Fusion method of vehicle end sensing result and field end sensing result

Abstract

The invention belongs to the technical field of automatic driving perception fusion, and discloses a method for fusing a vehicle end perception result and a field end perception result, and the method comprises the following steps: A1, a vehicle end calculation unit receives a field end perception result Ps and a perception result Pv transmitted by a vehicle end perception module at the same time; a2, the vehicle end calculation unit receives and analyzes the absolute positioning coordinates transmitted by the positioning module; a3, respectively calculating an absolute difference value Di between each element in Ps and P0; a4, querying to find the minimum value of Di, and recording the serial number as io; a5, taking the target as the vehicle, taking the coordinates of other elements in Ps as the original point and the positive direction of the y axis, and converting the coordinates into a vehicle coordinate system to obtain a converted set Psv; and A6, respectively calculating the absolute distance from the coordinate value in each element in the sets Pv and Psv to the original point, removing the elements with the distance from the original point being less than Dn in the Psv set, removing the elements with the distance from the original point being greater than Df in the Pv set, and obtaining updated Pv and Psv.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving perception fusion technology, and more specifically, to a method for fusing vehicle-side perception results with field-side perception results. Background Art

[0002] At present, the perception algorithm of the autonomous driving system mainly obtains data from sensors deployed on the vehicle side for calculation, thereby obtaining the perception results in the vehicle coordinate system, and using this perception result as input for subsequent decision-making and control. However, based on China's vehicle-road-cloud integration technology route, with the deployment of a large number of roadside perception devices, roadside perception devices can provide high-precision global perception results. When the autonomous driving vehicle can obtain both vehicle-side perception results and field-side perception results at the same time, how to make good use of these two perception results becomes the most critical issue.

[0003] There are currently two processing methods, one is to use only the perception results of the vehicle side, and the other is to use only the perception results of the field side. These two methods are the simplest, but they cannot combine the advantages of vehicle-side perception and field-side perception. A method that can integrate vehicle-side perception and field-side perception is needed to fully utilize the advantages of both, make up for the disadvantages of both, and achieve more stable and high-precision perception.

[0004] Advantages of vehicle-side perception: Vehicle-side perception has high perception accuracy for targets close to the vehicle, and has strong perception and resolution capabilities for extremely small targets near the vehicle;

[0005] Disadvantages of vehicle-side perception: The vehicle-side has poor perception ability for distant targets or targets partially blocked by other objects, with low accuracy and poor stability. The perception results often jitter.

[0006] Advantages of field-side perception: The field-side perception range is much longer than that of the vehicle-side, and can provide perception results across the entire field, and the perception accuracy of each target remains basically consistent;

[0007] Disadvantages of field-side perception: The field side can only provide the absolute positions of all perceived targets at the same time, and cannot provide the perception results in the vehicle coordinate system for a specific vehicle; the field side has poor perception capabilities for small objects near the vehicle and is basically unable to distinguish them.

[0008] To this end, the present application proposes a method for fusing vehicle-side perception results with field-side perception results to solve the above-mentioned problems. Summary of the invention

[0009] In order to solve the above problems, the present application provides a method for fusing vehicle-side perception results with field-side perception results, so that the perception results have a wider range, higher accuracy and stronger stability.

[0010] The present application provides a method for fusing vehicle-side perception results with field-side perception results using the following technical solution:

[0011] A method for fusing vehicle-side perception results with field-side perception results, the method comprising the following steps:

[0012] A1. The vehicle-side computing unit simultaneously receives the field-side sensing result Ps and the sensing result Pv from the vehicle-side sensing module;

[0013] A2. The vehicle-side computing unit receives and analyzes the absolute positioning coordinates transmitted by the positioning module;

[0014] A3. Calculate the absolute difference Di between each element in Ps and P0 respectively;

[0015] A4. Find the minimum value of Di and record the sequence number as io;

[0016] A5. Take the target (Xsio, Ysio, Hsio) as the vehicle, and transform the coordinates of other elements in Ps to the vehicle coordinate system with (Xsio, Ysio, Hsio) as the origin and the positive direction of the y-axis. The transformed set is Psv, and the formula of Psc is as follows:

[0017] Psv={(Xsv1,Ysv1,Hsv1),(Xsv2,Ysv2,Hsv2),…,(Xsvm,Ysvm,Hsvm)};

[0018] The three values ​​in the element are X for the x-axis coordinate, Y for the y-axis coordinate, and H for the heading angle;

[0019] A6. Calculate the absolute distance from the coordinate value of each element in the sets Pv and Psv to the origin respectively, remove the elements in the Psv set whose distance from the origin is less than Dn, and remove the elements in the Pv set whose distance from the origin is greater than Df, to obtain the updated Pv and Psv;

[0020] A7. Calculate the intersection-over-union (IOU) ratios between the elements in the sets Pv and Psv whose distances from the origin are between Dn and Df respectively.

[0021] A8. Combine the elements in Pv whose distance from the origin is less than Dn, the elements in Psv whose distance from the origin is greater than Df, all calculated (Xri, Yri, Hri) elements, and elements whose IOU values ​​are less than 0.5 into a new element set Pr. Pr is the set of all targets after the vehicle-side perception results are fused with the field-side perception results.

[0022] Furthermore, Ps and Pv are vector sets composed of coordinate values ​​of the sensed targets, the elements in Ps are in the absolute coordinate system, and the elements in Pv are in the vehicle coordinate system, as follows:

[0023] Ps={(Xs1,Ys1,Hs1),(Xs2,Ys2,Hs2),…,(Xsm,Ysm,Hsm)}m is a positive integer;

[0024] Pv={(Xv1, Yv1, Hv1),(Xv2, Yv2, Hv2),…,(Xv, Yvn, Hvn)}n is a positive integer;

[0025] The three values ​​in the element are X for the x-axis coordinate, Y for the y-axis coordinate, and H for the heading angle.

[0026] Through the above technical solution, the coordinate systems of Ps and Pv are clarified, which helps to understand the source and applicability of the data; the coordinate values ​​of the perceived targets are structured in the form of vector sets, which is convenient for processing and analysis; and m and n are used as positive integer parameters, which provides a flexible data representation method to adapt to different numbers of target objects.

[0027] Furthermore, in step A7, element pairs with an intersection over union (IOU) greater than or equal to 0.5 are considered to be the same target object, and element pairs with an IOU less than 0.5 are considered to be two different objects.

[0028] Through the above technical solution, the target objects can be identified and distinguished more accurately through the calculation of IOU, thereby improving the overall perception accuracy; by using a clear threshold to judge the identity of the target objects, the decision-making logic is simplified, the operability of the system is improved, and it helps to reduce misjudgments.

[0029] Furthermore, for element pairs with IOU greater than or equal to 0.5, the calculation is performed according to the following formula:

[0030]

[0031]

[0032] Get the new coordinate value (Xri, Yri, Hri) of the object, where i is a positive integer.

[0033] Furthermore, in step A6, Dn and Df are empirical values ​​obtained through actual testing.

[0034] In summary, the present application includes at least one of the following beneficial technical effects:

[0035] (1) The perception range is wider than that of a single vehicle, and the perception distance covers the entire intersection;

[0036] (2) Long-distance perception results are more accurate. Since the field-side perception is integrated, the final fusion result also has high accuracy for the perceived targets that are far away from the vehicle.

[0037] (3) The perception blind area is smaller than that of a single vehicle. Due to the integration of field-side perception, the perception results can also output high-precision coordinates for objects in the blind area of ​​the vehicle-side sensor. DETAILED DESCRIPTION

[0038] The technical solutions in the embodiments of the present application will be described clearly and completely below; obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments, and all other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application without making creative work are within the scope of protection of the present application.

[0039] In the description of the present application, it should be noted that the terms "upper", "lower", "inner", "outer", "top / bottom" and the like indicating directions or positional relationships are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present application. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0040] In the description of this application, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "provided with", "set / connected", "connected", etc. should be understood in a broad sense. For example, "connected" can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0041] Example:

[0042] The present application is described in further detail below.

[0043] The embodiment of the present application discloses a method for fusing vehicle-side perception results with field-side perception results, characterized in that the method comprises the following steps:

[0044] A1. The vehicle-side computing unit simultaneously receives the field-side sensing result Ps and the sensing result Pv from the vehicle-side sensing module;

[0045] A2. The vehicle-side computing unit receives and analyzes the absolute positioning coordinates transmitted by the positioning module;

[0046] A3. Calculate the absolute difference Di between each element in Ps and P0 respectively;

[0047] A4. Find the minimum value of Di and record the sequence number as io;

[0048] A5. Take the target (Xsio, Ysio, Hsio) as the vehicle, and transform the coordinates of other elements in Ps to the vehicle coordinate system with (Xsio, Ysio, Hsio) as the origin and the positive direction of the y-axis. The transformed set is Psv, and the formula of Psc is as follows:

[0049] Psv={(Xsv1,Ysv1,Hsv1),(Xsv2,Ysv2,Hsv2),…,(Xsvm,Ysvm,Hsvm)};

[0050] The three values ​​in the element are X for the x-axis coordinate, Y for the y-axis coordinate, and H for the heading angle;

[0051] A6. Calculate the absolute distance from the coordinate value of each element in the sets Pv and Psv to the origin respectively, remove the elements in the Psv set whose distance from the origin is less than Dn, and remove the elements in the Pv set whose distance from the origin is greater than Df, to obtain the updated Pv and Psv;

[0052] A7. Calculate the intersection-over-union (IOU) ratios between the elements in the sets Pv and Psv whose distances from the origin are between Dn and Df respectively.

[0053] A8. Combine the elements in Pv whose distance from the origin is less than Dn, the elements in Psv whose distance from the origin is greater than Df, all calculated (Xri, Yri, Hri) elements, and elements whose IOU values ​​are less than 0.5 into a new element set Pr. Pr is the set of all targets after the vehicle-side perception results are fused with the field-side perception results.

[0054] Ps and Pv are vector sets composed of the coordinate values ​​of the perceived target. The elements in Ps are in the absolute coordinate system, and the elements in Pv are in the vehicle coordinate system, as follows:

[0055] Ps={(Xs1,Ys1,Hs1),(Xs2,Ys2,Hs2),…,(Xsm,Ysm,Hsm)}m is a positive integer;

[0056] Pv={(Xv1, Yv1, Hv1),(Xv2, Yv2, Hv2),…,(Xv, Yvn, Hvn)}n is a positive integer;

[0057] The three values ​​in the element are X for the x-axis coordinate, Y for the y-axis coordinate and H for the heading angle; the coordinate systems of Ps and Pv are clarified, which helps to understand the source and applicability of the data; the coordinate values ​​of the perceived target are structured in the form of a vector set, which is easy to process and analyze; m and n are used as positive integer parameters, which provides a flexible data representation method to adapt to different numbers of target objects.

[0058] In step A7, element pairs with an intersection over union (IOU) greater than or equal to 0.5 are considered to be the same target object, and element pairs with an IOU less than 0.5 are considered to be two different objects. By calculating the IOU, the target objects can be identified and distinguished more accurately, thereby improving the overall perception accuracy. By using a clear threshold to determine the identity of the target objects, the decision logic is simplified, the operability of the system is improved, and it helps to reduce misjudgments.

[0059] For element pairs with IOU greater than or equal to 0.5, the calculation is performed according to the following formula:

[0060]

[0061] Get the new coordinate value (Xri, Yri, Hri) of the object, where i is a positive integer.

[0062] In step A6, Dn and Df are empirical values ​​obtained through actual tests.

[0063] The implementation principle of a method for fusing vehicle-side perception results with field-side perception results in an embodiment of the present application is as follows:

[0064] The present invention utilizes and integrates the vehicle-side perception results and the field-side perception results at the same time, giving full play to the advantages of vehicle-side perception and field-side perception, and makes up for the defects of using only one end perception result. It mainly has the following advantages:

[0065] The sensing range is wider than that of a single vehicle, and the sensing distance covers the entire intersection;

[0066] The long-distance perception results are more accurate. Since the field-side perception is integrated, the final fusion result also has high accuracy for the perception targets far away from the vehicle.

[0067] The perception blind spot is smaller than that of a single vehicle. Due to the integration of field-side perception, the perception results can also output high-precision coordinates for objects in the blind spot of the vehicle-side sensor.

[0068] The above are all preferred embodiments of the present application, and the protection scope of the present application is not limited thereto. Therefore, any equivalent changes made according to the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for fusing vehicle-side perception results with field-side perception results, characterized in that: The method comprises the following steps: A1. The vehicle-side computing unit simultaneously receives the field-side sensing result Ps and the sensing result Pv from the vehicle-side sensing module; A2. The vehicle-side computing unit receives and analyzes the absolute positioning coordinates transmitted by the positioning module; A3. Calculate the absolute difference Di between each element in Ps and P0 respectively; A4. Find the minimum value of Di and record the sequence number as io; A5. Take the target (Xsio, Ysio, Hsio) as the vehicle, and transform the coordinates of other elements in Ps to the vehicle coordinate system with (Xsio, Ysio, Hsio) as the origin and the positive direction of the y-axis. The transformed set is Psv, and the formula of Psc is as follows: Psv={(Xsv1,Ysv1,Hsv1),(Xsv2,Ysv2,Hsv2),…,(Xsvm,Ysvm,Hsvm)}; The three values ​​in the element are X for the x-axis coordinate, Y for the y-axis coordinate, and H for the heading angle; A6. Calculate the absolute distance from the coordinate value of each element in the sets Pv and Psv to the origin respectively, remove the elements in the Psv set whose distance from the origin is less than Dn, and remove the elements in the Pv set whose distance from the origin is greater than Df, to obtain the updated Pv and Psv; A7. Calculate the intersection-over-union (IOU) ratios between the elements in the sets Pv and Psv whose distances from the origin are between Dn and Df respectively. A8. Combine the elements in Pv whose distance from the origin is less than Dn, the elements in Psv whose distance from the origin is greater than Df, all calculated (Xri, Yri, Hri) elements, and elements whose IOU values ​​are less than 0.5 into a new element set Pr. Pr is the set of all targets after the vehicle-side perception results are fused with the field-side perception results.

2. The method for fusing vehicle-side perception results with field-side perception results according to claim 1, characterized in that: Ps and Pv are vector sets composed of the coordinate values ​​of the perceived target. The elements in Ps are in the absolute coordinate system, and the elements in Pv are in the vehicle coordinate system, as follows: Ps={(Xs1,Ys1,Hs1),(Xs2,Ys2,Hs2),…,(Xsm,Ysm,Hsm)}m is a positive integer; Pv={(Xv1, Yv1, Hv1),(Xv2, Yv2, Hv2),…,(Xv, Yvn, Hvn)}n is a positive integer; The three values ​​in the element are X for the x-axis coordinate, Y for the y-axis coordinate, and H for the heading angle.

3. The method for fusing vehicle-side perception results with field-side perception results according to claim 1, characterized in that: In step A7, element pairs with an intersection-over-union ratio (IOU) greater than or equal to 0.5 are considered to be the same target object, and element pairs with an IOU less than 0.5 are considered to be two different objects.

4. The method for fusing vehicle-side perception results with field-side perception results according to claim 3, characterized in that: For element pairs with IOU greater than or equal to 0.5, the calculation is performed according to the following formula: Get the new coordinate value (Xri, Yri, Hri) of the object, where i is a positive integer.

5. The method for fusing vehicle-side perception results with field-side perception results according to claim 1, characterized in that: In step A6, Dn and Df are empirical values ​​obtained through actual tests.

Images