Search

CN-121989995-A - Traffic participant trajectory prediction method, device, electronic equipment and storage medium

CN121989995ACN 121989995 ACN121989995 ACN 121989995ACN-121989995-A

Abstract

The application discloses a traffic participant track prediction method, a traffic participant track prediction device, electronic equipment and a storage medium, and relates to the technical field of automatic driving. The method comprises the steps of obtaining surrounding traffic participant information, vehicle state information and environment information, constructing a vehicle surrounding lane diagram, screening target traffic participants to obtain position parameters of the vehicle surrounding lane diagram, associating the target traffic participants with the lane diagram, constructing a corresponding coordinate system, carrying out coordinate projection on the position parameters, identifying driving intention according to a preset rule based on continuous frame quantification parameters, setting polynomial boundary conditions, determining fitting coefficients, obtaining transverse and longitudinal position coordinates of the target traffic participants within a future preset duration through polynomial fitting by using the fitting coefficients, generating an initial driving track, processing the initial track, and outputting a final predicted track. The application can improve the accuracy of identifying the driving intention of the traffic participant and the reliability of track prediction, and adapt to the track pre-judging requirement of automatic driving in complex road scenes.

Inventors

  • ZHAO HUITING
  • HAO ZHI
  • LIU GUIYU
  • ZHANG ZONGTIAN

Assignees

  • 一汽解放汽车有限公司

Dates

Publication Date
20260508
Application Date
20260325

Claims (10)

  1. 1. A method of traffic participant trajectory prediction comprising the steps of: Acquiring surrounding traffic participant information, own state information of a vehicle and map environment information; Constructing a lane diagram around the vehicle based on the acquired information, screening target traffic participants according to a preset space range, and acquiring position parameters of the target traffic participants; associating the target traffic participant with the peripheral lane diagram and constructing a corresponding coordinate system; Carrying out coordinate projection on the position parameters; identifying the driving intention of the traffic participant according to a preset rule based on the quantization parameters of the continuous frames; based on the driving intention, setting boundary conditions of a polynomial in combination with a preset road traffic rule, and determining fitting coefficients; fitting transverse and longitudinal position coordinates within a preset time period of a traffic participant in the future by using a polynomial according to the fitting coefficient to generate an initial running track; And after sampling and filtering smoothing the initial running track, obtaining and outputting the final predicted track of the traffic participant.
  2. 2. The method of claim 1, wherein the surrounding traffic participant information comprises a location, a speed, and 2D frame coordinates of the traffic participant; the self-vehicle state information is acquired through a vehicle-mounted positioning system, and comprises a self-vehicle position, a traveling speed and a traveling direction; The map environment information is obtained through a navigation map and comprises lane distribution, a road center line, traffic lights and stop line related information.
  3. 3. The traffic participant trajectory prediction method of claim 2, comprising: extracting the coordinates of the corner points and the central points of the 2D frame of the traffic participant, and dividing the current lane, the left lane and the right lane; Establishing a Frenet coordinate system by taking the central line of the current lane road as a reference, and defining the left side of the self-vehicle forward direction as the forward direction of the abscissa axis; and projecting the related coordinates of the traffic participants to the coordinate system to obtain left-side boundary abscissas, center point abscissas and right-side boundary abscissas.
  4. 4. The traffic participant trajectory prediction method of claim 1, wherein the travel intent comprises left lane change, right lane change and straight travel; the preset rules are specifically as follows: the left lane change is that the abscissa of the central point is larger than zero, the abscissa of the left boundary of the traffic participant is larger than the distance between the central point of the traffic participant and the left boundary of the current lane, and the abscissa of the central point is continuously increased by 3 frames or the transverse speed of the traffic participant is continuously increased by 3 frames and is larger than 0; Right lane change, namely, the condition that the absolute value of the abscissa of the central point is smaller than zero, the absolute value of the abscissa of the right side boundary is larger than the distance between the central point of the traffic participant and the right side boundary of the current lane, and the continuous 3 frames of the abscissa of the central point are smaller or the continuous 3 frames of the transverse speed of the traffic participant are smaller than 0; straight running, namely, the left lane change and right lane change judging conditions are not met; wherein, the continuous 3 frames are the current frame, the last frame and the last frame.
  5. 5. The traffic participant trajectory prediction method of claim 4, wherein the trajectory prediction is performed by: fitting the relation between the transverse coordinates and time by using a polynomial of 3 times, and fitting the relation between the longitudinal coordinates and time by using a polynomial of 4 times, wherein polynomial coefficients are respectively determined through corresponding constraint conditions; Generating an initial track by combining the transverse and longitudinal fitting results, and obtaining a final predicted track after sampling at a preset frequency and smoothing by Kalman filtering; the track end point is positioned on the target lane road center line corresponding to the intention by default, and the predicted duration can be customized.
  6. 6. The traffic participant trajectory prediction method of claim 5, comprising: the constraint conditions of the 3 th-degree polynomial fitting comprise an initial transverse position, an initial transverse speed, an end transverse position and an end transverse speed; constraints for the 4 th degree polynomial fit include initial longitudinal position, initial longitudinal velocity, initial longitudinal acceleration, terminal longitudinal velocity, and terminal longitudinal acceleration.
  7. 7. The traffic participant trajectory prediction method of claim 6, wherein the 3-degree polynomial fit transverse coordinates are specifically: ; Wherein, the To indicate the time of day of the traffic participant Lateral displacement relative to the road reference line; An accumulated time calculated for the traffic participant from the initial time; is an initial transverse position, Is an initial transverse velocity, Is a transverse acceleration related quantity, Is a lateral jerk related quantity.
  8. 8. The traffic participant trajectory prediction method of claim 7, wherein the 4 th order polynomial fit longitudinal coordinates are specifically: ; Wherein, the At the moment of time for the traffic participants Longitudinal displacement relative to the starting point; An accumulated time calculated for the traffic participant from the initial time; Is an initial longitudinal position coefficient, Is an initial longitudinal velocity coefficient, Is an initial longitudinal acceleration coefficient, Is a longitudinal Jerk (Jerk) coefficient, Is a fourth order term coefficient.
  9. 9. A traffic participant trajectory prediction device, comprising: the system comprises an information acquisition module, a target screening module, a coordinate construction module, a coordinate projection module, an intention recognition module, a fitting coefficient acquisition module, an initial track generation module and a final track output module; The information acquisition module is used for acquiring surrounding traffic participant information, own vehicle state information, map environment information, fitting coefficient acquisition module, initial track generation module and final track output module; The target screening module is used for constructing a lane diagram around the vehicle based on the acquired information, screening target traffic participants according to a preset space range and acquiring position parameters of the target traffic participants; the coordinate construction module is used for associating the target traffic participant with the peripheral lane diagram and constructing a corresponding coordinate system; The coordinate projection module is used for carrying out coordinate projection on the position parameters; the intention recognition module is used for recognizing the driving intention of the traffic participant according to a preset rule based on the quantization parameters of the continuous frames; the fitting coefficient acquisition module is used for determining fitting coefficients by combining boundary conditions of a polynomial set by a preset road traffic rule based on the driving intention; The initial track generation module is used for generating an initial running track by adopting a polynomial to fit the transverse and longitudinal position coordinates of the traffic participant within a future preset time length according to the fitting coefficient; and the final track output module is used for obtaining and outputting the final predicted track of the traffic participant after sampling, filtering and smoothing the initial running track.
  10. 10. The electronic equipment is characterized by comprising a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory are communicated with each other through the communication bus; The memory has stored therein a computer program which, when executed by the processor, causes the processor to perform the steps of a traffic participant trajectory prediction method as claimed in any one of claims 1 to 8.

Description

Traffic participant trajectory prediction method, device, electronic equipment and storage medium Technical Field The present invention relates to the field of automatic driving automobiles, and in particular, to a traffic participant trajectory prediction method, a traffic participant trajectory prediction apparatus, an electronic device, and a storage medium. Background With the increasing maturity of automatic driving technology and the increasing perfection of road facilities, the market share occupied by class 2/3 automatic driving vehicles is increasing according to the class of automatic driving classified by the american society of automotive engineers. The accurate prediction of the driving intention and the driving track of surrounding traffic participants by an automatic driving vehicle is important to driving safety, and how to accurately predict the driving track of the traffic participants under the condition of considering the calculation complexity of the whole automatic driving system becomes a popular research content. Disclosure of Invention In view of the above, the present invention aims to provide a traffic participant trajectory prediction method, a traffic participant trajectory prediction device, an electronic device and a storage medium, which aim to solve the technical problem of how to accurately predict the travel trajectory of a river copper participant in a structured road-oriented scene. The invention provides the following scheme: According to one aspect of the present invention, there is provided a traffic participant trajectory prediction method comprising the steps of: Acquiring surrounding traffic participant information, own state information of a vehicle and map environment information; Constructing a lane diagram around the vehicle based on the acquired information, screening target traffic participants according to a preset space range, and acquiring position parameters of the target traffic participants; associating the target traffic participant with the peripheral lane diagram and constructing a corresponding coordinate system; Carrying out coordinate projection on the position parameters; identifying the driving intention of the traffic participant according to a preset rule based on the quantization parameters of the continuous frames; based on the driving intention, setting boundary conditions of a polynomial in combination with a preset road traffic rule, and determining fitting coefficients; fitting transverse and longitudinal position coordinates within a preset time period of a traffic participant in the future by using a polynomial according to the fitting coefficient to generate an initial running track; And after sampling and filtering smoothing the initial running track, obtaining and outputting the final predicted track of the traffic participant. Further, the surrounding traffic participant information includes the location, speed, and 2D frame coordinates of the traffic participant; the self-vehicle state information is acquired through a vehicle-mounted positioning system, and comprises a self-vehicle position, a traveling speed and a traveling direction; The map environment information is obtained through a navigation map and comprises lane distribution, a road center line, traffic lights and stop line related information. Further, the method comprises the steps of: extracting the coordinates of the corner points and the central points of the 2D frame of the traffic participant, and dividing the current lane, the left lane and the right lane; Establishing a Frenet coordinate system by taking the central line of the current lane road as a reference, and defining the left side of the self-vehicle forward direction as the forward direction of the abscissa axis; and projecting the related coordinates of the traffic participants to the coordinate system to obtain left-side boundary abscissas, center point abscissas and right-side boundary abscissas. Further, the driving intention comprises left lane changing, right lane changing and straight running; the preset rules are specifically as follows: the left lane change is that the abscissa of the central point is larger than zero, the abscissa of the left boundary of the traffic participant is larger than the distance between the central point of the traffic participant and the left boundary of the current lane, and the abscissa of the central point is continuously increased by 3 frames or the transverse speed of the traffic participant is continuously increased by 3 frames and is larger than 0; Right lane change, namely, the condition that the absolute value of the abscissa of the central point is smaller than zero, the absolute value of the abscissa of the right side boundary is larger than the distance between the central point of the traffic participant and the right side boundary of the current lane, and the continuous 3 frames of the abscissa of the central point are smaller or the continuous 3 frame