment of road entrance / exit potential functions and road boundary potential functions; Step 3, establishment of pedestrian dynamic potential functions based on fuzzy rules, considering pedestrian hazard level and mobility level; Step 4, comprehensive superposition of road entrance / exit potential functions, road boundary potential functions, and pedestrian dynamic potential functions to obtain a resultant potential field considering pedestrian trajectory, and planning the vehicle trajectory based on the resultant potential field using a gradient descent method. This invention not only considers road change trends and road boundary ranges, and details the road risk field by establishing road entrance / exit potential functions and road boundary potential functions, but also incorporates pedestrian interaction dynamics, enabling a more accurate depiction of traffic scenarios involving mixed pedestrian and vehicle traffic.

Description

Technical Field

[0001] This invention belongs to the field of intelligent connected vehicle technology, and in particular relates to an intelligent vehicle trajectory planning method that takes into account dynamic pedestrian interaction. Background Technology

[0002] Intelligent vehicles are ordinary vehicles with the addition of advanced sensors (such as radar, cameras, etc.), controllers, actuators, and other devices. Through onboard environmental perception systems and information terminals, they can exchange information with people, vehicles, roads, etc., enabling the vehicle to have intelligent environmental perception capabilities, automatically analyze the safety and danger status of the vehicle, and reach the destination according to the driver's wishes, ultimately achieving the goal of replacing human operation.

[0003] In recent years, intelligent vehicles have become a research hotspot in the field of vehicle engineering worldwide and a new driving force for the growth of the automotive industry. Many developed countries have incorporated them into their key intelligent transportation systems.

[0004] The goal of intelligent vehicle trajectory planning is to obtain safe, efficient, and collision-free paths, ensuring that the vehicle accurately drives from the planned starting point to its destination. In mixed pedestrian and vehicle traffic environments, traditional artificial potential field methods treat pedestrians merely as simple static or dynamic obstacles. This not only fails to consider the trajectory uncertainty caused by pedestrian interactions but also easily overlooks the intentions and changes in movement trajectories under interactive relationships. Consequently, in real traffic scenarios, autonomous vehicles either become too conservative, resulting in reduced efficiency, or become too aggressive, leading to collision risks, making it difficult to achieve good trajectory planning results. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the purpose of this invention is to provide an intelligent vehicle trajectory planning method that takes into account dynamic pedestrian interaction, so as to improve the accuracy of intelligent vehicle trajectory planning.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] The present invention provides an intelligent vehicle trajectory planning method considering dynamic pedestrian interaction, comprising the following steps:

[0008] Step 1: Collect relevant road and pedestrian information in real time;

[0009] Step 2: Establish the potential functions for the road entrances and exits and the potential functions for the road boundaries;

[0010] Step 3: Considering pedestrian hazard level and mobility level, establish a pedestrian dynamic potential function based on fuzzy rules;

[0011] Step 4: Combine and superimpose the potential functions of the road entrance / exit, the potential function of the road boundary, and the dynamic potential function of the pedestrian to obtain the resultant potential field considering the pedestrian trajectory. Based on the resultant potential field, plan the vehicle trajectory according to the gradient descent method.

[0012] Preferably, the information collected in step 1 specifically includes: changes in the road ahead, road boundary range, pedestrian movement trajectory, and pedestrian category.

[0013] Preferably, the road entrance / exit potential function and the road boundary potential function in step 2 are respectively:

[0014] U rep-re =α(y-Y0) 2

[0015]

[0016] Among them, U rep-re It is the potential function of the road entrance and exit, U att-rb Y0 is the potential function of the road boundary; Y0 is the road length, L is the road width, and x is the road width. l x is the lateral position of the road centerline, x and y are the lateral and longitudinal positions of the road, α is the potential function coefficient of the road entrance / exit, β is the potential function coefficient of the road plane, and v is the vehicle speed. lim It is the maximum speed limit on the road.

[0017] Preferably, in step 3, a circular obstacle potential field is used for pedestrians, and the effective range of the pedestrian obstacle in the road is:

[0018] (XX (t) ) 2 +(YY (t) ) 2 =γ

[0019] Where X and Y are the horizontal and vertical position coordinates, respectively, X (t) and Y (t) These are the x and y coordinates of the pedestrian's centroid, respectively; γ is the normalization coefficient.

[0020] Preferably, step 3, establishing the pedestrian dynamic potential function based on fuzzy rules, specifically includes:

[0021] Step 31: Based on the pedestrian category results collected by the vehicle, the mobility of different types of pedestrians is divided into four mobility levels: F4, F3, F2, and F1. Among them, pedestrian types include: children, teenagers, young and middle-aged adults, and the elderly; the corresponding mobility levels are: F4, F3, F2, and F1, respectively.

[0022] Step 32: Calculate the collision time TTC:

[0023]

[0024] In the formula, t TTC It is the collision avoidance time; ΔS is the distance from the vehicle's center of gravity to the pedestrian's center of gravity; ΔS' is the longitudinal relative distance between the vehicle and the pedestrian; d safety It is the safe radius of the space occupied by pedestrians; d vehicle It is the distance from the vehicle's center of mass to the front profile of the vehicle; Δv is the resultant velocity of the vehicle and the pedestrian in the direction of the vehicle's travel. It is the vehicle speed; It is the speed of a pedestrian along the direction of vehicle travel;

[0025] Step 33: Based on the collision time calculation results, classify different collision times into four danger levels: R4, R3, R2, and R1; where t TTC When the value is between 0 and 1.5, it corresponds to R4; t TTC When the value is between 1.6 and 2.7, it corresponds to R3; t TTC When the value is between 2.8 and 3.6, it corresponds to R²; TTC When it is greater than 3.6, it corresponds to R1;

[0026] Step 34: Use triangular membership functions to describe the two fuzzy variables, mobility level and hazard level, construct fuzzy surfaces, and establish the relationship between the four comprehensive trajectory hazard levels (I, II, III, IV, V) and mobility level and hazard level;

[0027] Step 35: Assign different normalization coefficient γ values ​​to different comprehensive hazard levels, and use them as the safety radius under interactive conditions; where comprehensive trajectory hazard level I corresponds to γ ​​= 0.3, comprehensive trajectory hazard level II corresponds to γ ​​= 0.5, comprehensive trajectory hazard level III corresponds to γ ​​= 0.8, comprehensive trajectory hazard level IV corresponds to γ ​​= 0.9, and comprehensive trajectory hazard level V corresponds to γ ​​= 1.3.

[0028] Step 36: Integrate the pedestrian trajectory into the three-dimensional potential field space to obtain the pedestrian dynamic potential function:

[0029]

[0030] In the formula, U rep-pedestrian It is the pedestrian dynamic obstacle potential function, U p It is the balance quantity of the pedestrian potential function, and ρ is the horizontal and vertical correlation coefficient.

[0031] Preferably, in step 32, the safe radius of the space occupied by pedestrians is related to the interaction between pedestrians and the interaction between pedestrians and vehicles. The interaction force between pedestrians is calculated using a social force model, and the interaction force between pedestrians and vehicles is calculated using a virtual contour, as follows:

[0032]

[0033]

[0034] In the formula, It is the force of interaction between pedestrians. It is the interaction force between pedestrians and vehicles. These are the repulsive force, collision force, and navigational force exerted by pedestrian j on pedestrian i, respectively. veh and b veh It is the characteristic parameter, λ veh These are the anisotropic characteristic parameters of vehicle forces. It is the direction of the interaction force between pedestrians and vehicles. It is the minimum distance from the pedestrian to the front outline of the vehicle, f exp It is an exponentially decaying function. It is a sinusoidal anisotropic function. Indicates the interaction perspective.

[0035] The beneficial effects of this invention are:

[0036] 1. The artificial potential field trajectory planning method proposed in this invention, which integrates pedestrian trajectory information, not only considers the road change trend and road boundary range, but also details the road risk field by establishing the potential function of the road entrance and exit and the potential function of the road boundary. Furthermore, it incorporates the interactive dynamics of pedestrians, which can more accurately depict the traffic scenario of mixed pedestrian and vehicle traffic.

[0037] 2. The pedestrian dynamic potential function of the present invention takes into account the pedestrian danger level and mobility level, and performs trajectory planning by integrating the combined potential field of the road and pedestrian potential functions, which can achieve better trajectory smoothness and improve pedestrian protection. Attached Figure Description

[0038] Figure 1 This is a flowchart of the method of the present invention.

[0039] Figure 2 This is a schematic diagram of the blurred surface.

[0040] Figure 3 This is a schematic diagram of the resultant potential function. Detailed Implementation

[0041] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to embodiments and accompanying drawings. The content mentioned in the embodiments is not intended to limit the present invention.

[0042] Reference Figures 1-3 As shown, the present invention provides an intelligent vehicle trajectory planning method considering dynamic pedestrian interaction, comprising the following steps:

[0043] Step 1: Collect relevant road and pedestrian information in real time;

[0044] The information collected specifically includes: changes ahead of the road, road boundary range, pedestrian movement trajectory, and pedestrian category.

[0045] Step 2: Establish the potential functions for the road entrances and exits and the potential functions for the road boundaries;

[0046] The potential functions at the road entrances / exits and the potential functions at the road boundaries are as follows:

[0047] U rep-re =α(y-Y0) 2

[0048]

[0049] Among them, U rep-re It is the potential function of the road entrance and exit, U att-rb Y0 is the potential function of the road boundary; Y0 is the road length, L is the road width, and x is the road width. l x is the lateral position of the road centerline, x and y are the lateral and longitudinal positions of the road, α is the potential function coefficient of the road entrance / exit, β is the potential function coefficient of the road plane, and v is the vehicle speed. lim It is the maximum speed limit on the road.

[0050] Step 3: Considering pedestrian hazard level and mobility level, establish a pedestrian dynamic potential function based on fuzzy rules;

[0051] In step 3, a circular obstacle potential field is used for pedestrians, and the effective range of the pedestrian obstacle in the road is as follows:

[0052] (XX (t) ) 2 +(YY (t) ) 2 =γ

[0053] In the formula, X and Y are the horizontal and vertical position coordinates, respectively. (t) and Y (t) These are the x and y coordinates of the pedestrian's centroid, respectively; γ is the normalization coefficient.

[0054] Furthermore, step 3, establishing the pedestrian dynamic potential function based on fuzzy rules, specifically includes:

[0055] Step 31: Based on the pedestrian category results collected by the vehicle, the mobility of different types of pedestrians is divided into four mobility levels: F4, F3, F2, and F1. Pedestrian types include: children, teenagers, young and middle-aged adults, and the elderly; the corresponding mobility levels are F4, F3, F2, and F1, respectively; see Table 1.

[0056] Table 1

[0057]

[0058] Step 32: Calculate the collision time TTC:

[0059]

[0060] In the formula, t TTC It is the collision avoidance time; ΔS is the distance from the vehicle's center of gravity to the pedestrian's center of gravity; ΔS' is the longitudinal relative distance between the vehicle and the pedestrian; d safety It is the safe radius of the space occupied by pedestrians; d vehicle It is the distance from the vehicle's center of gravity to its front profile; Δv is the resultant velocity of the vehicle and the pedestrian in the direction of the vehicle's travel. It is the vehicle speed; It is the speed of a pedestrian along the direction of vehicle travel;

[0061] Step 33: Based on the collision time calculation results, classify different collision times into four danger levels: R4, R3, R2, and R1; where t TTC When the value is between 0 and 1.5, it corresponds to R4; t TTC When the value is between 1.6 and 2.7, it corresponds to R3; t TTC When the value is between 2.8 and 3.6, it corresponds to R²; TTC When the value is greater than 3.6, it corresponds to R1; as shown in Table 2:

[0062] Table 2

[0063]

[0064] Step 34: Use triangular membership functions to describe the two fuzzy variables, mobility level and hazard level, construct fuzzy surfaces, and establish the relationship between the four comprehensive trajectory hazard levels (I, II, III, IV, V) and the mobility level and hazard level, as shown in Table 3:

[0065] Table 3

[0066]

[0067] Step 35: Assign different normalization coefficient γ values ​​to different comprehensive hazard levels, and use them as the safety radius under interactive conditions; where comprehensive trajectory hazard level I corresponds to γ ​​= 0.3, comprehensive trajectory hazard level II corresponds to γ ​​= 0.5, comprehensive trajectory hazard level III corresponds to γ ​​= 0.8, comprehensive trajectory hazard level IV corresponds to γ ​​= 0.9, and comprehensive trajectory hazard level V corresponds to γ ​​= 1.3; see Table 4:

[0068] Table 4

[0069]

[0070]

[0071] Step 36: Integrate the pedestrian trajectory into the three-dimensional potential field space to obtain the pedestrian dynamic potential function:

[0072]

[0073] In the formula, U rep-pedestrian It is the pedestrian dynamic obstacle potential function, U p It is the balance quantity of the pedestrian potential function, and ρ is the horizontal and vertical correlation coefficient.

[0074] In step 32, the safe radius of the space occupied by pedestrians is related to the interaction between pedestrians and the interaction between pedestrians and vehicles. The interaction force between pedestrians is calculated using a social force model, and the interaction force between pedestrians and vehicles is calculated using a virtual contour, as follows:

[0075]

[0076]

[0077] In the formula, It is the force of interaction between pedestrians. It is the interaction force between pedestrians and vehicles. These are the repulsive force, collision force, and navigational force exerted by pedestrian j on pedestrian i, respectively. veh and b veh It is the characteristic parameter, λ veh These are the anisotropic characteristic parameters of vehicle forces. It is the direction of the interaction force between pedestrians and vehicles. It is the minimum distance from the pedestrian to the front outline of the vehicle, f exp It is an exponentially decaying function. It is a sinusoidal anisotropic function. Indicates the interaction perspective.

[0078] Step 4: Combine and superimpose the potential functions of the road entrance / exit, the potential function of the road boundary, and the dynamic potential function of the pedestrian to obtain the resultant potential field considering the pedestrian trajectory. Based on the resultant potential field, plan the vehicle trajectory according to the gradient descent method.

[0079] This invention has many specific applications. The above description is only a preferred embodiment of this invention. It should be noted that for those skilled in the art, several improvements can be made without departing from the principle of this invention, and these improvements should also be considered within the scope of protection of this invention.

Claims

1. A method for intelligent vehicle trajectory planning that considers dynamic pedestrian interaction, characterized in that, The steps are as follows: Step 1: Collect relevant road and pedestrian information in real time; Step 2: Establish the potential functions for the road entrances and exits and the potential functions for the road boundaries; Step 3: Considering pedestrian hazard level and mobility level, establish a pedestrian dynamic potential function based on fuzzy rules; Step 4: Combine and superimpose the potential functions of road entrances and exits, road boundaries, and pedestrian dynamic potential functions to obtain the resultant potential field considering pedestrian trajectories. Based on the resultant potential field, plan the vehicle trajectory according to the gradient descent method. The information collected in step 1 specifically includes: changes in the road ahead, road boundary range, pedestrian movement trajectory, and pedestrian category; The potential functions of the road entrance / exit and the potential function of the road boundary in step 2 are as follows: U rep-re =α(y-Y0) 2 In the formula, U rep-re It is the potential function of the road entrance and exit, U att-rb Y0 is the potential function of the road boundary; Y0 is the road length, L is the road width, and x is the road width. l x is the lateral position of the road centerline, x and y are the lateral and longitudinal positions of the road, α is the potential function coefficient of the road entrance / exit, β is the potential function coefficient of the road plane, and v is the vehicle speed. lim It is the maximum speed limit on the road; In step 3, a circular obstacle potential field is used for pedestrians. The effective range of the pedestrian obstacle in the road is as follows: (XX (t) ) 2 +(YY (t) ) 2 =γ In the formula, X and Y are the horizontal and vertical position coordinates, respectively. (t) and Y (t) These are the x and y coordinates of the pedestrian's centroid, respectively; γ is the normalization coefficient.

2. The intelligent vehicle trajectory planning method considering dynamic pedestrian interaction according to claim 1, characterized in that, Step 3, establishing the pedestrian dynamic potential function based on fuzzy rules, specifically includes: Step 31: Based on the pedestrian category results collected by the vehicle, the mobility of different types of pedestrians is divided into four mobility levels: F4, F3, F2, and F1. Among them, pedestrian types include: children, teenagers, young and middle-aged adults, and the elderly; the corresponding mobility levels are: F4, F3, F2, and F1, respectively. Step 32: Calculate the collision time TTC: In the formula, t TTC It is the collision avoidance time; ΔS is the distance from the vehicle's center of gravity to the pedestrian's center of gravity; ΔS' is the longitudinal relative distance between the vehicle and the pedestrian; d safety It is the safe radius of the space occupied by pedestrians; d vehicle It is the distance from the vehicle's center of mass to the front profile of the vehicle; Δv is the resultant velocity of the vehicle and the pedestrian in the direction of the vehicle's travel. It is the vehicle speed; It is the speed of a pedestrian along the direction of vehicle travel; Step 33: Based on the collision time calculation results, classify different collision times into four danger levels: R4, R3, R2, and R1; where t TTC When the value is between 0 and 1.5, it corresponds to R4; t TTC When the value is between 1.6 and 2.7, it corresponds to R3; t TTC When the value is between 2.8 and 3.6, it corresponds to R²; TTC When it is greater than 3.6, it corresponds to R1; Step 34: Use triangular membership functions to describe the two fuzzy variables, mobility level and hazard level, construct fuzzy surfaces, and establish the relationship between the hazard level of the four comprehensive trajectories and the mobility level and hazard level; Step 35: Assign different normalization coefficient γ values ​​to different comprehensive hazard levels, and use them as the safety radius under interactive conditions; where comprehensive trajectory hazard level I corresponds to γ ​​= 0.3, comprehensive trajectory hazard level II corresponds to γ ​​= 0.5, comprehensive trajectory hazard level III corresponds to γ ​​= 0.8, comprehensive trajectory hazard level IV corresponds to γ ​​= 0.9, and comprehensive trajectory hazard level V corresponds to γ ​​= 1.

3. Step 36: Integrate the pedestrian trajectory into the three-dimensional potential field space to obtain the pedestrian dynamic potential function: In the formula, U rep-pedestrian It is the pedestrian dynamic obstacle potential function, U p It is the balance quantity of the pedestrian potential function, and ρ is the horizontal and vertical correlation coefficient.

3. The intelligent vehicle trajectory planning method considering dynamic pedestrian interaction according to claim 2, characterized in that, In step 32, the safe radius of the space occupied by pedestrians is related to the interaction between pedestrians and the interaction between pedestrians and vehicles. The interaction force between pedestrians is calculated using a social force model, and the interaction force between pedestrians and vehicles is calculated using a virtual contour, as follows: In the formula, It is the force of interaction between pedestrians. It is the interaction force between pedestrians and vehicles. These are the repulsive force, collision force, and navigational force exerted by pedestrian j on pedestrian i, respectively. veh and b veh It is the characteristic parameter, λ veh These are the anisotropic characteristic parameters of vehicle forces. It is the direction of the interaction force between pedestrians and vehicles. It is the minimum distance from the pedestrian to the front outline of the vehicle, f exp It is an exponentially decaying function. It is a sinusoidal anisotropic function. Indicates the perspective of interaction.

Images