A 3D path planning method for intelligent driving vehicles to prevent rollover

Abstract

A three-dimensional path planning method for an intelligent driving vehicle to prevent rollover, comprising: First, constructing a vehicle roll estimation model containing three-dimensional terrain information and vehicle heading angle, and designing a roll cost function; Second, designing a path length cost function that combines two-dimensional and three-dimensional paths; Third, designing a fuzzy adaptive weight optimizer to optimize the weights of the vehicle roll cost function and the path length cost function; Fourth, calculating the sum of the weighted vehicle roll cost function and the weighted path length cost function, using the gradient descent method to find its negative gradient, and determining the next path direction according to the negative gradient direction; Fifth, determining whether the planned path reaches the destination, if not, repeating the above steps. The method proposed by the present invention effectively helps the intelligent driving vehicle to plan a reasonable path that comprehensively considers the rollover risk and the path length in the three-dimensional terrain, providing a safe and economical path planning solution for the intelligent driving vehicle in the complex three-dimensional terrain environment.

Description

Technical Field

[0001] The present invention relates to the field of vehicle path planning, and particularly to a three-dimensional path planning method for an intelligent driving vehicle that prevents rollover. Background Art

[0002] With the upgrading of the automotive industry, the development and application of intelligent vehicle technology have penetrated into people's production and life, becoming an increasingly prominent trend. The wide application of intelligent vehicle technology is considered to greatly improve people's driving experience and effectively reduce the incidence of traffic accidents. Therefore, it has received great attention from major countries around the world, and vehicle intelligence has been taken as the main research and investment direction for the development of the automotive industry. In recent years, the development of vehicle intelligent technology in China has been rapid. However, vehicle intelligent technology involves a wide range and complex application scenarios, and there are still many technical problems to be solved. Therefore, breaking through the bottleneck of vehicle intelligent technology and improving the intelligent vehicle technology system have far-reaching social and economic benefits for improving the traffic situation in China.

[0003] Path planning, as an essential core technology for intelligent vehicles, has always received extensive attention from researchers. However, with the in-depth research, the application scenario of intelligent vehicle path planning has expanded from two-dimensional structured roads to three-dimensional roadless roads. The change in the application scenario has brought new problems to the implementation of path planning and also put forward new requirements for path planning. Therefore, in this situation, there is an urgent need for a path planning method that meets the requirements of a three-dimensional environment. Summary of the Invention

[0004] The present invention provides a three-dimensional path planning method for an intelligent driving vehicle that prevents rollover to solve the technical problem that existing three-dimensional path planning technologies cannot effectively prevent vehicle rollover.

[0005] To achieve the above object, the technical solution of the present invention is realized as follows:

[0006] The present invention provides a three-dimensional path planning method for an intelligent driving vehicle that prevents rollover, including the following steps:

[0007] Step S10: Construct a vehicle roll estimation model that includes three-dimensional terrain information and vehicle heading angle, and design a roll cost function;

[0008] Step S20: Design a path length cost function that combines two-dimensional and three-dimensional paths;

[0009] Step S30: Design a fuzzy adaptive weight optimizer to optimize the weights of the vehicle roll cost function and the path length cost function;

[0010] Step S40: Calculate the sum of the weighted vehicle roll cost function and the weighted path length cost function, and then use the gradient descent method to find its negative gradient. Determine the next path direction according to the negative gradient direction;

[0011] Step S50: Determine whether the planned path reaches the destination. If not, repeat the above steps until the destination is reached.

[0012] Furthermore, in the vehicle roll estimation model including three-dimensional terrain information and vehicle heading angle in step S10, the three-dimensional terrain information can be expressed as a function h(x, y), where (x, y) are the abscissa and ordinate in the three-dimensional terrain, and h(x, y) is the elevation at the (x, y) coordinate. The gradient corresponding to h(x, y) can be expressed as:

[0013]

[0014]

[0015] In the formula, γ(x, y) represents the angle between the gradient of h(x, y) and the x-axis, i represents the unit vector in the x direction, j represents the unit vector in the y direction, and arctan2 is an extended form of the arctan function, and the range of its return value is (-π, π]. Since the vehicle axle is perpendicular to the vehicle heading angle, the directional derivative of h(x, y) along the vehicle axle can be expressed as:

[0016]

[0017] In the formula, represents the 2-norm of represents the heading angle of the vehicle centroid at (x, y). The coordinates corresponding to the left and right tires of the vehicle are (x 1 , y 1 ) and (x 2 , y 2 ), respectively, and their calculation expressions are:

[0018]

[0019]

[0020]

[0021]

[0022] In the formula, l xy is the projected length of the distance l between the left and right tires in the x - y plane when the vehicle is at (x, y), and its calculation expression is:

[0023]

[0024] When the vehicle heading angle is and the vehicle coordinates are (x, y), the roll angle of the vehicle caused by the three-dimensional terrain can be expressed as:

[0025]

[0026] Taking the vehicle roll angle expression as the vehicle roll estimation model, the vehicle roll cost function in the three-dimensional terrain is designed as:

[0027]

[0028] In the formula, ξ d (x, y) is the distance attenuation coefficient, which can make the roll cost function of the vehicle match the path length cost function when the vehicle is at different positions. ξ d (x, y) calculation expression is designed as:

[0029]

[0030] In the formula, c 1 is the intensity coefficient, d s is the x-y plane distance from the starting point to the destination, d(x, y) is the x-y plane distance from the point (x, y) to the destination, d s and the calculation expressions of d(x, y) are:

[0031]

[0032]

[0033] In the formula, (x q , y q ) is the plane coordinates of the starting point, (x m , y m ) is the plane coordinates of the starting point.

[0034] Furthermore, the path length cost function for integrating the two-dimensional and three-dimensional paths in step S20 is designed as:

[0035] C l (x, y) = c 2 C 2l (x, y) + c 3 C 3l (x, y)

[0036] In the formula, c 2 is the intensity coefficient of C 2l (x, y), c 3 is the intensity coefficient of C 3l (x, y), C2l (x, y) represents the x - y plane distance cost function from (x, y) to the destination, and its calculation expression is:

[0037]

[0038] C 3l (x, y) is used to suppress the three - dimensional terrain undulation, and its calculation expression is:

[0039]

[0040] In the formula, h d (x, y) is the height parameter related to the position. At the starting point, its value is equal to the actual elevation h(x q , y q ) of the starting point; at the destination, its value is equal to the actual elevation h(x m , y m ) of the destination. At any position on the map, its calculation expression is designed as:

[0041]

[0042] Furthermore, the design method of the fuzzy adaptive weight optimizer in step S30 includes the following steps:

[0043] Step 31: Define the input variables as ξ 1 and ξ 2 , and the output variable is which represents the fuzzy weight of the roll cost function C ρ . The calculation expression of ξ 1 is:

[0044]

[0045] In the formula, δ(x, y) is the direction angle from the coordinate (x, y) to the destination, and its calculation expression is:

[0046]

[0047] The calculation expression of ξ 2 is:

[0048]

[0049] Step 32: Normalize ξ 1 and ξ 2 , and set its range to [-5, 5].

[0050]

[0051]

[0052] In the formula, Max(ξ 1 ) and Min(ξ 1 ) represent the maximum and minimum values of ξ 1 ; Max(ξ 2 ) and Min(ξ 2 ) represent the maximum and minimum values of ξ 2 .

[0053] Step 33: Establish the first membership function, the second membership function, and the third membership function, which are respectively used to transform and into corresponding fuzzy variables;

[0054] Step 34: Design fuzzy inference rules to obtain the inference rule table;

[0055] ξ 1 represents the difference between the vehicle heading angle and the direction angle required to reach the destination. The larger ξ 1 is, the more serious the vehicle yaw is, and it is necessary to increase the weight of the path length cost function to suppress vehicle yaw. On the contrary, if the vehicle yaw is not obvious, the weight of the path length cost function can be reduced. ξ 2 represents the difference between the vehicle heading angle and the terrain gradient direction. The smaller ξ 2 is, the more the vehicle tends to move along the terrain contour line, which is more conducive to highlighting the amplification effect of the terrain on the roll angle. It is necessary to increase the weight of the roll cost function to suppress the increase of the roll angle. On the contrary, if the vehicle tends to move along the terrain gradient direction, which is beneficial to suppressing the increase of the roll angle, the weight of the roll cost function can be reduced. Therefore, for the value, there are the following inference rules:

[0056] When is small and is large, should take the median value;

[0057] When is small and is small, should take the large value;

[0058] When is large and is small, should take the median value;

[0059] When is large and is large, should take the small value;

[0060] Step 35: Defuzzify the fuzzy variables by using the centroid method, and the defuzzification method may also include but is not limited to the maximum membership degree method and the weighted average method; of the fuzzy variables into clear variables, and the defuzzification method may also include but is not limited to the maximum membership degree method and the weighted average method;

[0061] Step 36: Normalize the roll cost function weight to the range of [0, 1], and its normalization formula is as follows;

[0062]

[0063] In the formula, is the minimum value of, represents the maximum value of, represents the roll cost function weight after normalization; in order to reduce the calculation, the weight of the path length cost function can be simply calculated as:

[0064]

[0065] Furthermore, the expression of the sum of the weighted vehicle roll cost function and the weighted path length cost function in step S40 is:

[0066]

[0067] Taking the negative gradient of the above formula gives:

[0068]

[0069] The modulus of the gradient is:

[0070]

[0071] The next vehicle position is:

[0072]

[0073]

[0074] In the formula, x next and y next represent the position of the vehicle planned for the next step, v represents the speed of the vehicle, and t is the sampling interval time.

[0075] Furthermore, the method for judging whether the planned path reaches the destination in step S50 is:

[0076]

[0077] In the formula, d pFor the determination of distance, if the above equation holds, it is determined that the vehicle has reached the end point. d p The expression of can be set as:

[0078] d p = vt

[0079] Advantages of the present invention:

[0080] 1. Based on terrain information, the present invention considers the vehicle rollover problem at the path planning level, not only avoiding complex vehicle dynamics, greatly reducing the computational complexity of the method, but also effectively suppressing the vehicle roll angle generated by terrain undulations when planning the path, thereby reducing the vehicle rollover risk.

[0081] 2. In addition, the present invention also weighs the path length under three-dimensional terrain, planning a safe path with anti-rollover and short path length for intelligent vehicles, which has important theoretical and practical significance. Brief Description of the Drawings

[0082] Figure 1 is the flowchart of the method of the present invention;

[0083] Figure 2 is the vehicle roll angle generated by the terrain and its calculation schematic diagram. Detailed Description of the Preferred Embodiment

[0084] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0085] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0086] Referring to Figure 1 , the embodiment of the present application provides a three-dimensional path planning method for anti-rollover, including the following steps:

[0087] Step S10: Construct a vehicle roll estimation model including three-dimensional terrain information and vehicle heading angle, and design a roll cost function;

[0088] Step S20: Design a path length cost function that combines two-dimensional and three-dimensional paths;

[0089] Step S30: Design a fuzzy adaptive weight optimizer to optimize the weights of the vehicle roll cost function and the path length cost function;

[0090] Step S40: Calculate the sum of the weighted vehicle roll cost function and the weighted path length cost function, and then use the gradient descent method to calculate its negative gradient. Determine the next path direction according to the negative gradient direction;

[0091] Step S50: Determine whether the planned path reaches the destination. If it does not reach, repeat the above steps until reaching the destination.

[0092] The method proposed by the present invention effectively helps an intelligent driving vehicle to plan a reasonable path that comprehensively considers the roll risk and the path length in a three-dimensional terrain, providing a safe and economical path planning solution for the intelligent driving vehicle in a complex three-dimensional terrain environment.

[0093] The vehicle roll estimation model including three-dimensional terrain information and vehicle heading angle in step S10 is characterized in that the three-dimensional terrain information in step S10 can be expressed as a function h(x, y), where (x, y) are the abscissa and ordinate in the three-dimensional terrain, and h(x, y) is the elevation at the (x, y) coordinate. The gradient corresponding to h(x, y) can be expressed as:

[0094]

[0095]

[0096] In the formula, γ(x, y) represents the angle between the gradient of h(x, y) and the x-axis, i represents the unit vector in the x direction, j represents the unit vector in the y direction, and arctan2 is the extended form of the arctan function, and its return value range is (-π, π]. Since the vehicle axle is perpendicular to the vehicle heading angle, the directional derivative of h(x, y) along the vehicle axle can be expressed as:

[0097]

[0098] In the formula, represents the 2-norm of represents the heading angle of the vehicle centroid at (x, y). The coordinates corresponding to the left and right tires of the vehicle are (x 1 , y 1 ) and (x 2 , y 2 ), respectively, and their calculation expressions are:

[0099]

[0100]

[0101]

[0102]

[0103] In the formula, l xy is the projected length of the distance l between the left and right tires in the x-y plane when the vehicle is at (x, y), and its calculation expression is:

[0104]

[0105] When the vehicle heading angle is and the vehicle coordinates are (x, y), the roll angle of the vehicle caused by the three-dimensional terrain can be expressed as:

[0106]

[0107] Taking the roll angle expression of the vehicle as the vehicle roll estimation model, the vehicle roll cost function in the three-dimensional terrain is designed as:

[0108]

[0109] In the formula, ξ d (x, y) is the distance attenuation coefficient, which can make the roll cost function of the vehicle match the path length cost function when the vehicle is at different positions. The calculation expression of ξ d (x, y) is designed as:

[0110]

[0111] In the formula, c 1 is the intensity coefficient, d s is the x-y plane distance from the starting point to the destination, d(x, y) is the x-y plane distance from the point (x, y) to the destination, and the calculation expressions of d s and d(x, y) are:

[0112]

[0113]

[0114] In the formula, (x q , y q ) are the plane coordinates of the starting point, and (x m , y m ) are the plane coordinates of the starting point.

[0115] The path length cost function of the integrated two-dimensional and three-dimensional path in the step S20 is characterized in that the path length cost function of the integrated two-dimensional and three-dimensional path is designed as:

[0116] C l (x, y) = c 2 C 2l (x, y) + c 3 C3l (x, y)

[0117] Wherein, c 2 is the intensity coefficient of C 2l (x, y), and c 3 is the intensity coefficient of C 3l (x, y). C 2l (x, y) represents the x - y plane distance cost function from (x, y) to the destination, and its calculation expression is:

[0118]

[0119] C 3l (x, y) is used to suppress the three - dimensional terrain undulation, and its calculation expression is:

[0120]

[0121] Wherein, h d (x, y) is the height parameter related to the position. At the starting point, its value is equal to the actual elevation h(x q , y q ) of the starting point; at the destination, its value is equal to the actual elevation h(x m , y m ) of the destination. At any position on the map, its calculation expression is designed as:

[0122]

[0123] The fuzzy adaptive weight optimizer in the step S30 is characterized in that the design method of the fuzzy adaptive weight optimizer includes the following steps:

[0124] Step 31: Define the input variables as ξ 1 and ξ 2 , and the output variable is which represents the fuzzy weight of the roll cost function C ρ 1 2 . The calculation expression of ξ

[0125]

[0126] is: Wherein, δ(x, y) is the direction angle from the coordinate (x, y) to the destination, and its calculation expression is:

[0127]

[0128] The calculation expression of ξ 2 is:

[0129]

[0130] Step 32: For ξ 1 and ξ 2 perform normalization, and set its range to [-5, 5].

[0131]

[0132]

[0133] In the formula, Max(ξ 1 ) and Min(ξ 1 ) represent the maximum and minimum values of ξ 1 ; Max(ξ 2 ) and Min(ξ 2 ) represent the maximum and minimum values of ξ 2 .

[0134] Step 33: Establish the first membership function, the second membership function and the third membership function, which are respectively used to convert and into corresponding fuzzy variables;

[0135] Step 34: Design fuzzy inference rules to obtain the inference rule table of ;

[0136] ξ 1 represents the difference between the vehicle heading angle and the direction angle required to reach the destination. The larger ξ 1 is, the more serious the vehicle yaw is, and it is necessary to increase the weight of the path length cost function to suppress vehicle yaw. On the contrary, if the vehicle yaw is not obvious, the weight of the path length cost function can be reduced. ξ 2 represents the difference between the vehicle heading angle and the terrain gradient direction. The smaller ξ 2 is, the more the vehicle tends to move along the terrain contour line, which is more conducive to highlighting the amplification effect of the terrain on the roll angle. It is necessary to increase the weight of the roll cost function to suppress the increase of the roll angle. On the contrary, if the vehicle tends to move along the terrain gradient direction, which is beneficial to suppressing the increase of the roll angle, the weight of the roll cost function can be reduced. Therefore, for the value of , there are the following inference rules:

[0137] When is small and is large, should take the median value;

[0138] When is small and is small, should take the large value;

[0139] When large and small, the median value should be taken;

[0140] When large and large, the small value should be taken;

[0141] For the detailed inference rules of [[[ID=]], please refer to Table 1:

[0142] Table 1 Inference rule table of [[[ID=]]

[0143]

[0144] In Table 1, the fuzzy sets {NB, NM, NS, ZO, PS, PM, PB} represent {Negative Big, Negative Medium, Negative Small, Zero, Positive Small, Positive Medium, Positive Big}.

[0145] Step 35: Use the centroid method for defuzzification to transform the [[[ID=]] fuzzy variable into a crisp variable. The defuzzification method can also include but is not limited to the maximum membership degree method and the weighted average method;

[0146] Step 36: Normalize the roll cost function weight [[[ID=]] to the range of [0, 1]. The normalization formula is as follows;

[0147]

[0148] In the formula, is the minimum value of [[[ID=]] represents the maximum value of [[[ID=]] represents the normalized roll cost function weight; To reduce the calculation, the weight of the path length cost function can be simply calculated as:

[0149]

[0150] The sum of the weighted vehicle roll cost function and the weighted path length cost function in step S40 above, characterized in that the sum of the weighted vehicle roll cost function and the weighted path length cost function is expressed as:

[0151]

[0152] Taking the negative gradient of the above formula gives:

[0153]

[0154] The modulus of the gradient is:

[0155]

[0156] The next vehicle position is:

[0157]

[0158]

[0159] where, x next and y next represent the position of the next step of the planned vehicle, v represents the speed of the vehicle, and t is the sampling interval time.

[0160] The method for judging whether the planned path reaches the destination in step S50 is:

[0161]

[0162] where, d p is the judgment distance. If the above formula holds, it is determined that the vehicle has reached the end point. The expression of d p can be set as: d p = vt.

[0163] See Figure 2 , Figure 2 which is the roll angle of the vehicle generated by the terrain and its calculation schematic diagram.

[0164] The plane coordinates of the left and right wheels of the vehicle are (x 1 , y 1 ) and (x 2 , y 2 ). According to the directional derivative and the distance between the left and right tires, l xy can be calculated as:

[0165]

[0166] It can be seen from this that the calculation expression of the roll angle of the vehicle generated by the terrain is:

[0167]

[0168] As described above, the present invention considers the vehicle rollover problem at the path planning level based on terrain information, not only avoiding complex vehicle dynamics and greatly reducing the computational complexity of the method, but also effectively suppressing the roll angle of the vehicle generated by terrain undulations when planning the path, thereby reducing the vehicle rollover risk. In addition, the present invention also weighs the path length under three-dimensional terrain and plans a safe path for intelligent vehicles that prevents rollover and has a short path length, which has important theoretical and practical significance.

[0169] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should be covered within the protection scope of the present invention. Moreover, the technical solutions between various embodiments of the present invention can be combined with each other, but it must be based on the fact that those skilled in the art can implement it. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claimed rights.

Claims

1. A three-dimensional path planning method for an intelligent driving vehicle to prevent rollover, characterized in that: It includes the following steps: Step S10: Construct a vehicle roll estimation model based on three-dimensional terrain information and vehicle heading angle, and design a roll cost function; Step S20: Design a path length cost function that combines two-dimensional and three-dimensional paths; Step S30: Design a fuzzy adaptive weight optimizer to optimize the weights of the vehicle roll cost function and the path length cost function; Step S40: Calculate the sum of the weighted vehicle roll cost function and the weighted path length cost function, and then use the gradient descent method to find its negative gradient, and determine the next path direction according to the negative gradient direction; Step S50: Determine whether the planned path reaches the destination. If not, repeat the above steps until reaching the destination; Among them, in the step S10, the three-dimensional terrain information is expressed by the function h(x, y), where (x, y) are the abscissa and ordinate in the three-dimensional terrain, and h(x, y) is the elevation at the (x, y) coordinate. When the vehicle heading angle is When the vehicle's coordinates are (x, y), the roll angle of the vehicle caused by the three-dimensional terrain is expressed as: where (x 1 , y 1 ) and (x 2 , y 2 ) are the coordinates corresponding to the tires on the left and right sides of the vehicle, and l xy is the projected length of the distance l between the left and right tires in the x-y plane when the vehicle is at (x, y); Taking the vehicle roll angle expression as the vehicle roll estimation model, the vehicle roll cost function in the three-dimensional terrain is designed as: where ξ d (x, y) is the distance attenuation coefficient.

2. The three-dimensional path planning method for an intelligent driving vehicle to prevent rollover according to claim 1, characterized in that, The gradient expression corresponding to h(x, y) is: In the formula, γ(x, y) represents the angle between the gradient of h(x, y) and the x-axis, i represents the unit vector in the x direction, j represents the unit vector in the y direction, and arctan2 is an extended form of the arctan function, and its return value range is (-π, π]; Since the vehicle axle is perpendicular to the vehicle heading angle, the directional derivative of h(x, y) along the vehicle axle is expressed as: In the formula, represents the 2-norm of and represents the heading angle of the vehicle's center of mass at (x, y); The coordinates corresponding to the tires on the left and right sides of the vehicle are:

3. The three-dimensional path planning method for an intelligent driving vehicle to prevent rollover according to claim 2, characterized in that, The projected length \(l\) of the distance \(l\) between the left and right tires in the \(x - y\) plane xy has the following calculation expression: Distance attenuation coefficient ξ d (x,y) is designed to be calculated as follows: where c 1 is the strength coefficient, d q is the x-y plane distance from the starting point to the destination, d(x, y) is the x-y plane distance from the point (x, y) to the destination, and the calculation expressions for d q and d(x, y) are as follows: Wherein, (x q , y q ) are the planar coordinates of the starting point, and (x m , y m ) are the planar coordinates of the ending point.

4. The three-dimensional path planning method for an intelligent driving vehicle to prevent rollover according to claim 3, characterized in that, In the step S20, the path length cost function that combines two-dimensional and three-dimensional paths is designed as: C l (x,y) = c 2 C 2l (x,y) + c 3 C 3l (x,y) where c 2 is the intensity coefficient of C 2l (x, y), and c 3 is the intensity coefficient of C 3l (x, y). C 2l (x, y) represents the x-y plane distance cost function from (x, y) to the destination, and its calculation expression is: C 3l (x, y) represents the cost function of the undulating terrain, which is used to suppress the undulation of the three-dimensional terrain. Its calculation expression is: where h d (x, y) is a height parameter related to the position. At the starting point, the height parameter h d (x, y) value is equal to the actual elevation h(x q , y q ) at the starting point; at the destination, the height parameter h d (x, y) value is equal to the actual elevation h(x m , y m ) at the destination; at any position on the map, the calculation expression of the height parameter h d (x, y) is designed as:

5. The three-dimensional path planning method for an intelligent driving vehicle to prevent rollover according to claim 4, characterized in that, In the step S30, the design method of the fuzzy adaptive weight optimizer includes the following steps: Step 31: Define the input variables as ξ 1 and ξ 2 , and the output variable is denotes the weight of the roll cost function C ρ . The calculation expression of the input variable ξ 1 is as follows: In the formula, δ(x, y) is the direction angle from the coordinate (x, y) to the destination, and its calculation expression is: Input variable ξ 2 The calculation expression is as follows: Step 32: Normalize the input variables ξ 1 and ξ 2 with the range set to [-5, 5]. The normalization method is as follows: Wherein, Max(ξ 1 ) and Min(ξ 1 ) represent the maximum and minimum values of the input variable ξ 1 ; Max(ξ 2 ) and Min(ξ 2 ) represent the maximum and minimum values of the input variable ξ 2 ; Step 33: Establish a first membership function, a second membership function, and a third membership function, where the first membership function, the second membership function, and the third membership function are respectively used to transform the input variable and the output variable into corresponding fuzzy variables; Step 34: Design fuzzy inference rules to obtain the inference rule table of the output variable ; Step 35: Use the centroid method for defuzzification to convert the fuzzy variable of the output variable into a crisp variable. The defuzzification method can include, but is not limited to, the maximum membership degree method and the weighted average method; Step 36: Normalize the weight of the roll cost function to the range of [0, 1], and the normalization formula is as follows; wherein, is the minimum value of the roll cost function weight , represents the maximum value of the roll cost function weight , represents the roll cost function weight after normalization; Path length cost function weight Expressed as:

6. The three-dimensional path planning method for an intelligent driving vehicle to prevent rollover according to claim 5, characterized in that, In the step S40, the sum of the weighted vehicle roll cost function and the weighted path length cost function is expressed as: Taking the negative gradient of the above formula gives: The modulus of the gradient is: The next vehicle position is: where, x next and y next represent the next position of the planned vehicle, v represents the speed of the vehicle, and t is the sampling interval time.

7. The three-dimensional path planning method for an intelligent driving vehicle to prevent rollover according to claim 6, characterized in that, In the step S50, the method for determining whether the planned path reaches the destination is: where d p is the determination distance. If the above equation holds, it is determined that the vehicle has reached the end point; Determination distance d p The expression of p is: d = vt.

Images