A lateral dynamic avoidance decision method for unmanned vehicle and unmanned vehicle system

Abstract

The present invention provides a lateral dynamic avoidance decision method for an unmanned vehicle and an unmanned vehicle system, the decision method comprising the following steps: based on a driving scene, obtaining an obstacle consideration area according to vehicle state information and lane information or according to vehicle state information, lane information and the horizontal coordinate of a planned space, judging whether a suspected obstacle is an actual obstacle based on the obstacle consideration area: if so, obtaining the relative state information of the actual obstacle according to the vehicle state information and the state information of the actual obstacle; obtaining an avoidance decision according to the vehicle state information, lane information, lateral reference position and the relative state information of the actual obstacle; the unmanned vehicle performs dynamic avoidance based on the avoidance decision. The lateral dynamic avoidance decision method for an unmanned vehicle provided by the present invention has high accuracy in dynamic obstacle detection and recognition, and strong stability in avoidance decision.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned vehicles, and in particular to an unmanned vehicle lateral dynamic avoidance decision method and an unmanned vehicle system. Background Art

[0002] When driving on motorway, unmanned vehicles need to make decisions based on information such as static obstacles, dynamic obstacles, lane lines, and curbs provided by perception or maps. For example, for static objects such as static obstacles, it is relatively easy for the vehicle to make a detour decision, while for dynamic obstacles, it is necessary to consider the relative position, speed, acceleration and other information between the vehicle and the obstacle, as well as traffic information such as lane lines to make a comprehensive decision. Due to the stability of perception detection, there are certain errors in the detection and identification of dynamic obstacles. Over-confidence in the high-level information provided by perception will challenge the stability of decision-making. Summary of the invention

[0003] In view of the problems existing in the prior art, the present invention provides a lateral dynamic avoidance decision method for an unmanned vehicle, which screens and locates actual obstacles from suspected obstacles through an obstacle consideration area to ensure the accuracy of dynamic obstacle detection and identification. The method obtains avoidance decisions on the premise of fully considering lane information, the vehicle status information of the unmanned vehicle itself, the lateral reference position and the status information of the actual obstacle, thereby ensuring the stability of the avoidance decisions.

[0004] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is as follows: a method for making a lateral dynamic avoidance decision for an unmanned vehicle, comprising the following steps:

[0005] Obtain driving scene, lane information, planning space horizontal coordinates, lateral reference position and vehicle status information of the unmanned vehicle;

[0006] Based on the driving scenario, obtaining an obstacle consideration area according to the vehicle state information and the lane information, or obtaining an obstacle consideration area according to the lane information, the horizontal coordinate of the planning space and the vehicle state information;

[0007] Obtain status information of suspected obstacles;

[0008] Determining whether the suspected obstacle is an actual obstacle based on the obstacle consideration area: if so, obtaining relative state information of the actual obstacle according to the vehicle state information and the state information of the actual obstacle;

[0009] Obtaining an avoidance decision based on the vehicle state information, the lane information, and the relative state information between the lateral reference position and the actual obstacle;

[0010] The unmanned vehicle performs dynamic avoidance based on the avoidance decision.

[0011] In some of these embodiments, the vehicle state information includes vehicle boundary coordinates and vehicle width;

[0012] The lane information includes the lane boundary horizontal coordinate and the lane line type;

[0013] The obstacle consideration area includes the horizontal coordinates of the area limit position and the area boundary coordinates;

[0014] The driving scenarios include normal driving scenarios and preset driving scenarios;

[0015] The relative state information includes a relative driving state and a relative position.

[0016] In some embodiments, based on the normal driving scenario, the step of obtaining the horizontal coordinate of the area limit position is:

[0017] Obtaining the first coefficient, the second coefficient and the safety distance margin;

[0018] A first abscissa is obtained by multiplying the lane boundary abscissa inward by a first product, where the first product is the product of the vehicle width and the first coefficient;

[0019] Obtaining the initial area limit position horizontal coordinate according to the first horizontal coordinate, or obtaining the initial area limit position horizontal coordinate according to the first horizontal coordinate and the planned space horizontal coordinate;

[0020] Based on the lane line type, the horizontal coordinate of the area limit position is obtained by retracting the safety distance margin inward from the horizontal coordinate of the initial area limit position, or by multiplying the horizontal coordinate of the initial area limit position inward by a second product, wherein the second product is the product of the safety distance margin and the second coefficient.

[0021] In some embodiments, the vehicle boundary coordinates include a vehicle boundary horizontal coordinate and a vehicle boundary vertical coordinate; the region boundary coordinates include a region boundary horizontal coordinate and a region boundary vertical coordinate;

[0022] Based on the normal driving scenario, the steps of obtaining the horizontal coordinate of the area boundary are:

[0023] Obtaining a third coefficient, a detection distance margin, and a detection distance;

[0024] Based on the lane line type, the detection distance margin is expanded outward according to the lane boundary horizontal coordinate to obtain the initial area boundary horizontal coordinate, or the lane boundary horizontal coordinate is expanded outward according to a third product to obtain the initial area boundary horizontal coordinate, where the third product is the product of the detection distance margin and the third coefficient;

[0025] Obtaining the region boundary horizontal coordinate according to the initial region boundary horizontal coordinate, the vehicle boundary horizontal coordinate and the detection distance;

[0026] Based on the normal driving scenario, the steps of obtaining the vertical coordinate of the area boundary are:

[0027] obtaining a relative driving state and a fourth coefficient;

[0028] Based on the relative driving state, the area boundary ordinate is obtained according to the vehicle boundary ordinate, or according to the vehicle boundary ordinate and the detection distance, or according to the vehicle boundary ordinate, the fourth coefficient and the detection distance.

[0029] In some embodiments, the vehicle state information also includes current heading angle information; the step of obtaining the relative driving state is:

[0030] Obtain the same-direction detection threshold and the heading angle information of the obstacle;

[0031] The obtaining of a heading angle difference according to the current heading angle information and the heading angle information of the obstacle;

[0032] The relative driving state is obtained by comparing the heading angle difference with the same-direction detection threshold.

[0033] In some embodiments, the state information of the suspected obstacle includes obstacle boundary coordinates and obstacle longitudinal speed, and the obstacle boundary coordinates include at least obstacle boundary longitudinal coordinates; the vehicle state information also includes vehicle longitudinal speed; the relative position includes at least longitudinal relative position;

[0034] The steps of judging whether the suspected obstacle is an actual obstacle based on the obstacle consideration area are as follows:

[0035] Determine whether the suspected obstacle is within the obstacle consideration area according to the obstacle consideration area and the obstacle boundary coordinates: if so, the suspected obstacle is an obstacle to be confirmed;

[0036] Obtaining a longitudinal relative position of the obstacle to be confirmed according to the obstacle boundary ordinate, the obstacle longitudinal speed, the vehicle boundary ordinate and the vehicle longitudinal speed of the obstacle to be confirmed;

[0037] Based on the longitudinal relative position, or based on the longitudinal relative position and the relative state information, it is determined whether there is a possibility of contact between the obstacle to be confirmed and the unmanned vehicle according to the state information of the obstacle to be confirmed and the vehicle state information: if so, the obstacle to be confirmed is the actual obstacle.

[0038] In some embodiments, the obstacle boundary coordinates also include obstacle boundary horizontal coordinates; the vehicle state information also includes its own horizontal coordinates; the relative position also includes a lateral relative position;

[0039] The step S4 further includes: obtaining the lateral relative position of the actual obstacle according to the obstacle boundary lateral coordinate of the actual obstacle and the own lateral coordinate of the actual obstacle, or obtaining the lateral relative position of the actual obstacle according to the obstacle boundary lateral coordinate of the actual obstacle, the area boundary lateral coordinate and the own lateral coordinate of the actual obstacle.

[0040] In some of these embodiments, the avoidance decision includes a hold decision, a center decision, a left avoidance decision, and a right avoidance decision.

[0041] In some of the embodiments, the left side avoidance decision includes a left side lateral avoidance decision and a left side lateral and longitudinal comprehensive avoidance decision; the right side avoidance decision includes a right side lateral avoidance decision and a right side lateral and longitudinal comprehensive avoidance decision.

[0042] An unmanned vehicle system comprises an unmanned vehicle and a dynamic avoidance module, wherein the dynamic avoidance module is used to implement a lateral dynamic avoidance decision method of the unmanned vehicle.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] The present invention screens and locates actual obstacles from suspected obstacles through the obstacle consideration area, thereby ensuring the accuracy of dynamic obstacle detection and recognition, and obtaining avoidance decisions on the premise of fully considering lane information, the vehicle status information of the unmanned vehicle itself, the lateral reference position and the status information of the actual obstacle, thereby ensuring the stability of the avoidance decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a flow chart of the dynamic lateral avoidance decision-making method for an unmanned vehicle of the present invention;

[0046] Figure 2 A schematic diagram of the process of obtaining the horizontal coordinates of the regional limit position of the present invention;

[0047] Figure 3 A schematic diagram of a process for obtaining the horizontal coordinates of a region boundary according to the present invention;

[0048] Figure 4 A schematic diagram of a flow chart for obtaining a relative driving state according to the present invention;

[0049] Figure 5 A schematic diagram of a process for determining whether a suspected obstacle is an actual obstacle according to the present invention;

[0050] Figure 6A schematic diagram of a process of traversing an actual obstacle set to obtain an obstacle set to be decided according to the present invention;

[0051] Figure 7 The figure is a flow chart of obtaining an avoidance decision according to a set of obstacles to be decided in the present invention. DETAILED DESCRIPTION

[0052] In order to clearly illustrate the technical features of this solution, the following will be combined with the accompanying drawings and examples to describe in detail the implementation methods of this application, so that the application can fully understand how to use technical means to solve technical problems and achieve the corresponding technical effects and implement them accordingly. The embodiments of this application and the various features in the embodiments can be combined with each other without conflict, and the technical solutions formed are all within the scope of protection of this application.

[0053] The present disclosure is defined under the Frenet system, that is, in the following embodiments, the left side of the lateral coordinate value is positive and the right side is negative; in the PNC framework with lateral and longitudinal decoupling, the lateral direction needs to provide a lateral dynamic avoidance function and provide a priori information for longitudinal decision-making. The present invention designs a framework to coordinate these factors to achieve dynamic avoidance of unmanned vehicles.

[0054] See also Figure 1 The embodiment of the present invention provides a method for making decisions on lateral dynamic avoidance of an unmanned vehicle, comprising the following steps:

[0055] Acquire driving scenarios, lane information, horizontal coordinates of planning space, horizontal reference positions and vehicle status information of the unmanned vehicle; the vehicle will experience different driving scenarios during driving. Usually, the normal driving scenario is the main one, that is, driving on linear roads. It will also drive into preset driving scenarios, including left-turn scenarios at intersections, straight-ahead scenarios in conflict zones, and gate scenarios. The horizontal reference position indicates the expected horizontal position of the unmanned vehicle, i.e. the vehicle itself. It is a global variable in a planning process of one frame. The upstream module can adjust the horizontal reference position horizontal coordinate according to its decision. Make updates;

[0056] The planning space is the horizontal planning space, which means the horizontal space where the unmanned vehicle can travel. The horizontal coordinate of the left boundary of the planning space can be given. and the horizontal coordinate of the right boundary of the planning space ; Theoretically, all lateral spaces within the lane boundary can be used for driving, but the upstream module may strategically narrow some lateral planning spaces for driving safety considerations. If given, it can be used here;

[0057] Obtain obstacle consideration areas based on driving scenarios and vehicle status information and lane information , or obtain the obstacle consideration area based on lane information, planning space horizontal coordinates and vehicle status information ; It is worth noting that the entry barrier considers the area Obstacles within the area are likely to affect the normal driving of the unmanned vehicle and are obstacles that need further consideration. Confirm the obstacle consideration area It is a prerequisite for obstacle identification and avoidance decision making;

[0058] Obtain the status information of suspected obstacles. All obstacles that can be identified around the unmanned vehicle are suspected obstacles. The specific identification distance varies depending on factors such as vehicle model, sensor configuration, and algorithm optimization. This disclosure only considers dynamic obstacles and does not consider the avoidance of static obstacles. For dynamic obstacles, if the type of obstacle does not conform to the driving logic (such as dynamic garbage bags, etc.), it is not considered as a suspected obstacle;

[0059] Considering regions based on barriers Determine whether the suspected obstacle is an actual obstacle. If the suspected obstacle intersects with the obstacle consideration area, it is considered that the suspected obstacle has a high probability of affecting the normal driving of the unmanned vehicle, and the suspected obstacle will be recognized as an actual obstacle. If so, the relative status information of the actual obstacle is obtained based on the vehicle status information and the status information of the actual obstacle.

[0060] Obtain avoidance decisions based on vehicle status information, lane information, lateral reference position, and relative status information of actual obstacles;

[0061] The unmanned vehicle performs dynamic avoidance based on avoidance decisions.

[0062] Considering areas by barriers The actual obstacles are screened and located from the suspected obstacles to ensure the accuracy of dynamic obstacle detection and identification, and avoidance decisions are made under the premise of fully considering the lane information, the vehicle status information of the unmanned vehicle itself, the lateral reference position and the status information of the actual obstacles to ensure the stability of the avoidance decisions.

[0063] Lane information includes lane boundary coordinates and lane line type. Lane boundary coordinates include the horizontal coordinates of the left lane boundary. and the horizontal coordinate of the right lane boundary ,The lane line type mainly based on the process of obtaining avoidance decisions in this disclosure is the white dotted line;

[0064] Vehicle status information includes vehicle boundary coordinates and vehicle width ;

[0065] Obstacle consideration area Includes the horizontal coordinates of the area limit position and the area boundary coordinates; the horizontal coordinates of the area limit position include the horizontal coordinates of the left limit position of the area The horizontal coordinate of the right limit position of the region , consider the area for obstacles The horizontal coordinate of the vehicle's driving limit position;

[0066] Relative status information includes relative driving status and relative position; relative driving status is the driving state of the obstacle vehicle relative to the unmanned vehicle, such as , Reverse driving , vertical driving etc.; relative position refers to the position relationship of the obstacle vehicle in front, back, left, and right of the unmanned vehicle;

[0067] See also Figure 2 In some embodiments, based on a normal driving scenario, the steps of obtaining the horizontal coordinates of the area limit position are:

[0068] Get the first coefficient , the second coefficient and safety distance margin , the first coefficient , the second coefficient and safety distance margin All can be obtained through customization;

[0069] The first horizontal coordinate is obtained by multiplying the first product inward according to the horizontal coordinate of the lane boundary. The first product is the vehicle width The product of the left horizontal coordinate value and the first coefficient is that in the Frenet system, the horizontal coordinate value of the left side is positive and the horizontal coordinate value of the right side is negative. The left boundary is reduced when it is contracted inward and increased when it is expanded outward, and the right boundary is increased when it is contracted inward and decreased when it is expanded outward. Therefore, the horizontal coordinate of the left boundary of the lane is The inward contraction is decreasing, and the horizontal coordinate of the right limit position of the area To gather inward is to add;

[0070] The initial area limit position abscissa is obtained according to the first abscissa, or the initial area limit position abscissa is obtained according to the first abscissa and the abscissa of the planning space (the abscissa of the left boundary of the planning space and the horizontal coordinate of the right boundary of the planning space ) obtains the horizontal coordinate of the initial area limit position. Correspondingly, the horizontal coordinate of the initial area limit position also includes the horizontal coordinate of the left limit position of the initial area and the horizontal coordinate of the right limit position of the initial area ;

[0071] If the planning space is given, the horizontal coordinate of the planning space is given and ,but:

[0072] The horizontal coordinate of the left limit position of the initial area for:

[0073] ;

[0074] The horizontal coordinate of the right limit position of the initial area for:

[0075] ;

[0076] If no planning space is given, then:

[0077] The horizontal coordinate of the left limit position of the initial area for:

[0078] ;

[0079] The horizontal coordinate of the right limit position of the initial area for:

[0080] ;

[0081] Usually, the first coefficient Take 0.5;

[0082] Based on the lane line type, the safety distance margin is adjusted inward according to the horizontal coordinate of the initial area limit position Get the horizontal coordinate of the area limit position, or multiply the initial area limit position horizontal coordinate inward by the second product to get the horizontal coordinate of the area limit position. The second product is the safety distance margin. With the second coefficient The product of

[0083] If the left lane line is not a white dashed line, the horizontal coordinate of the left limit position of the initial area Inwards customizable safety distance margin , get the horizontal coordinate of the left limit position of the area :

[0084] ;

[0085] If not, the horizontal coordinate of the left limit position of the initial area is Inward times the safety distance margin , get the horizontal coordinate of the left limit position of the area :

[0086] ;

[0087] If the right lane line is not a white dashed line, the horizontal coordinate of the right limit position of the initial area Inward safety distance margin , get the horizontal coordinate of the right limit position of the area :

[0088] ;

[0089] If not, the horizontal coordinate of the right limit position of the initial area is Inward times the safety distance margin , get the horizontal coordinate of the right limit position of the area :

[0090] ;

[0091] Usually, the second coefficient Take 2;

[0092] In some embodiments, the vehicle boundary coordinates include a vehicle boundary horizontal coordinate and a vehicle boundary vertical coordinate; the region boundary coordinates include a region boundary horizontal coordinate and a region boundary vertical coordinate;

[0093] The vehicle boundary horizontal coordinate includes the vehicle left boundary horizontal coordinate and the horizontal coordinate of the right boundary of the vehicle , the vehicle boundary ordinate includes the vehicle front boundary ordinate and the ordinate of the rear boundary of the vehicle ;

[0094] The horizontal coordinate of the region boundary includes the horizontal coordinate of the left boundary of the region and the abscissa of the right border of the region , the region boundary ordinate includes the region front boundary ordinate and the ordinate of the rear boundary of the region ;

[0095] See also Figure 3 , based on the normal driving scenario, the steps to obtain the horizontal coordinates of the area boundary are:

[0096] Get the third coefficient , Detection distance margin and detection distance, the third coefficient , Detection distance margin The detection distance and detection distance are obtained by customization. The detection distance includes the left detection distance , Right detection distance , front detection distance And rear detection distance ;

[0097] Based on the lane line type, the detection distance margin is expanded outward according to the horizontal coordinate of the lane boundary Get the initial area boundary horizontal coordinate, or expand the third product outward according to the lane boundary horizontal coordinate to get the initial area boundary horizontal coordinate. The third product is the detection distance margin. With the third coefficient The product of

[0098] If the left lane line is a white dashed line, the horizontal coordinate of the left boundary of the lane Expand the custom detection distance margin , get the horizontal coordinate of the left border of the initial area :

[0099] ;

[0100] No, then the horizontal coordinate of the left boundary of the lane Expansion times the detection distance margin , get the horizontal coordinate of the left boundary of the initial area :

[0101] ;

[0102] If the right lane line is a white dashed line, the horizontal coordinate of the right lane boundary Outward expansion detection distance margin , get the horizontal coordinate of the right boundary of the initial area :

[0103] ;

[0104] No, then the horizontal coordinate of the right edge of the lane Expand outward times the detection distance margin , get the horizontal coordinate of the right boundary of the initial area :

[0105] ;

[0106] Usually, the third coefficient Take 2;

[0107] Obtaining the region boundary horizontal coordinate according to the initial region boundary horizontal coordinate, the vehicle boundary horizontal coordinate and the detection distance;

[0108] ;

[0109] ;

[0110] Based on the normal driving scenario, the steps to obtain the vertical coordinates of the area boundary are:

[0111] Get relative driving status and the fourth coefficient , the fourth coefficient Obtained through customization;

[0112] In some embodiments, the vehicle status information also includes current heading angle information; see Figure 4 , get the relative driving state The steps are:

[0113] Get the same direction detection threshold And the heading angle information of the obstacle; among them, the same direction detection threshold Obtained through custom methods;

[0114] Get the heading angle difference based on the current heading angle information and the obstacle's heading angle information ;

[0115] Compare the heading angle difference and the same-direction detection threshold to obtain the relative driving state :

[0116] like , then the relative driving state The obstacle is moving in the same direction as the unmanned vehicle. ;

[0117] like , then the relative driving state The obstacle is moving in the opposite direction of the unmanned vehicle. ;

[0118] No, then the relative driving state The obstacle is moving perpendicular to the unmanned vehicle. ;

[0119] Based on the relative driving state, the region boundary ordinate is obtained according to the vehicle boundary ordinate, or the region boundary ordinate is obtained according to the vehicle boundary ordinate and the detection distance, or the vehicle boundary ordinate, the fourth coefficient and the detection distance to obtain the vertical coordinate of the area boundary;

[0120] When the relative driving state The obstacle is moving in the same direction as the unmanned vehicle. hour:

[0121] ;

[0122] ;

[0123] When the relative driving state The obstacle is moving in the opposite direction of the unmanned vehicle. hour:

[0124] ;

[0125] ;

[0126] When the relative driving state The obstacle is moving perpendicular to the unmanned vehicle. hour:

[0127] ;

[0128] ;

[0129] Usually, when the relative driving state The obstacle is moving in the same direction as the unmanned vehicle. , or when the relative driving state The obstacle is moving in the opposite direction of the unmanned vehicle. hour Take 5, when the relative driving state The obstacle is moving perpendicular to the unmanned vehicle. hour Take 2;

[0130] In some embodiments, the driving scene also includes a preset driving scene. When the unmanned vehicle is in the preset driving scene, the region boundary coordinates need to be updated to obtain the region boundary coordinates in the corresponding scene, that is, when the unmanned vehicle is in the preset driving scene, the obstacle consideration area is determined according to the region boundary coordinates in the corresponding scene:

[0131] When the driverless car is in a left turn scenario at an intersection: it is necessary to update the horizontal coordinates of the left boundary of the area, the horizontal coordinates of the right boundary of the area, and the relative driving status The ordinate of the rear boundary of the area, the abscissa of the left extreme position of the area, and the abscissa of the right extreme position of the area when the obstacle is traveling in the same direction as the unmanned vehicle;

[0132] Get the scene adjustment margin , scene adjustment margin Obtained through custom methods;

[0133] According to the horizontal coordinate of the left border of the area Expand the scene adjustment margin outward Get the fifth horizontal coordinate , the fifth horizontal axis The fifth horizontal coordinate is the horizontal coordinate of the left boundary of the area updated in the left turn scene at the intersection. for:

[0134] ;

[0135] The horizontal coordinate of the right boundary of the area updated in the left turn scenario at the intersection for:

[0136] ;

[0137] Updated relative driving status for left turn scenarios at intersections is the ordinate of the rear boundary of the obstacle area when the unmanned vehicle is traveling in the same direction for:

[0138] ;

[0139] In the formula, is the seventh coefficient, usually Take 0.5;

[0140] In addition, the horizontal coordinate of the left limit position of the area updated in the left turn scene at the intersection for:

[0141] ;

[0142] The horizontal coordinate of the right limit position of the area updated in the left turn scenario at the intersection for:

[0143] ;

[0144] When the driverless car is in a straight-ahead scenario in the conflict zone: only the relative driving status needs to be updated is the ordinate of the rear boundary of the obstacle area relative to the unmanned vehicle when traveling in the same direction;

[0145] Updated relative driving status in the conflict zone straight driving scenario is the ordinate of the rear boundary of the obstacle area when the unmanned vehicle is traveling in the same direction for:

[0146] ;

[0147] In the formula, is the eighth coefficient, usually Take 0.5;

[0148] When the driverless car is in the gate scene: only the relative driving status needs to be updated is the ordinate of the rear boundary of the obstacle area relative to the unmanned vehicle when traveling in the same direction;

[0149] Updated relative driving status in the gate scenario is the ordinate of the rear boundary of the obstacle area when the unmanned vehicle is traveling in the same direction for:

[0150] ;

[0151] The preset driving scenarios can also be customized or new scenarios can be added according to actual needs. The scenes must be designed to be absolutely mutually exclusive and cannot affect each other.

[0152] In some embodiments, the state information of the suspected obstacle includes the obstacle boundary coordinates and the obstacle longitudinal speed. The obstacle boundary coordinates at least include the obstacle boundary ordinate, which includes the obstacle front boundary ordinate and the vertical coordinate of the boundary behind the obstacle It is worth noting that the rectangular frame based on obstacle recognition has the same direction as the front boundary of the unmanned vehicle as the front boundary of the obstacle, and the same direction as the rear boundary of the unmanned vehicle as the rear boundary of the obstacle, rather than the front and rear along the direction of the obstacle. The vehicle status information also includes the longitudinal speed of the vehicle. ;

[0153] See also Figure 5 , considering regions based on obstacles The steps to determine whether a suspected obstacle is an actual obstacle are:

[0154] Considering areas based on obstacles and obstacle boundary coordinates to determine whether the suspected obstacle is in the obstacle consideration area Inside: If yes, the suspected obstacle is an obstacle to be confirmed; specifically:

[0155] Get the first function and the second function ;

[0156] First function for:

[0157] ;

[0158] First function Used to determine whether the vertical coordinate of the obstacle boundary of the suspected obstacle is within the obstacle consideration area within the vertical coordinate of the region boundary;

[0159] Second function for:

[0160] ;

[0161] Second function Used to determine whether the obstacle boundary coordinate of the suspected obstacle is within the obstacle consideration area within the abscissa of the region boundary;

[0162] Call the first function ;

[0163] When the first function outputs a positive value, that is, the obstacle boundary ordinate of the suspected obstacle is in the obstacle consideration area Within the vertical coordinate of the region boundary, call the second function ;

[0164] When the second function When the output value is set, the horizontal coordinate of the obstacle boundary of the suspected obstacle is within the obstacle consideration area. The suspected obstacles within the abscissa of the area boundary are obstacles to be confirmed;

[0165] When the first function or the second function outputs a negative value, the judgment of the current suspected obstacle is completed, that is, the current suspected obstacle is an obstacle that does not need to be further considered. At this time, the state information of the new suspected obstacle is returned to judge the new suspected obstacle;

[0166] According to the obstacle boundary ordinate and obstacle longitudinal speed of the obstacle to be confirmed , the vehicle boundary ordinate and the vehicle longitudinal velocity Obtain the longitudinal relative position of the obstacle to be confirmed;

[0167] when When the obstacle is completely behind the unmanned vehicle,

[0168] when When the obstacle is completely in front of the unmanned vehicle,

[0169] If not, wait for confirmation that the obstacle is approaching the unmanned vehicle;

[0170] Based on the longitudinal relative position, or based on the longitudinal relative position and relative state information, it is determined whether there is a possibility of contact between the obstacle to be confirmed and the unmanned vehicle according to the state information of the obstacle to be confirmed and the vehicle state information: if so, the obstacle to be confirmed is an actual obstacle.

[0171] If the obstacle is completely behind the unmanned vehicle, , and it is confirmed that the obstacle is moving in the same direction as the unmanned vehicle:

[0172] If and only if When the obstacle is confirmed, it is possible to catch up with the unmanned vehicle;

[0173] Otherwise, the obstacle to be confirmed cannot catch up with the unmanned vehicle, that is, there is no possibility of contact. The current obstacle to be confirmed is an obstacle that does not need to be further considered, and the judgment of the current obstacle to be confirmed is completed;

[0174] when hour:

[0175] Get the first relative distance and the first relative velocity ;

[0176] The first relative distance for: ;

[0177] The first relative speed for: ;

[0178] According to the first relative distance and the first relative velocity Calculate the first relative time ;

[0179] First relative time for: ;

[0180] Get time threshold and distance threshold , time threshold and distance threshold All are obtained through customization;

[0181] when and When the obstacle to be confirmed cannot catch up with the unmanned vehicle, that is, there is no possibility of contact, the obstacle to be confirmed is an obstacle that does not need to be further considered, and the judgment of the obstacle to be confirmed is completed;

[0182] If no, the obstacle to be confirmed is located at the actual obstacle. At this time, the actual obstacle is located behind the unmanned vehicle;

[0183] If the obstacle is completely behind the unmanned vehicle, , and the obstacle to be confirmed is not traveling in the same direction as the unmanned vehicle, and there is no possibility that the obstacle to be confirmed and the unmanned vehicle will come into contact. The current obstacle to be confirmed is an obstacle that does not need to be further considered, and the judgment of the current obstacle to be confirmed is completed;

[0184] If the obstacle is completely in front of the unmanned vehicle , and it is confirmed that the obstacle is moving in the same direction as the unmanned vehicle:

[0185] when hour:

[0186] Get the second relative distance and the second relative speed ;

[0187] The second relative distance for: ;

[0188] The second relative speed for: ;

[0189] According to the second relative distance and the second relative speed Calculate the second relative time ;

[0190] The second relative time for: ;

[0191] when and When the unmanned vehicle cannot catch up with the obstacle to be confirmed, that is, there is no possibility of contact, the obstacle to be confirmed is an obstacle that does not need to be further considered, and the judgment of the obstacle to be confirmed is completed;

[0192] If no, the obstacle to be confirmed is an actual obstacle. At this time, the actual obstacle is located in front of the unmanned vehicle;

[0193] If the obstacle is completely in front of the unmanned vehicle , and the obstacle to be confirmed is not traveling in the same direction as the unmanned vehicle, and the unmanned vehicle has the possibility of contacting the obstacle to be confirmed, and the obstacle to be confirmed is the actual obstacle. At this time, the actual obstacle is located in front of the unmanned vehicle;

[0194] If an obstacle is confirmed to be approaching the unmanned vehicle:

[0195] If the obstacle to be confirmed is moving perpendicular to the unmanned vehicle:

[0196] When the obstacle type to be confirmed is a pedestrian or other slow moving object, since the obstacle consideration area has been The construction of the pedestrian and other types of confirmed obstacles has no possibility of contact with the unmanned vehicle. The current obstacle to be confirmed is an obstacle that does not need to be further considered, and the judgment of the current obstacle to be confirmed is completed;

[0197] No (i.e. the obstacle to be confirmed is close to the unmanned vehicle except for other situations where the obstacle type of the obstacle to be confirmed is a pedestrian or other slow moving object), then the unmanned vehicle and the obstacle to be confirmed are likely to be in contact, and the obstacle to be confirmed is an actual obstacle. At this time, the actual obstacle is located on the side of the unmanned vehicle;

[0198] In some embodiments, the obstacle boundary coordinates also include the obstacle boundary horizontal coordinates, and the obstacle boundary horizontal coordinates include the obstacle left boundary vertical coordinates. and the ordinate of the right boundary of the obstacle It is worth noting that based on the rectangular box of obstacle recognition, the boundary in the same direction as the left boundary of the unmanned vehicle is the left boundary of the obstacle, and the boundary in the same direction as the right boundary of the unmanned vehicle is the right boundary of the obstacle, while the left and right of the obstacle itself; the vehicle status information also includes its own horizontal coordinate , its own horizontal coordinate is the horizontal coordinate of the current driving position of the unmanned vehicle;

[0199] Obtain the horizontal relative position of the actual obstacle based on the horizontal coordinate of the obstacle boundary and its own horizontal coordinate , or obtain the horizontal relative position of the actual obstacle based on the horizontal coordinate of the obstacle boundary, the horizontal coordinate of the area boundary and the horizontal coordinate of the actual obstacle , that is, to judge the horizontal relative relationship between the actual obstacle and the unmanned vehicle;

[0200] like and , indicating that the actual obstacle occupies both the left and right spaces of the unmanned vehicle, then the horizontal relative position For left and right sides ;

[0201] like and , indicating that the actual obstacle occupies the space on the right side of the unmanned vehicle, then the lateral relative position For the right side ;

[0202] like and , indicating that the actual obstacle occupies the space on the right side of the unmanned vehicle, then the lateral relative position For the left .

[0203] In some of these embodiments, the avoidance decision includes a maintenance decision , Central Decision , Left-side avoidance decision and right avoidance decision In addition, in the process of obtaining the avoidance decision, that is, in the decision-making process, the first transient decision and the second transient decision will also be called.

[0204] Keep the decision horizontal axis of the decision as the second horizontal axis , the second horizontal axis is:

[0205] ;

[0206] In the formula, is a function that limits the range of values, The three values ​​in the function represent from left to right: the value to be limited (value), the lower limit of the range (min_value) and the upper limit of the range (max_value): if the value is less than the lower limit, the lower limit value is returned; if the value is greater than the upper limit, the upper limit value is returned; otherwise, the original value is returned. Limit to the lower limit of the range, that is, the horizontal coordinate of the right limit position of the area The upper limit of the range is the horizontal coordinate of the left limit position of the area between;

[0207] The decision abscissa of the central decision is the horizontal reference position abscissa ;

[0208] The abscissa of the first transient decision is the third abscissa , the third horizontal axis is :

[0209] ;

[0210] In the formula, is the fifth coefficient, usually, Take 2;

[0211] The decision horizontal coordinate of the left avoidance decision is composed of the second horizontal coordinate , horizontal reference position abscissa and the third horizontal axis Joint decision making;

[0212] The abscissa of the second transient decision is the fourth abscissa , the fourth horizontal coordinate is :

[0213] ;

[0214] In the formula, is the fifth coefficient, usually, Take 2;

[0215] The decision abscissa of the right avoidance decision is composed of the second abscissa , horizontal reference position abscissa and the fourth horizontal coordinate Joint decision making;

[0216] All actual obstacles in each frame planning form an actual obstacle set, and the number of actual obstacles in the actual obstacle set is obtained, which is calculated as ; See Figure 6 , traverse each actual obstacle in the actual obstacle set, the traversal steps are;

[0217] Step S1: Initialize the current static count , let the current static count be zero, that is ;

[0218] Step S2: Based on the third function , the third distance is obtained based on the obstacle boundary coordinates of the actual obstacle and the vehicle boundary coordinates , and the current static count is increased by 1, that is, ;

[0219] The third function That is to calculate the third distance function; specifically: obtaining the Cartesian outer rectangle of the obstacle and the geometric center point of the obstacle according to the obstacle boundary coordinates of the actual obstacle, obtaining the Cartesian outer rectangle of the vehicle and the geometric center point of the vehicle according to the vehicle boundary coordinates, connecting the geometric center point of the obstacle and the geometric center point of the vehicle to obtain a center line, the center line and the Cartesian outer rectangle of the obstacle generate a first intersection point, and the center line and the Cartesian outer rectangle of the vehicle generate a second intersection point; the third distance is the distance between the first intersection point and the second intersection point; it can be calculated based on geometric principles, and the calculation process is prior art and will not be repeated here;

[0220] Step S3: Determination Is it established:

[0221] If yes, repeat steps S2 to S3, and add the actual obstacle to the list of obstacles to be decided;

[0222] Otherwise, a set of obstacles to be decided is obtained according to the row of obstacles to be decided; the set of obstacles to be decided is a collection of obstacles to be decided;

[0223] See also Figure 7 , when the actual obstacle set is empty: at this time, the current static count Not initialized yet, that is, the current static count The current static count of the history obtained after the previous frame traversal ends ;

[0224] Based on historical current static count Calculate the current static count :

[0225] ;

[0226] like , then the central decision is executed;

[0227] Otherwise, the hold decision is executed;

[0228] When the actual obstacle set is not empty: After traversing the actual obstacle set to obtain the obstacle set to be decided, according to the third distance Sort the values ​​from small to large and take the third distance The obstacle to be decided corresponding to the minimum value is the final decision obstacle;

[0229] Get the close distance threshold ; Distance threshold Obtained through customization;

[0230] Final decision based on obstacle and distance threshold , according to the fourth function Get the output result of the fourth function; the fourth function The calculation process is:

[0231] The third distance to determine the final decision obstacle Is the distance less than the threshold met? , that is, to determine whether the final decision obstacle is near the unmanned vehicle, and to determine the lateral relative position of the final decision obstacle is satisfied or , determine the lateral relative position of the final decision obstacle This is so that subsequent decisions can cover all possible situations that may hinder the final decision;

[0232] When the output result is no, it indicates an error;

[0233] When the output result is yes, there are four cases:

[0234] Case 1: If and , then the final decision obstacle's lateral relative position For left and right sides ;

[0235] When the lane line types of the left lane line and the right lane line are both white dotted lines, a keep decision is executed;

[0236] When only the left lane line is a white dashed line: If the second horizontal coordinate If it is greater than 0, it means that the driverless car is on the left side of the lane center, and the center decision is executed; otherwise, the hold decision is executed;

[0237] When only the lane line type of the right lane line is a white dashed line: If the second horizontal coordinate If it is less than 0, it means that the driverless car is on the right side of the lane center, and the hold decision is executed; otherwise, the center decision is executed;

[0238] When the lane line types of the left lane line and the right lane line are not white dashed lines, the central decision is executed;

[0239] Case 2: If and , then the final decision obstacle's lateral relative position For the left ;

[0240] At this time, the abscissa of the second transient decision is called, that is, the fourth abscissa , compared with the second horizontal axis , horizontal reference position abscissa and the fourth horizontal coordinate , the decision corresponding to the smallest horizontal coordinate value among the three is output as the right avoidance decision and executed ;

[0241] Case 3: If and , then the final decision obstacle's lateral relative position For the right side ;

[0242] At this time, the abscissa of the first transient decision is called, that is, the third abscissa , compared with the second horizontal axis , horizontal reference position abscissa and the third horizontal axis , the decision corresponding to the largest horizontal coordinate value among the three is output as the left avoidance decision and executed ;

[0243] Case 4: If and and and , it means that the unmanned vehicle is near the left side and right side There are dynamic obstacles on both sides; the unmanned vehicle is encouraged to maintain its current lateral position without making left or right avoidance movements, that is, to execute a hold decision.

[0244] That is, when the decision of the current frame planning is completed, the decision output of the nearest actual obstacle is used as the current final lateral avoidance decision;

[0245] The horizontal coordinate of the current final lateral avoidance decision is complementary filtered with the horizontal coordinate of the historical global lateral avoidance decision planned in the previous frame, and the filtered result is assigned to the current global lateral avoidance decision planned in the current frame to provide a reference for lateral planning. At the same time, the current global lateral avoidance decision of the current frame is used as the historical global lateral avoidance decision planned in the next frame.

[0246] In some of the embodiments, the left side avoidance decision includes a left side lateral avoidance decision and a left side lateral and longitudinal comprehensive avoidance decision; the right side avoidance decision includes a right side lateral avoidance decision and a right side lateral and longitudinal comprehensive avoidance decision.

[0247] Get the speed change threshold , speed change threshold Obtained through custom methods;

[0248] When calculating the horizontal coordinate of the current final lateral avoidance decision and its own horizontal coordinate The distance between

[0249] If the right side avoidance decision is executed, the horizontal coordinate of the current final lateral avoidance decision and its own horizontal coordinate The distance between is the fourth relative distance; compare the fourth relative distance with the speed change threshold :When the fourth relative distance is less than the speed change threshold When the right-side lateral and longitudinal comprehensive avoidance decision is executed, the right-side lateral avoidance decision is executed; otherwise, the right-side lateral avoidance decision is executed, that is, the longitudinal decision is not processed;

[0250] If the left avoidance decision is executed, the horizontal coordinate of the current final lateral avoidance decision and its own horizontal coordinate The distance between is the fifth relative distance; compare the fifth relative distance with the speed change threshold :When the fifth relative distance is less than the speed change threshold When the vehicle is in the opposite direction, the left-side lateral and longitudinal comprehensive avoidance decision is executed; otherwise, the left-side lateral avoidance decision is executed, that is, the longitudinal decision is not processed.

[0251] This embodiment also provides an unmanned vehicle system, including an unmanned vehicle and a dynamic avoidance module, wherein the dynamic avoidance module is used to implement a lateral dynamic avoidance decision method for the unmanned vehicle.

[0252] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.

Claims

1. A lateral dynamic avoidance decision-making method for an unmanned vehicle, characterized by: The following steps are involved: Acquire driving scenes, lane information, horizontal coordinates of planning space, horizontal reference positions and vehicle status information of the unmanned vehicle; the vehicle status information includes vehicle boundary coordinates and vehicle width; the lane information includes lane boundary horizontal coordinates and lane line type; the driving scenes include normal driving scenes and preset driving scenes; Based on the driving scenario, obtaining an obstacle consideration area according to the vehicle state information and the lane information, or obtaining an obstacle consideration area according to the lane information, the horizontal coordinate of the planning space and the vehicle state information; The obstacle consideration area includes a region limit position horizontal coordinate and a region boundary coordinate; the region limit position horizontal coordinate is the horizontal coordinate of the driving limit position of the unmanned vehicle in the obstacle consideration area; the region boundary coordinate is the coordinate used to determine the boundary of the obstacle consideration area; Obtain status information of suspected obstacles; Determine whether the suspected obstacle is an actual obstacle based on the obstacle consideration area: if so, obtain relative state information of the actual obstacle according to the vehicle state information and the state information of the actual obstacle; the relative state information includes relative driving state and relative position; Obtaining an avoidance decision based on the vehicle state information, the lane information, the lateral reference position, and the relative state information of the actual obstacle; The unmanned vehicle performs dynamic avoidance based on the avoidance decision.

2. The unmanned vehicle lateral dynamic avoidance decision-making method according to claim 1, characterized in that: Based on the normal driving scenario, the steps of obtaining the horizontal coordinate of the area limit position are: Obtaining the first coefficient, the second coefficient and the safety distance margin; A first abscissa is obtained by multiplying the lane boundary abscissa inward by a first product, where the first product is the product of the vehicle width and the first coefficient; Obtaining the initial area limit position horizontal coordinate according to the first horizontal coordinate, or obtaining the initial area limit position horizontal coordinate according to the first horizontal coordinate and the planned space horizontal coordinate; Based on the lane line type, the horizontal coordinate of the area limit position is obtained by retracting the safety distance margin inward from the horizontal coordinate of the initial area limit position, or by multiplying the horizontal coordinate of the initial area limit position inward by a second product, wherein the second product is the product of the safety distance margin and the second coefficient.

3. The unmanned vehicle lateral dynamic avoidance decision-making method according to claim 1, characterized in that: The vehicle boundary coordinates include a vehicle boundary horizontal coordinate and a vehicle boundary vertical coordinate; the area boundary coordinates include an area boundary horizontal coordinate and an area boundary vertical coordinate; Based on the normal driving scenario, the steps of obtaining the horizontal coordinate of the area boundary are: Obtaining a third coefficient, a detection distance margin, and a detection distance; Based on the lane line type, the detection distance margin is expanded outward according to the lane boundary horizontal coordinate to obtain the initial area boundary horizontal coordinate, or the lane boundary horizontal coordinate is expanded outward according to a third product to obtain the initial area boundary horizontal coordinate, where the third product is the product of the detection distance margin and the third coefficient; Obtaining the region boundary horizontal coordinate according to the initial region boundary horizontal coordinate, the vehicle boundary horizontal coordinate and the detection distance; Based on the normal driving scenario, the steps of obtaining the vertical coordinate of the area boundary are: obtaining a relative driving state and a fourth coefficient; Based on the relative driving state, the area boundary ordinate is obtained according to the vehicle boundary ordinate, or according to the vehicle boundary ordinate and the detection distance, or according to the vehicle boundary ordinate, the fourth coefficient and the detection distance.

4. The unmanned vehicle lateral dynamic avoidance decision-making method according to claim 3, characterized in that: The vehicle state information also includes current heading angle information; the steps of obtaining the relative driving state are: Obtain the same-direction detection threshold and the heading angle information of the obstacle; The obtaining of a heading angle difference according to the current heading angle information and the heading angle information of the obstacle; The relative driving state is obtained by comparing the heading angle difference with the same-direction detection threshold.

5. The unmanned vehicle lateral dynamic avoidance decision-making method according to claim 3, characterized in that: The state information of the suspected obstacle includes obstacle boundary coordinates and obstacle longitudinal speed, and the obstacle boundary coordinates at least include obstacle boundary longitudinal coordinates; The vehicle state information also includes the longitudinal speed of the vehicle; the relative position at least includes the longitudinal relative position; The steps of judging whether the suspected obstacle is an actual obstacle based on the obstacle consideration area are as follows: Determine whether the suspected obstacle is within the obstacle consideration area according to the obstacle consideration area and the obstacle boundary coordinates: if so, the suspected obstacle is an obstacle to be confirmed; Obtaining a longitudinal relative position of the obstacle to be confirmed according to the obstacle boundary ordinate, the obstacle longitudinal speed, the vehicle boundary ordinate and the vehicle longitudinal speed of the obstacle to be confirmed; Based on the longitudinal relative position, or based on the longitudinal relative position and the relative state information, it is determined whether there is a possibility of contact between the obstacle to be confirmed and the unmanned vehicle according to the state information of the obstacle to be confirmed and the vehicle state information: if so, the obstacle to be confirmed is the actual obstacle.

6. The unmanned vehicle lateral dynamic avoidance decision-making method according to claim 5, characterized in that: The obstacle boundary coordinates also include the obstacle boundary horizontal coordinates; the vehicle state information also includes its own horizontal coordinates; the relative position also includes the lateral relative position; The lateral relative position of the actual obstacle is obtained according to the obstacle boundary lateral coordinate of the actual obstacle and the own lateral coordinate, or the lateral relative position of the actual obstacle is obtained according to the obstacle boundary lateral coordinate of the actual obstacle, the area boundary lateral coordinate and the own lateral coordinate.

7. The unmanned vehicle lateral dynamic avoidance decision-making method according to claim 6, characterized in that: The avoidance decision includes a hold decision, a center decision, a left avoidance decision and a right avoidance decision.

8. The unmanned vehicle lateral dynamic avoidance decision-making method according to claim 7, characterized in that: The left side avoidance decision includes a left side lateral avoidance decision and a left side lateral and longitudinal comprehensive avoidance decision; the right side avoidance decision includes a right side lateral avoidance decision and a right side lateral and longitudinal comprehensive avoidance decision.

9. An unmanned vehicle system, comprising an unmanned vehicle and a dynamic avoidance module, characterized in that: The dynamic avoidance module is used to implement the unmanned vehicle lateral dynamic avoidance decision method described in any one of claims 1-8.

Images