Method for geometric representation of vehicle region of a vehicle for collision detection

Abstract

A method for geometrically representing a vehicle region (1) of a vehicle for collision detection, wherein the vehicle region (1) has a boundary (2), the method comprising the following method steps: performing a medial axis transformation on the vehicle region (1) to generate a vehicle region skeleton (4); performing point classification on points of the vehicle region skeleton (4) to determine front corner region points (5, 6) and rear corner region points (7, 8), as well as front wheelbase points (9) and rear wheelbase points (10); and also performing a circle decomposition on the vehicle region (1), wherein each circle of the circle decomposition has a maximum area excess value (17).

Description

Technical Field

[0001] The invention relates to a method for geometric representation of a vehicle region of a vehicle for collision detection, wherein the vehicle region has a boundary. Background Art

[0002] The geometric representation of the vehicle region is known from the prior art. The vehicle region is surrounded by a rectangular bounding box, and the bounding box is decomposed and covered by a fixed number of circles with a fixed radius. This type of method can be used in Figure 1 found in the prior art.

[0003] This type of decomposition therefore results in a large overflow, such that during collision detection ("collision estimation" or "collision detection"), regions of the occupancy map produced by the decomposition appear to be occupied but are not actually occupied.

[0004] This is disadvantageous in the case of parking assistance systems, for example.

[0005] It is therefore an object of the present invention to specify a method which overcomes the disadvantages of the prior art and provides improved collision detection.

[0006] This object is achieved by a method having the features of claim 1 . Summary of the invention

[0007] The core idea of ​​the present invention is to provide a method for geometric representation of a vehicle region of a vehicle for collision detection, wherein the vehicle region has a boundary, the method comprising the following steps:

[0008] (a) performing a medial axis transformation on the vehicle region to generate a vehicle region skeleton, and performing point classification on points of the vehicle region skeleton to determine front corner region points and rear corner region points, as well as front wheelbase points and rear wheelbase points;

[0009] (b) performing circle decomposition on the vehicle area, wherein each circle of the circle decomposition has a maximum area excess value, including:

[0010] (i)(I) generating a first circle, wherein the center of the first circle is the front wheelbase point and the first circle has a first radius, and generating a second circle, wherein the center of the second circle is the rear wheelbase point and the second circle has the first radius; wherein the connection between the front wheelbase point and the rear wheelbase point corresponds to the first skeleton connection line;

[0011] (II) performing a verification step to determine whether the area of ​​the vehicle area assigned to the first skeleton connection line is completely covered, wherein the rear wheelbase point and the front wheelbase point serve as endpoints;

[0012] (III) if the result of the verification step is no, generating an nth circle with a first radius on the first skeleton connection line, where n=3, ...; incrementing n, and executing step (II) again;

[0013] (ii) Along the front wheelbase point and the nth v The nth point between the front corner area points v The front frame connection line is from the front wheelbase point to the nth v The front corner area starts at point n v =1,2,…;

[0014] (I) in the nth v Generate the nth with the smallest radius on the front skeleton connection line v front circle, and iteratively adjust the nth v The minimum radius of the front circle until the maximum area exceeds the value condition;

[0015] (II) performs a verification step to determine the assignment to the nth v Whether the area of ​​the vehicle area where the front frame is connected is completely covered;

[0016] (III) If the result of the verification step is no, then in the nth v Generate the first (n) with the smallest radius on the front skeleton connection line v +1) previous circle, and iteratively adjust the (nth v +1) The minimum radius of the front circle, until the maximum area exceeds the value condition, and increases by n v , and execute step (II) again;

[0017] (iii) Along the rear wheelbase point and the nth h The nth point between the points in the back corner area h The rear frame connection line extends from the rear wheelbase point to the nth h The starting point of the back corner area, n h =1,2,…;

[0018] (I) Generate the nth h After the circle and iteratively adjust the nth h The minimum radius of the back circle until the maximum area exceeds the value condition;

[0019] (II) performs a verification step to determine the assignment to the nth h Whether the area of ​​the vehicle area where the rear frame is connected is completely covered;

[0020] (III) If the result of the verification step is negative, then generate the (nth)th skeleton connection line with the minimum radius. h +1) after the circle, and iteratively adjust the (n h+1) The minimum radius of the circle after the maximum area exceeds the value condition, and increases by n h , and execute step (II) again.

[0021] The term "vehicle area" refers to the area occupied by the vehicle when the vehicle is viewed from above. According to the present invention, the vehicle area has a border or edge surrounding the vehicle area.

[0022] A minimum radius can be pre-specified, the minimum radius being the smallest possible radius of the corresponding circle. No radius smaller than the minimum radius is provided. In this case, the adjustment of the minimum radius is an increase in the radius value, for example incrementally, until the area exceeding the value condition is met. Preferably, it is provided that the radius is less than or equal to the first radius, which means: minimum radius ≤ radius ≤ first radius.

[0023] Methods of medial axis transformation are well known from the prior art.Performing a medial axis transformation results in a skeleton, in this case a vehicle region skeleton according to the present invention.

[0024] A vehicle area skeleton is formed by a plurality of points, wherein, according to the present invention, a certain proportion of the points are classified by performing point classification according to the present invention, wherein the points of the vehicle area skeleton are classified into front corner area points, rear corner area points, and front wheelbase points and rear wheelbase points through point classification.

[0025] According to a preferred embodiment, point classification is performed from a pre-specified vehicle center so that the point with the largest distance from the vehicle center is defined as a corner area point, and the point of the vehicle center that constitutes the Y node or the vehicle center that is arranged close to the Y node is defined as a wheelbase point.

[0026] From the coordinates (x c ,y c The term “vehicle center” denoted by ) is used to denote the average value of the skeleton points, where the set of skeleton points P s (x s ,y s ) is represented by the following formula:

[0027]

[0028] Among them, x s，max is the maximum value of the x coordinate and x s，min is the minimum value of the x coordinate, y s，max is the maximum value of the y coordinate and y s，min is the minimum y-coordinate.

[0029] The term "Y node" is used to refer to the point where the lines of the Y intersect.

[0030] Furthermore, further points corresponding to side corner area points may be classified, which correspond to the positions of the vehicle's rearview mirrors. Side corner area points are those points which are at a certain distance from the vehicle center and do not correspond to any other corner area points.

[0031] The term "area excess value" is used to indicate that a circle covering a part of the vehicle area has an area portion that does not cover any part of the vehicle area. This area portion corresponds to the area excess value which, according to the invention, is allowed to assume a maximum value. The area portion can consist of a single area or several partial areas.

[0032] The fact that a given circle of the circle decomposition has such an area-exceeding value means that a portion of the given circle is outside the boundary of the vehicle area and thus intersects the boundary.

[0033] The storage unit and the calculation unit are particularly preferably arranged in the vehicle or connected to the vehicle. The storage unit is preferably provided and designed so that vehicle regions are stored therein and the results of each method step can be stored, i.e. the vehicle region skeleton from the median axis transformation and the point classification with the resulting classification points are stored. The same applies to the circle decomposition and further embodiments. The calculation unit is preferably provided and designed to perform the method steps, i.e. perform the median axis transformation, the point classification and the circle decomposition, wherein further method steps of further embodiments can also be performed by the calculation unit. More preferably, the calculation unit is connected to the storage unit in a manner that allows signaling, in this way the calculation unit can read out the required parameters and the operation results can be stored in the storage unit.

[0034] According to the present invention, as described above, circle decomposition is performed starting from the first circle and the second circle.

[0035] According to a preferred embodiment, the first radius is such that the first radius corresponds to a maximum radius and / or a maximum allowed radius. More preferably, the first radius depends on the vehicle width and a maximum area excess value.

[0036] In order to meet the specification of the maximum area excess, it is provided that the first radius is greater than half the vehicle width and / or the diameter is greater than the vehicle width. The vehicle width can be the width of the vehicle without the exterior rear-view mirrors, that is to say, for example, from one side to the other. It is also conceivable that the vehicle width is defined by the exterior rear-view mirrors.

[0037] The first radius is determined by the width of the vehicle, preferably without considering the exterior mirrors, the maximum allowed area excess value and the grid resolution. For example, given a vehicle width of 2.06m and a grid resolution of 0.1m, the first radius is selected to be 2.1m in order to cover the width of the vehicle while still satisfying the grid resolution constraint.

[0038] Therefore, the maximum area excess value may be an adjustable parameter for the quality of the proposed method. The maximum area excess value should preferably be greater than or equal to (first radius - vehicle width) / 2. This gives the maximum area excess value for the defined nth circle.

[0039] If the greater than or equal to condition (first radius - vehicle width) / 2 is not met, the grid resolution should be adjusted, although this may slow down the method.

[0040] The fact that the exterior rear view mirrors should preferably be ignored means that, just to cover the side areas assigned to the exterior rear view mirrors, no too large an nth circle is provided, resulting in the need to select an area excess value that is too large.

[0041] According to the invention, a verification step is provided during or after each generation of a circle in order to verify whether the entire vehicle area is covered. The verification step is preferably performed by a computing unit.

[0042] According to a further embodiment, it is provided that when point classification is performed, a first side corner area point and a second side corner area point are defined, wherein a connecting line between the first side corner area point and the second side corner area point corresponds to a side skeleton connecting line, wherein the first side skeleton connecting line and the second side skeleton connecting line intersect at a first point, wherein the first side skeleton connecting line is formed by the skeleton connecting line between the first point and the first corner area point, and the second side skeleton connecting line is formed by the skeleton connecting line between the first point and the second side corner area point, wherein starting from the first point to the first side corner area point along the first side skeleton connecting line, or starting from the first point to the second side corner area point along the second side skeleton connecting line:

[0043] (I) Generate an nth segment with the smallest radius on the first side skeleton connection line or the second side skeleton connection line s side circle, and iteratively adjust the nth s The minimum radius of the side circle until the maximum area exceeds the value condition;

[0044] (II) performing a verification step to determine whether the area of ​​the vehicle area assigned to the first side frame connection line or the second side frame connection line is completely covered;

[0045] (III) If the result of the verification step is no, then generating a (nth) minimum radius on the first side skeleton connection line or the second side skeleton connection line. s +1) side circle, and iteratively adjust the (n s +1) The minimum radius of the side circle, until the maximum area exceeds the value condition, and increases by n s , and execute step (II) again.

[0046] The side corner area points may preferably be exterior rear view mirrors, since these points represent the outermost points of the vehicle in the lateral direction.

[0047] If there are side mirrors on the vehicle, these are taken into account in the central axis transformation, so that corresponding skeleton connecting lines are also provided there in order to describe the specified area.

[0048] The process of covering the side frame connection lines is performed in the same manner as the previous frame connection lines.

[0049] Before executing the circle covering process of the side skeleton connection line, a verification step is preferably performed to determine whether the side skeleton connection line has been covered by other circles. If this is the case, a new circle covering process does not need to be performed. If the result of the verification step is no, that is, the side skeleton connection line has not been covered, the process of covering the side skeleton connection line is performed.

[0050] According to a further embodiment, it is provided that in case the result of the verification step is negative, a third circle is determined starting from the first circle, wherein the first circle intersects the boundary at a minimum of four intersection points, wherein the first intersection point and the second intersection point are opposite to each other with respect to the first skeleton connection line, and wherein the center of the third circle is determined by the first intersection point and the second intersection point;

[0051] or, in the case of a negative result of the verification step, determining a third circle starting from the second circle, wherein the second circle intersects the boundary at a minimum of four intersection points, wherein the first intersection point and the second intersection point are opposite to each other with respect to the first skeleton connecting line, and wherein the center of the third circle is determined by the first intersection point and the second intersection point;

[0052] And wherein, in the case of repeating the verification step of the result, the (n+1)th circle is determined starting from the nth circle, wherein the nth circle intersects with the boundary at a minimum of four intersection points, wherein the first intersection point and the second intersection point are opposite to each other relative to the first skeleton connecting line, and wherein the center of the (n+1)th circle is determined by the first intersection point and the second intersection point.

[0053] The center of the third circle may be determined by the intersection of the first circle or the second circle, and the third circle may be generated accordingly.

[0054] Therefore, in case of further repetition of the verification step with a negative result, the 4th circle with n=4 must be generated and is generated based on the intersection of the third circle with the boundary. This also applies to all subsequent circles associated with the first skeleton connection line.

[0055] In this case, the nth circle of n=3, ... is generated from the first circle or the second circle.

[0056] According to another embodiment, it is provided that, starting from a first circle, a first front circle is determined, wherein the first circle intersects the boundary at a minimum of four intersection points, wherein a third intersection point and a fourth intersection point are opposite one another with respect to the first front skeleton connection line, wherein the center of the first front circle is determined by the third intersection point and the fourth intersection point, and wherein, in the case of a negative verification step, the (nth) v +1) Circle from nth v The circle begins to determine, where the nth v The circle intersects the boundary at at least two intersection points, wherein the first intersection point and the second intersection point are opposite to each other with respect to the first front skeleton connection line, and wherein the (nth) v +1) The center of the circle is determined by the first intersection point and the second intersection point.

[0057] The process of covering the area assigned to the first front frame connection line is performed until the assigned area is completely covered.

[0058] When the maximum area exceeding value condition is satisfied, the circle is determined, which means that if the corresponding area has not been completely covered, a new circle is determined accordingly until the area is completely covered.

[0059] A similar procedure is provided for the rear frame connection line. A similar procedure applies to the side frame connection lines.

[0060] According to a further preferred embodiment, a circle decomposition is provided with a total area excess value for the vehicle area, wherein the total area excess value depends on a maximum area excess value of each circle, a grid resolution and a minimum radius of each circle.

[0061] Preferably, it can be provided that the total area excess value has a minimum value, which can be pre-specified. The minimum value can be selected according to design and application.

[0062] The grid resolution is preferably between 5 cm and 50 cm, particularly preferably between 10 cm and 30 cm, particularly preferably 10 cm.

[0063] The minimum radius of each circle also depends on the grid resolution.

[0064] According to a further embodiment, when it is provided that, after the circle decomposition has been performed, each circle is decomposed into rectangles in such a way that each circle is covered by a rectangle, special consideration has to be given to the grid resolution.

[0065] In this case, the minimum radius has to be chosen so that the circle can be represented by at least two rectangles.

[0066] More preferably, an occupancy map (or "collision map" or "collision map grid") is calculated as the vehicle moves, in particular an integral image of the occupancy map, wherein the movement of the vehicle results in applying a transformation matrix to each circle, and wherein the occupancy of each rectangle within a predetermined time period is checked by fast collision detection.

[0067] Fast collision detection can be performed by a rectangular decomposition of the circle decomposition, since the correspondence calculations that must be performed are greatly simplified.

[0068] Environmental collision checking is performed by a circle-rectangle decomposition based method, where the shape of the vehicle is first decomposed into circles and then into rectangles.

[0069] Due to the rotation invariance of the circle, only the center of the circle needs to be transformed based on the updated vehicle position, and all decomposed rectangles of the circle are positioned based on the updated center of the circle, where the fixed direction is aligned with the open space map of the environment. In order to check the occupancy of the rectangle faster, due to the update of the occupancy, the integral image of the occupancy map is calculated at each time frame, as described below:

[0070]

[0071] Using the integral image I0, the occupancy of the rectangular area in the occupancy map can be easily checked by the following arithmetic equation:

[0072]

[0073] Among them, if ∑ x，y∈A o(x,y) = 0, then the rectangle indicates no collision. Thus, this check is performed for each decomposed rectangle at each pose or position of the vehicle along the trajectory.

[0074] If the value of the above sum is greater than zero, the rectangle is occupied, which indicates a possible collision. These two equations (computation of the integral image once for each image update, and calculation of the rectangle occupancy for each rectangle of the circle-rectangle decomposition of the vehicle area) are referred to as "fast collision detection" or "fast collision estimation". In addition, the invariance of the circle under rotation, and the introduction of axis-aligned rectangles in each oriented vehicle, make the algorithm faster.

[0075] In practical application, the vehicle position and occupancy map are calculated for each image and / or at each time period using dynamic and / or static objects by means of sensors and the method. Based on this information, during trajectory tracking, the vehicle area of ​​the vehicle is checked for collisions at each time period for the remainder of the trajectory. If there is a collision prediction, i.e. the calculated sum is not equal to zero, the vehicle performs measured braking or emergency braking of the vehicle depending on the distance to the predicted collision.

[0076] According to a further preferred embodiment, provision is made for the circle decomposition to be performed exactly once. It is also advantageous if the rectangle decomposition is performed exactly once.

[0077] This means that only the information obtained is used for later applications, without the need to perform new calculations. The corresponding values ​​and calculation results are stored in memory cells for later use.

[0078] The dependent claims give further advantageous embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Additional objects, advantages and convenient uses of the present invention can be found in the following description in conjunction with the accompanying drawings. In the accompanying drawings:

[0080] Figure 1 shows a geometrical representation of a vehicle region according to the prior art;

[0081] Figure 2A It is the medial axis transformation and point classification;

[0082] Figure 2B The process of generating the first circle and the second circle is shown;

[0083] Figure 2C The process of generating the nth circle is shown;

[0084] Figure 2D The process of generating the first back circle is shown;

[0085] Figure 2E The full circle decomposition process of the vehicle area is shown;

[0086] Figure 2F A schematic diagram showing the coverage of the rear skeleton connection line and the determination of the first rear circle;

[0087] Figure 2G Shown according to Figure 2F Schematic diagram of, and determination of the second posterior circle;

[0088] Figure 3 The determination of the minimum radius is shown;

[0089] Figure 4 Shown according to Figure 2E The rectangular decomposition of the circle decomposition.

[0090] In the drawings, identical components always have the same corresponding reference numerals. For the sake of clarity, in some drawings, components that have been identified elsewhere may not be provided with reference numerals. DETAILED DESCRIPTION

[0091] Figure 1A geometrical representation of a vehicle region 1 with a boundary 2 is shown, wherein a bounding box 3 is provided in the form of a rectangle enclosing the vehicle region 1. The bounding box 3 has been covered by a circle decomposition, so that a fixed number of circles, in this case 10 circles with a fixed radius, are provided.

[0092] It can be seen that the corresponding area excess value 17 of each circle is shown and is relatively large relative to the vehicle area 1 of the vehicle, in the current case approximately one third of the area of ​​each circle, and there is a considerable amount in the front and rear areas, so that a large distance must be maintained between the object and the vehicle to prevent detection of a collision, which is disadvantageous when maneuvering or parking.

[0093] The total area excess value, which can be formed, for example, from the sum of the individual area excess values, is therefore relatively large.

[0094] According to the invention, the entire area excess is minimized by the method according to the invention in order to be able to improve collision detection.

[0095] In the following figures, the method according to the invention is described on the basis of particularly preferred exemplary embodiments.

[0096] Figure 2A The vehicle region skeleton 4 of the vehicle region 1 is shown, which is obtained by a median axis transformation (MAT). In the present case, the vehicle region skeleton 4 comprises a first skeleton connection line 16, a first front skeleton connection line 20, a second front skeleton connection line 21, a first rear skeleton connection line 22, a second rear skeleton connection line 23, a first side skeleton connection line 24 and a second side skeleton connection line 25.

[0097] In addition, point classification has been performed, and these points have been classified relative to the vehicle center 26, wherein the point classification is performed starting from the vehicle center 26, so that the point with the largest distance from the vehicle center 26 is defined as the corner area point, and the point of the vehicle center representing the Y node or arranged near the Y node is defined as the wheelbase point. In this process, the first front corner area point 5, the second front corner area point 6, the first rear corner area point 7, the second rear corner area point 8, the first side corner area point 11, the second side corner area point 12, the front wheelbase point 9 and the rear wheelbase point 10 are obtained.

[0098] The first front corner area point 5 and the front wheelbase point 9 form edge points of the first front frame connecting line 20, the second front corner area point 6 and the front wheelbase point 9 form edge points of the second front frame connecting line 21, the first rear corner area point 7 and the rear wheelbase point 10 form edge points of the first rear frame connecting line 22, and the second rear corner area point 8 and the rear wheelbase point 10 form edge points of the second rear frame connecting line 23.

[0099] The line connecting the first side corner area point 11 and the second side corner area point 12 corresponds to the side skeleton connection line 28, wherein the first skeleton connection line 16 and the side skeleton connection line 28 intersect at the first point 27, wherein the first side skeleton connection line 24 is formed by the skeleton connection line between the first point 27 and the first side corner area point 11, and the second side skeleton connection line 25 is formed by the skeleton connection line between the first point 27 and the second side corner area point 12.

[0100] The first side corner area point 11 and the second corner area point respectively correspond to each exterior rearview mirror of the vehicle.

[0101] From the coordinates (x c ,y c ), the term "vehicle center 26" is used to represent the average value of the skeleton points, wherein the set of skeleton points P s (x x ,y s ) is represented by the following formula:

[0102]

[0103] Among them, x s，max is the maximum value of the x coordinate and x s，min is the minimum value of the x coordinate, y s，max is the maximum value of the y coordinate and x s，min is the minimum y-coordinate.

[0104] The term "Y node" is used to refer to the point where the lines of the Y intersect.

[0105] Furthermore, further points corresponding to side corner area points may be classified, which correspond to the positions of the vehicle's rearview mirrors. Side corner area points are those points which are at a certain distance from the vehicle center and do not correspond to any other corner area points.

[0106] The subsequent method steps describe the implementation of a circle decomposition of the vehicle region, wherein each circle of the circle decomposition has a maximum region excess value.

[0107] According to the invention, a first circle 13 and a second circle 14 are generated, wherein the first circle 13 and the second circle 14 have a first radius 18. The first radius 18 depends on the vehicle width 29 and the maximum area excess value. The vehicle width 29 corresponds to the extension of the vehicle in the width direction B, but preferably without taking into account the exterior rearview mirrors.

[0108] The first radius 18 is then selected based on the vehicle width specification and the maximum zone excess value.

[0109] from Figure 2BIt can also be seen in FIG. 1 that the area allocated to the first skeleton connection line 16 is not yet covered by the first circle 13 and the second circle 14 , so further circles are generated starting from the third circle 15 .

[0110] In the present case, starting from the second circle 14, it can be seen that the second circle 14 intersects the boundary 2 at a minimum of two intersection points, namely a first intersection point 30 and a second intersection point 31, due to the need for the first radius 18. These intersection points 30, 31 are opposite to each other with respect to the first skeleton connection line 16.

[0111] like Figure 2C As shown, starting from the intersection points 30 , 31 , a third circle 15 can be easily constructed which has a center 32 and a first radius 18 .

[0112] After the third circle has been generated, a verification step is performed to verify that the area assigned to the first skeleton connection line 14 is completely covered. This is not the case here, so a fourth circle 33 must be generated with a center 34. In this case, starting from the intersection points 30, 31 of the third circle 15 with the boundary 2, the center 34 is determined and a fourth circle 33 with a first radius 18 is generated.

[0113] If necessary, additional circles can be generated in the same way until the entire area assigned to the first skeleton connection line 16 is covered.

[0114] The remaining areas to be covered which are assigned to the front, rear and side frame connection lines 20 , 21 , 22 , 23 , 24 , 25 , 28 must also be covered completely accordingly.

[0115] The process of the first rear frame connection line 22 will be described below in conjunction with the attached Figure 2D This is described by way of example only. The procedure for the other skeleton connection lines 20, 21, 23, 24, 25, 28 will be performed similarly.

[0116] Further, the second circle 14 intersects the boundary 2 at the third intersection 35 and the fourth intersection 36, so the center of the first rear circle 37 can be determined. The third intersection 35 and the fourth intersection 36 are opposite to each other with respect to the first rear frame connection line 22.

[0117] The center 38 of the first rear circle 37 is determined as a function of the intersection points 35 and 36, the location of the center on the first rear frame connection line 22, and the minimum radius. The radius of the first rear circle 37 is iteratively increased until the maximum area excess value requirement is met.

[0118] Preferably, provision is made to ignore the subsequent circles located in the previously formed circle of the first skeleton connection line 16, ie these circles are not generated. Instead, further circles of the first and subsequent skeleton connection line 22 are generated iteratively.

[0119] Figure 2E A full circle decomposition is shown.

[0120] After the vehicle region 1 has been decomposed into circles, it is particularly preferred to provide that the generated circles are further simplified by a rectangular decomposition.

[0121] It is particularly preferred that the rectangular decomposition of the circles is performed in such a way that each circle is covered by a rectangle.

[0122] In this case, a minimum radius has to be chosen, optionally as a function of the grid resolution, such that the circle can be represented by at least two rectangles.

[0123] exist Figure 2F In the example of , multiple back circles are needed to completely cover the area assigned to the corresponding back skeleton connection line (not shown here), because the maximum area excess value condition has been met and the first back circle 37 does not completely cover it. Figure 2G As shown, in order to completely cover the area assigned to the rear frame connecting line, a further rear circle of the corresponding rear frame connecting line, namely, a second rear circle 39 is determined.

[0124] Figure 3 The importance of this is illustrated in , where, for the example of a grid with a resolution of 0.1 m, different circles with corresponding radii are shown.

[0125] It can be seen that for a circle with a diameter of 0.6m, the circle is represented by a single rectangle, since each cell is at least partially located within the circle. This is no longer the case for other larger diameters, so each circle can be covered by at least two rectangles.

[0126] The idea behind the minimum circle radius is to minimize the total number of circles. According to the invention, the proposed method attempts to cover the vehicle area 1 with a minimum number of circles within the allowed boundary conditions (eg maximum area excess and minimum radius).

[0127] Figure 4 A circular decomposition of the vehicle region 1 is shown, wherein a rectangular decomposition has been performed. Figure 1 As can be seen from the comparison of , the coverage of vehicle region 1 is selected to be significantly smaller, which is also shown by the bounding box 3 . Figure 1 and Figure 4 The vehicle region 1 with the boundary 2 and the bounding box 3 is the same, so that it can be clearly seen that Figure 1 The differences from the prior art shown.

[0128] A further advantage of the decomposition and coverage of the vehicle region 1 shown is that the circle is invariant when rotated, so that only the center of the circle needs to be multiplied by the rotation matrix to adjust for the movement of the vehicle.

[0129] By breaking it up into rectangles, the overall calculation time for collision detection can be reduced, since only a few parameters must be used for the calculation.

[0130] Generally, an occupancy map is generated by a circle-rectangle decomposition, in which way possible collisions can be indicated.

[0131] In this case, the occupancy map is monitored in each time period by sensors arranged on or in the vehicle. The sensors are specially designed and provided for detecting distances.

[0132] The various embodiments with all features may be combined and exchanged in any manner.

[0133] All features disclosed in the application documents are regarded as essential features of the invention insofar as they are superior to the prior art either individually or in combination.

[0134] Reference numerals list

[0135] 1 Vehicle area

[0136] 2 Boundaries

[0137] 3 Bounding Box

[0138] 4 Vehicle Area Skeleton

[0139] 5 First front angle area point

[0140] 6 Second front angle area point

[0141] 7 First back angle area point

[0142] 8 Second rear corner area point

[0143] 9 Front wheelbase point

[0144] 10 Rear wheelbase point

[0145] 11 First side angle area point

[0146] 12 Second side angle area point

[0147] 13 First Circle

[0148] 14 Second Circle

[0149] 15 Third Circle

[0150] 16 First skeleton connection line

[0151] 17 Maximum area exceeded

[0152] 18 First Radius

[0153] 19 Y Node

[0154] 20 First front frame connection line

[0155] 21 Second front frame connection line

[0156] 22 First rear frame connection line

[0157] 23 Second rear frame connection line

[0158] 24 First side skeleton connection line

[0159] 25 Second side frame connection line

[0160] 26 Vehicle Center

[0161] 27 First Point

[0162] 28 side frame connection line

[0163] 29 Vehicle Width

[0164] 30 First intersection

[0165] 31 Second intersection

[0166] 32 Center of the third circle

[0167] 33 Fourth Circle

[0168] 34 Center of the fourth circle

[0169] 35 Third intersection

[0170] 36 Fourth Node

[0171] 37 First Back Circle

[0172] 38 Center of the first rear circle

[0173] 39 Second Back Circle

[0174] 40 First front circle

[0175] 41 Second front circle

[0176] 42 First side circle

[0177] 43 Second side circle

Claims

1. A method for geometrical representation of a vehicle region (1) of a vehicle for collision detection, wherein: The vehicle area (1) has a boundary (2), and the method comprises the following steps: (a) performing a medial axis transformation on a vehicle region (1) to generate a vehicle region skeleton (4), and performing point classification on points of the vehicle region skeleton (4) to determine front corner region points (5, 6) and rear corner region points (7, 8), as well as front wheelbase points (9) and rear wheelbase points (10); (b) performing circle decomposition on the vehicle area (1), wherein each circle of the circle decomposition has a maximum area excess value (17), comprising: (i) (I) generating a first circle (13), wherein the center of the first circle (13) is the front wheelbase point (9) and the first circle (13) has a first radius (18), and generating a second circle (14), wherein the center of the second circle (14) is the rear wheelbase point (10) and the second circle (14) has a first radius (18); Wherein, the connection between the front wheelbase point (9) and the rear wheelbase point (10) corresponds to the first skeleton connection line (16); (II) performing a verification step to determine whether the area of ​​the vehicle region (1) assigned to the first skeleton connection line (16) is completely covered, wherein the rear wheelbase point (10) and the front wheelbase point (9) serve as endpoints; (III) if the result of the verification step is negative, generating an nth circle having the first radius (18) on the first skeleton connecting line (16), wherein n=3, ...; increasing n, and executing step (II) again; (ii) along the front wheelbase point (9) and the nth v The nth point between the front corner area points (5, 6) v The front frame connection line extends from the front wheelbase point (9) to the nth v Starting from the front corner point (5, 6), n v =1,2,…; (I) in the nth v Generate the nth with the smallest radius on the front skeleton connection line v previous circle, and iteratively adjust the nth v The minimum radius of the front circle, until the maximum area exceeds the value (17) condition is met; (II) performing a verification step to determine the assignment to the nth v Whether the area of ​​the vehicle area (1) of the front frame connection line is completely covered; (III) If the result of the verification step is no, then in the nth v Generate the first (n) with the minimum radius on the front skeleton connection line v +1) previous circle, and iteratively adjust the (nth v +1) the minimum radius of the previous circle, until the maximum area exceeds the value (17) condition is met, increment n v , and execute step (II) again; (iii) along the rear wheelbase point (10) and the nth h Back corner area point (7, 8) The nth h The rear frame connection line (22, 23) extends from the rear wheelbase point (10) To the nth h Starting from the back corner point (7, 8), n h =1,2,…; (I) Generate an nth h After the circle, and iteratively adjust the nth h The minimum radius of the following circle until the maximum area exceeds value (17) The conditions are met; (II) performing a verification step to determine the assignment to the nth h whether the area of ​​the vehicle area (1) of the rear frame connection line (22, 23) is completely covered; (III) If the result of the verification step is no, then in the nth h The (nth) having the minimum radius is generated on the rear skeleton connection line (22, 23) h +1) back circle, And iteratively adjust the (nth h +1) the minimum radius of the circle after the maximum area exceeds the value (17) and the condition is met, and n is incremented h , and execute step (II) again.

2. The method according to claim 1, Features: When the point classification is performed, a first side angle area point (11) and a second side angle area point (12) are defined, wherein a connecting line between the first side angle area point (11) and the second side angle area point (12) corresponds to a side skeleton connecting line (28), wherein a first side skeleton connecting line (24) and a second side skeleton connecting line (25) intersect at a first point (27), wherein the first side skeleton connecting line (24) is formed by a skeleton connecting line between the first point (27) and the first side angle area point (11), and the second side skeleton connecting line (25) is formed by a skeleton connecting line between the first point (27) and the second side angle area point (12), Wherein, starting from the first point (27) to the first side angle region point (11) along the first side frame connecting line (24), or starting from the first point (27) to the second side angle region point (12) along the second side frame connecting line (25): (I) generating an nth side skeleton connecting line (24) or a second side skeleton connecting line (25) having the minimum radius s side circle, and iteratively adjust the nth s The minimum radius of the side circle, until the maximum area exceeds the value (17) condition is met; (II) performing a verification step to determine whether the area of ​​the vehicle region (1) assigned to the first side frame connection line (24) or the second side frame connection line (25) is completely covered; (III) If the result of the verification step is negative, then generating a (nth) side skeleton connecting line (24) or a second side skeleton connecting line (25) having the minimum radius s +1) side circle, and iteratively adjust the (nth s +1) the minimum radius of the side circle, until the maximum area exceeds the value (17) condition is met, and n is incremented s , and execute step (II) again.

3. The method according to claim 1 or 2, Features: In the case of a negative result of the verification step, determining a third circle (15) starting from the first circle (13), wherein the first circle (13) intersects the boundary (2) at least at a first intersection point (30), a second intersection point (31), a third intersection point (35) and a fourth intersection point (36), wherein the first intersection point (30) and the second intersection point (31) are opposite to each other with respect to the first skeleton connecting line (16), and wherein the center (32) of the third circle (15) is determined by the first intersection point (30) and the second intersection point (31); Alternatively, in the case of a negative result of the verification step, a third circle (15) is determined starting from the second circle (14), wherein the second circle (14) intersects the boundary (2) at least at a first intersection point (30), a second intersection point (31), a third intersection point (35) and a fourth intersection point (36), wherein the first intersection point (30) and the second intersection point (31) are opposite to each other with respect to the first skeleton connecting line (16), and wherein the center (32) of the third circle (15) is determined by the first intersection point (30) and the second intersection point (31); And wherein, in the case of repeating the verification step of the result, the (n+1)th circle is determined starting from the nth circle, wherein the nth circle intersects with the boundary (2) at a first intersection (30), a second intersection (31), a third intersection (35) and a fourth intersection (36), wherein the first intersection (30) and the second intersection (31) are opposite to each other relative to the first skeleton connecting line (16), and wherein the center (34) of the (n+1)th circle is determined by the first intersection (30) and the second intersection (31).

4. The method according to claim 1 or 2, Features: Starting from the first circle (13), a first front circle (40) is determined, wherein the first circle (13) intersects the boundary (2) at a minimum of a first intersection point (30), a second intersection point (31), a third intersection point (35) and a fourth intersection point (36), wherein the third intersection point (35) and the fourth intersection point (36) are opposite to each other relative to the first front skeleton connection line (20), and wherein the center of the first front circle (40) is determined by the third intersection point (35) and the fourth intersection point (36), and wherein, In the case of a negative verification step, from the nth v Starting from the previous circle, determine the (n v +1) front circle, where the nth v The front circle intersects the boundary (2) at least at the first intersection point (30) and the second intersection point (31), wherein the first intersection point (30) and the second intersection point (31) are opposite to each other relative to the first front skeleton connection line (20), wherein the first (n v +1) The center of the front circle is determined by the first intersection point (30) and the second intersection point (31).

5. The method according to claim 1 or 2, Features: The first radius (18) corresponds to a maximum permissible radius which is formed as a function of the vehicle width (29) and the maximum area excess value (17).

6. The method according to claim 1 or 2, Features: The circle decomposition has a total area excess value relative to the vehicle area (1), wherein the total area excess value depends on the maximum area excess value (17) of each circle, a grid resolution and a minimum radius of each circle.

7. The method according to claim 2, Features: The point classification is performed from a pre-specified vehicle center so that the point with the largest distance from the vehicle center is defined as the corner area point (5, 6, 7, 8, 11, 12), and the point of the vehicle center constituting a Y node (19) or arranged near the Y node (19) is defined as the wheelbase point (9, 10).

8. The method according to claim 1 or 2, Features: After the circle decomposition is performed, a rectangle decomposition of each circle is performed such that each circle is covered by a rectangle, wherein an integral image of an occupancy map is calculated while the vehicle moves, wherein the movement of the vehicle causes a transformation matrix to be applied to each circle, wherein the occupancy of each rectangle is checked by fast collision detection.

9. The method according to claim 1 or 2, Features: The circle decomposition is performed exactly once.

Images