CN-121990001-A - Automatic driving lane changing method integrating intention risk field and state machine decision reversal mechanism

CN121990001ACN 121990001 ACN121990001 ACN 121990001ACN-121990001-A

Abstract

The invention discloses an automatic driving behavior decision method based on an intention driving risk field, which comprises the steps of 1 eliminating geometric distortion by constructing a generalized bounding box, 2 introducing a momentum correction and intention coupling mechanism, dynamically remolding the risk field topology to simulate defensive driving psychology, 3 utilizing a space-time worst principle score, combining a double risk threshold state machine to perform decision, and introducing a decision reversal mechanism to break deadlock. The method effectively solves the problem of decision concussion of driving behaviors, and remarkably improves the passing efficiency and safety under complex interaction scenes.

Inventors

BAI HAIJIAN
HONG YU
CHENG HAO

Assignees

合肥工业大学

Dates

Publication Date: 20260508
Application Date: 20260408

Claims (8)

1. An automatic driving behavior decision method based on an intention-driven risk field is characterized by comprising the following steps of: Step 1, establishing a local Cartesian coordinate system with a vehicle dynamics center as an origin and a two-dimensional coordinate system with an obstacle vehicle dynamics center as the origin; Under a two-dimensional coordinate system, calculating the generalized bounding box distance of the obstacle vehicle relative to the own vehicle, so as to construct an intention-driven generalized bounding box risk field; Step 2, introducing a momentum correction factor, and constructing a longitudinal risk field based on a Sigmoid function, so as to obtain a longitudinal normalized distance of the self-vehicle under a two-dimensional coordinate system; Step 3, introducing an intention driving correction mechanism and according to a transverse driving intention instruction of the own vehicle Constructing a transverse risk field which is intended to be driven, so as to obtain a transverse normalized distance under a two-dimensional coordinate system; step 4, candidate track set generated based on decision layer Calculating and planning time domain by adopting space-time worst principle Any candidate track Risk score of (2); step 5, setting a desired risk threshold As a state admittance condition, setting a maximum acceptable risk threshold As a condition for maintaining state, determining that the vehicle is in planning horizon by using double risk threshold according to risk score And the following behavior driving decision instruction enables the vehicle to jump between different driving states.
2. The method for automatically determining driving behavior based on intention-driven risk field according to claim 1, wherein in step 1, a generalized bounding box risk field is constructed by using formula (1): (1) in the formula (1), the components are as follows, Vehicle position indicating that obstacle vehicle i at time t is in two-dimensional coordinate system Risk potential energy generated at the site; Representing the position of an obstacle vehicle i in a two-dimensional coordinate system at t The generalized bounding box distance at (1) is as follows: (2) in the formula (2), the amino acid sequence of the compound, Representing the longitudinal coordinates of the obstacle vehicle i at the time t under the two-dimensional coordinate system Velocity in the machine and longitudinal directions The generalized bounding box distance under the lower part, Representing the longitudinal coordinates of the obstacle vehicle i at the time t under the two-dimensional coordinate system Velocity at and transverse Lower generalized bounding box distance.
3. An automatic driving behavior decision method based on intention-driven risk field according to claim 2, wherein said step 2 comprises: step 2.1, obtaining the longitudinal positions of the obstacle vehicle i and the own vehicle at the time t under a two-dimensional coordinate system by utilizing the step (3) Basic physical distance at : (3) In the formula (3), the amino acid sequence of the compound, In order to obstruct the length of the vehicle i, The longitudinal coordinate of the self-vehicle at the moment t under a two-dimensional coordinate system; Step 2.2, defining a momentum correction factor, wherein the momentum correction factor comprises the longitudinal speed of the vehicle as shown in the formula (4) Forward momentum parameter of the lower part And a backward buffering parameter represented by formula (5) : (4) In the formula (4), the amino acid sequence of the compound, As the momentum gain coefficient, Is the minimum safe distance coefficient; step 2.3, obtaining the longitudinal coordinates of the vehicle at the time t by using the method (5) Velocity in the machine and longitudinal directions Dynamic longitudinal shape parameters below : (5) In the formula (5), the amino acid sequence of the compound, Is a longitudinal coordinate Is a function of the standard Sigmoid of (c), Is a smoothing factor; step 2.4, obtaining the longitudinal coordinate of the self-vehicle at the moment t under a two-dimensional coordinate system by utilizing the step (6) Velocity in the machine and longitudinal directions Lower longitudinal normalized distance : (6)。
4. An automatic driving behavior decision method based on intention-driven risk field according to claim 3, wherein said step 3 comprises: step 3.1, obtaining the transverse positions of the obstacle vehicle i and the own vehicle under the two-dimensional coordinate system by utilizing the step (7) Basic physical distance at : (7) In the formula (7), the amino acid sequence of the compound, Vehicle width is the obstacle vehicle i; Step 3.2, defining an intent-to-drive correction mechanism using equations (8) and (9), including the vehicle being at lateral position y and lateral velocity in a two-dimensional coordinate system Lower physical layer speed correction term With the vehicle in a two-dimensional coordinate system at a lateral position y and driving intent instructions Underlying cognitive layer intent correction term : (8) (9) In the formulas (8) and (9), For the lateral speed of the own vehicle in relation to the obstacle vehicle i in a two-dimensional coordinate system, Is a gain coefficient; In order to be aware of the coefficient of expansion, If the own vehicle is on the left side of the obstacle vehicle i, the azimuth symbol of the own vehicle relative to the center of the obstacle vehicle i is given by If the own vehicle is on the right side of the obstacle vehicle i, the following procedure is made = +1 =-1, Representing the lateral position of the vehicle in a cartesian coordinate system, Indicating the position of the obstacle vehicle in the Cartesian coordinate system, and if the obstacle vehicle is on the left side of the own vehicle, the obstacle vehicle is given the following order = -1, Otherwise, let =+1, Is a cognitive shear indication function, and comprises: If it is And is also provided with Order in principle =1; If it is And is also provided with Order in principle =1; Otherwise, let =0; Step 3.3, obtaining the transverse coordinate y and the transverse speed of the self-vehicle under the two-dimensional coordinate system by utilizing the step (11) Dynamic lateral shape parameters below : (10) In the formula (10), the amino acid sequence of the compound, Is the base risk width; Step 3.4, obtaining a driving intention instruction by using the method (11) Lateral coordinate y and lateral speed of the lower vehicle in two-dimensional coordinate system Lower transverse normalized distance : (11)。
5. The method for automatically determining driving behavior based on intention-driven risk field according to claim 1, wherein the candidate trajectory is obtained by using formula (12) in step 4 Final risk score of (2) : (12) In the formula (13), the amino acid sequence of the compound, Representing the candidate track of the own vehicle at the moment t A position on the upper surface; a generalized bounding box risk field representing an obstacle vehicle i, And the set of obstacle vehicles around the vehicle at the moment t is represented, and t is [0, T ].
6. An automatic driving behavior decision method based on intention-driven risk field according to claim 5, wherein said step 5 comprises: step 5.1, defining the vehicle driving state including following the state Execution state of lane change State of stopping channel change ; Step 5.2, setting the expected risk threshold as Setting the maximum acceptable risk threshold as ; Step 5.3, if < Outputting the jump instruction to program the time domain Inner bicycle is in following state Jump to the transition execution state If (1) > Outputting a termination instruction to program the time domain Inner self-vehicle channel changing execution state Jump to transition termination state ; If jump to the transition termination state In the process of (2) < Outputting a return instruction to enable the self-vehicle to be in a channel changing suspension state Jump back to track execution state ; If the vehicle transverse position deviation is converged to the target lane allowable range in the lane change execution process, the lane change is completed, and a following instruction is output, so that the running state of the vehicle is converted into a following state ; If the vehicle completely retreats to the original lane center line in the lane change stopping process, the vehicle indicates the end of the danger avoidance, and outputs a reset instruction to reset the running state of the vehicle to the following state 。
7. An electronic device comprising a memory and a processor, wherein the memory is configured to store a program that supports the processor to perform the method of any of claims 1-6, the processor being configured to execute the program stored in the memory.
8. A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when run by a processor performs the steps of the method according to any of claims 1-6.

Description

Automatic driving lane changing method integrating intention risk field and state machine decision reversal mechanism Technical Field The invention belongs to the technical field of automatic driving, and particularly relates to an automatic driving behavior decision method based on an intention driving risk field. Background In the automatic driving layered architecture, the behavior decision layer plays a key role in the top and bottom. However, the existing method still faces three theoretical bottlenecks when complex dynamic interaction is processed, namely firstly, collision risks are caused by geometric characterization distortion, because a particle model cannot cover the actual physical outline of a vehicle, safety envelope deletion is extremely easy to generate when the vehicle works in a narrow space or interacts with a large vehicle in a short distance, collision hidden danger is increased, secondly, decision logic lacks defensiveness and anthropomorphic property, because a system cannot simulate the psychology of a person for first looking ahead, channel switching triggering time is too mechanical, enough safety margin is difficult to reserve in game interaction, high-frequency oscillation of instructions is extremely easy to generate under small fluctuation of environment, thirdly, decision stiffness leads to low passing efficiency, a traditional state machine often falls into deadlock logic which is needed to be completely retracted once stopping under complex working conditions, a dynamically-changed traffic window cannot be flexibly captured, and the passing capacity of the vehicle under complex scenes is severely limited. Disclosure of Invention In order to overcome the defects that geometric distortion exists in environmental characterization and the defensive cognition mechanism of a human driver is lacked in the prior art, the invention provides an automatic driving method based on an intention-driven risk field and a decision reversal mechanism, and aims to realize automatic driving behavior decision with geometric accuracy, cognition humanization and decision stability by constructing an asymmetric risk field conforming to the real contour of a vehicle and introducing an intention-driven field intensity remodelling mechanism and dual threshold decision logic, so that the driving behavior decision concussion problem can be effectively solved, and the traffic efficiency and safety under a complex interaction scene are remarkably improved. In order to achieve the aim of the invention, the invention adopts the following technical scheme: the invention discloses an automatic driving behavior decision method based on an intention driving risk field, which is characterized by comprising the following steps of: Step 1, establishing a local Cartesian coordinate system with a vehicle dynamics center as an origin and a two-dimensional coordinate system with an obstacle vehicle dynamics center as the origin; Under a two-dimensional coordinate system, calculating the generalized bounding box distance of the obstacle vehicle relative to the own vehicle, so as to construct an intention-driven generalized bounding box risk field; Step 2, introducing a momentum correction factor, and constructing a longitudinal risk field based on a Sigmoid function, so as to obtain a longitudinal normalized distance of the self-vehicle under a two-dimensional coordinate system; Step 3, introducing an intention driving correction mechanism and according to a transverse driving intention instruction of the own vehicle Constructing a transverse risk field which is intended to be driven, so as to obtain a transverse normalized distance under a two-dimensional coordinate system; step 4, candidate track set generated based on decision layer Calculating and planning time domain by adopting space-time worst principleAny candidate trackRisk score of (2); step 5, setting a desired risk threshold As a state admittance condition, setting a maximum acceptable risk thresholdAs a condition for maintaining state, determining that the vehicle is in planning horizon by using double risk threshold according to risk scoreAnd the following behavior driving decision instruction enables the vehicle to jump between different driving states. The automatic driving behavior decision method based on the intention-driven risk field is also characterized in that in the step 1, a generalized bounding box risk field is constructed by using the formula (1): (1) in the formula (1), the components are as follows, Vehicle position indicating that obstacle vehicle i at time t is in two-dimensional coordinate systemRisk potential energy generated at the site; Representing the position of an obstacle vehicle i in a two-dimensional coordinate system at t The generalized bounding box distance at (1) is as follows: (2) in the formula (2), the amino acid sequence of the compound, Representing the longitudinal coordinates of the obstacle vehicle i at the time t