Device and method for monitoring the surrounding environment of a vehicle

Abstract

An apparatus for monitoring a surrounding environment of a vehicle may include: a sensor unit including a plurality of detection sensors configured to detect an object outside the vehicle based on frames having a predetermined period; and a control unit configured to: extract a stationary object from external objects detected by the sensor unit by using behavior information of the vehicle, map the extracted stationary object to a preset grid map, add occupancy information to each grid constituting the grid map based on whether the stationary object is mapped to the grid map, calculate an occupancy probability parameter indicating a probability that the stationary object will be located at each grid based on the occupancy information of the grids within the grid map added to a plurality of frames, and monitor the surrounding environment of the vehicle based on the calculated occupancy probability parameter.

Description

Technical Field

[0001] Exemplary embodiments of the present disclosure relate to an apparatus and method for monitoring a surrounding environment of a vehicle, and more particularly, to an apparatus and method for monitoring a surrounding environment of a vehicle by using an OGM (Occupancy Grid Map). Background Art

[0002] Automotive radar refers to a device that detects external objects within a detection area while the vehicle is driving and alerts the driver to help the driver drive the vehicle safely. Figure 1A and Figure 1B FIG. 1 shows an area in which a general vehicle radar transmits a radar signal to detect an external object. The vehicle radar operates to transmit a radar signal according to a frame having a predetermined period and detect an external object. Figure 1B As shown, the signal characteristics of the transmitted radar signal (e.g., waveform, frequency, range resolution, angular resolution, maximum sensing range, and FoV (Field of View)) vary depending on the vehicle system in which the radar is applied. Examples of such systems include DAS (Driver Assistance Systems) such as BSD (Blind Spot Detection), LCA (Lane Change Assist), or RCTA (Rear Cross Traffic Alert).

[0003] A related art of the present disclosure is disclosed in Korean Patent Application Publication No. 10-2013-0130843 published on December 2, 2013. Summary of the Invention

[0004] Various embodiments relate to an apparatus and method for monitoring the surroundings of a vehicle, which can improve the detection accuracy of external objects when monitoring the surroundings of the vehicle by radar.

[0005] In an embodiment, a device for monitoring the surrounding environment of a vehicle is provided. The device may include: a sensor unit including a plurality of detection sensors configured to detect objects outside the vehicle based on frames at a predetermined period; and a control unit configured to: extract stationary objects from the external objects detected by the sensor unit using behavior information of the vehicle; map the extracted stationary objects to a preset grid map; add occupancy information to each grid constituting the grid map depending on whether the mapping is performed; calculate an occupancy probability parameter indicating a probability that a stationary object will be located on each grid based on the occupancy information of the grids in the grid map in a plurality of frames to be monitored; and monitor the surrounding environment of the vehicle based on the calculated occupancy probability parameter. The control unit differentially corrects shadow grids corresponding to shadow areas where the sensor unit cannot detect external objects, according to the speed of the vehicle.

[0006] When the speed of the vehicle is equal to or higher than a preset reference value, the control unit can correct the shadow grid corresponding to the shadow area in the Kth frame where the sensor unit cannot detect the external object with a first scheme, and the first scheme is a scheme of receiving the occupancy probability parameter in the (K-1)th frame.

[0007] When the speed of the vehicle is less than a preset reference value, the control unit may correct the shadow grid corresponding to the shadow area where the sensor unit cannot detect external objects in the Kth frame with a second scheme, where the second scheme is a scheme of receiving occupancy probability parameters of grids surrounding the shadow grid.

[0008] The control unit may correct the shadow grid from the outermost shadow grid in a second scheme and set the highest occupation probability parameter among the occupation probability parameters of the grids located within the setting range from the shadow grid as the occupation probability parameter of the shadow grid.

[0009] In another embodiment, an apparatus for monitoring a surrounding environment of a vehicle is provided, the apparatus comprising: a sensor unit comprising a plurality of detection sensors for detecting objects outside the vehicle according to frames at a predetermined period; and a control unit configured to: extract a stationary object from external objects detected by the sensor unit, map the extracted stationary object to a preset grid map, calculate an occupancy probability parameter indicating a probability that the stationary object will be located on a grid of the grid map based on a result of the mapping, and monitor the surrounding environment of the vehicle by specifying a grid in the grid map where the stationary object is located based on the calculated occupancy probability parameter, while identifying a free space around the vehicle using the specified grid.

[0010] The control unit may determine a peak grid having a maximum occupancy probability parameter among the grids in the grid map, and determine that the stationary object is located on the peak grid when the occupancy probability parameter of the peak grid is equal to or greater than a threshold value defined for the peak grid. The peak grid in which the stationary object is determined to be located may be composed of a plurality of peak grids related to vehicle driving.

[0011] The control unit can be configured to: map a stationary object corresponding to a peak grid to an azimuth map having a plurality of azimuth indexes divided at equal radial intervals according to a set radius and a set angle and sharing coordinates with the grid map; and identify the azimuth index to which the stationary object is not mapped as an idle space around the vehicle.

[0012] When the third azimuth index to which the stationary object is not mapped is adjacent to the first azimuth index and the second azimuth index to which the first stationary object and the second stationary object are mapped respectively, and when the third stationary object exists in the area radially extended by the third azimuth index, the control unit can correct the third azimuth index to the azimuth index to which the stationary object is mapped if a predefined correction condition is met.

[0013] When the distances from the vehicle to the first stationary object, the second stationary object, and the third stationary object are defined as the first distance, the second distance, and the third distance, respectively, the correction conditions may include the following conditions: i) the difference between the first distance and the second distance is less than a reference value; ii) the third azimuth index is between the first azimuth index and the second azimuth index; iii) the difference between the first distance and the third distance is greater than the reference value and the difference between the second distance and the third distance is greater than the reference value; and iv) the third distance is greater than the first distance and the second distance.

[0014] The azimuth indexes on the azimuth map may have index flags having sequential values set according to sequential positions of the respective azimuth indices. The control unit may correct the third azimuth index by incorporating a third stationary object into the third azimuth index using the ordinate and abscissa coordinates of the first stationary object, the ordinate and abscissa coordinates of the second stationary object, and the first and second index flags of the first and second azimuth indices, respectively.

[0015] The control unit can correct the third-party azimuth index using the following equation:

[0016] [equation]

[0017] Xpos(target3)=ΔXpos_unit x(index3-index1)+Xpos(target1)

[0018] Ypos(target3)=ΔXpos_unit x(index3-index1)+Ypos(target1)

[0019] in

[0020]

[0021]

[0022] Among them, Xpos(target3) and Ypos(target3) are the vertical coordinate and horizontal coordinate of the position where the third stationary object is incorporated into the third azimuth index, Xpos(target1) and Ypos(target1) are the vertical coordinate and horizontal coordinate of the first stationary object, Xpos(target2) and Ypos(target2) are the vertical coordinate and horizontal coordinate of the second stationary object, and index1 to index3 are the index marks of the first azimuth index to the third azimuth index respectively.

[0023] In another embodiment, an apparatus for monitoring a vehicle's surroundings is provided. The apparatus may include: a sensor unit including a plurality of detection sensors configured to detect objects outside the vehicle based on frames at a predetermined period; and a control unit configured to: extract stationary objects from the external objects detected by the sensor unit, map the extracted stationary objects onto a preset grid map, calculate an occupancy probability parameter indicating a probability that the stationary objects will be located on a grid of the grid map based on the mapping result, and monitor the vehicle's surroundings by specifying a grid in the grid map where the stationary objects are located based on the calculated occupancy probability parameter. The control unit is configured to apply a predefined clustering algorithm to the specified grids to create one or more clusters consisting of a plurality of grids having the same characteristics, and extract edge grids of each of the created clusters to monitor continuous structures around the vehicle.

[0024] The control unit may determine a peak grid having a maximum occupancy probability parameter among the grids in the grid map, and determine that the stationary object is located on the peak grid when the occupancy probability parameter of the peak grid is equal to or greater than a threshold value defined for the peak grid. The peak grid in which the stationary object is determined to be located may be composed of a plurality of peak grids related to vehicle driving.

[0025] The control unit may create the one or more clusters using a Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm as a clustering algorithm.

[0026] The clustering criterion of the DBSCAN algorithm corresponds to the distance between the peak grids where the stationary objects are located.

[0027] The control unit may extract, from the meshes constituting the cluster, a mesh in which no peak mesh exists in the adjacent meshes as an edge mesh of the cluster.

[0028] The control unit may determine a nearest edge mesh located at a position closest to the vehicle among the extracted edge meshes, and when the determined nearest edge meshes are arranged continuously, the control unit may determine the nearest edge meshes as meshes where continuous structures around the vehicle are located.

[0029] The grid map may have a vertical axis, a horizontal axis, and an index set relative to the vehicle. When determining the closest edge grid of the cluster, the control unit may determine the closest edge grid of the cluster in a variable manner depending on in which quadrant of the first to fourth quadrants divided for the vehicle in the grid map the cluster is located.

[0030] The control unit can be configured to: when the cluster is located in the first quadrant, determine the grid with the index of the smallest vertical coordinate and the smallest horizontal coordinate in the edge grid of the cluster as the nearest edge grid; when the cluster is located in the second quadrant, determine the grid with the index of the largest vertical coordinate and the smallest horizontal coordinate in the edge grid of the cluster as the nearest edge grid; when the cluster is located in the third quadrant, determine the grid with the index of the largest vertical coordinate and the largest horizontal coordinate in the edge grid of the cluster as the nearest edge grid; and when the cluster is located in the fourth quadrant, determine the grid with the index of the smallest vertical coordinate and the largest horizontal coordinate in the edge grid of the cluster as the nearest edge grid.

[0031] According to an embodiment of the present disclosure, the apparatus and method for monitoring a vehicle's surrounding environment can map stationary objects detected by radar onto a preset grid map, add occupancy information to each grid cell constituting the grid map based on whether the stationary object is mapped to the grid map, and then calculate an occupancy probability parameter based on the occupancy information for each grid cell within the grid map added to multiple frames to be monitored, thereby monitoring the vehicle's surrounding environment. The occupancy probability parameter indicates the probability that the stationary object will be located at the corresponding grid cell. Therefore, when monitoring the vehicle's surrounding environment via radar, the apparatus and method can improve the detection accuracy of external objects. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1A and Figure 1B FIG. 1 is a diagram showing an area in which a general vehicle radar transmits a radar signal to detect an external object.

[0033] Figure 2 is a block diagram for describing a configuration of an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure.

[0034] Figure 3 is a diagram illustrating a grid diagram in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure.

[0035] Figures 4 to 8 is a diagram illustrating a process of setting a threshold value of a grid map in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure.

[0036] Figures 9A to 9C and Figure 10 3 is a diagram illustrating a process of updating a grid map in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure.

[0037] Figure 11 is a diagram illustrating a process of mapping a stationary object to a grid map in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure.

[0038] Figures 12 to 14 FIG. 1 is a diagram illustrating a process of determining an extended mapping area in the apparatus for monitoring the surrounding environment of a vehicle according to an embodiment of the present disclosure.

[0039] Figure 15 and Figure 16A 、 Figure 16B is a diagram illustrating a process of correcting an occupancy probability parameter in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure.

[0040] Figures 17 to 20 is a diagram illustrating a process of correcting a shadow grid in the apparatus for monitoring the surrounding environment of a vehicle according to an embodiment of the present disclosure.

[0041] Figures 21 to 28 is a diagram illustrating a process of identifying a vacant space in an apparatus for monitoring surroundings of a vehicle according to an embodiment of the present disclosure.

[0042] Figures 29 to 34 FIG. 1 is a diagram illustrating a process of monitoring a continuous structure in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure.

[0043] Figure 35 is a flowchart for explaining a method for monitoring the surrounding environment of a vehicle according to an embodiment of the present disclosure. Specific embodiments

[0044] Hereinafter, an apparatus and method for monitoring the surrounding environment of a vehicle will be described with reference to the accompanying drawings using various exemplary embodiments. It should be noted that the drawings are not drawn to exact scale and that the thickness of lines or the dimensions of components may be exaggerated for ease of description and clarity. Furthermore, the terms used herein are defined with consideration of the functionality of the present disclosure and may vary based on the customization or intent of the user or operator. Therefore, terms should be defined in accordance with the overall disclosure set forth herein.

[0045] Figure 2 is a block diagram for describing a configuration of an apparatus for monitoring the surrounding environment of a vehicle according to an embodiment of the present disclosure, Figure 3 is a diagram showing a grid diagram in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure, Figures 4 to 8 is a diagram showing a process of setting a threshold value of a grid map in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure, Figures 9A to 9C and Figure 10is a diagram showing a process of updating a grid map in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure, Figure 11 is a diagram showing a process of mapping a stationary object to a grid map in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure, Figures 12 to 14 is a diagram showing a process of determining an extended mapping area in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure, Figure 15 and Figure 16A 、 Figure 16B is a diagram showing a process of correcting an occupancy probability parameter in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure, and Figures 17 to 20 is a diagram illustrating a process of correcting a shadow grid in the apparatus for monitoring the surrounding environment of a vehicle according to an embodiment of the present disclosure. Figures 21 to 28 is a diagram illustrating a process of identifying a vacant space in an apparatus for monitoring surroundings of a vehicle according to an embodiment of the present disclosure. Figures 29 to 34 FIG. 1 is a diagram illustrating a process of monitoring a continuous structure in an apparatus for monitoring a surrounding environment of a vehicle according to an embodiment of the present disclosure. Figure 35 is a flowchart for explaining a method for monitoring the surrounding environment of a vehicle according to an embodiment of the present disclosure.

[0046] refer to Figure 2 , the apparatus for monitoring the surrounding environment of a vehicle according to an embodiment of the present disclosure may include a sensor unit 100 and a control unit 200 .

[0047] The sensor unit 100 may include first to fourth detecting sensors 110, 120, 130, and 140 corresponding to the radar sensor of the vehicle. Figure 2 As shown, the first detection sensor 110 may correspond to a right rear (RR) radar sensor, the second detection sensor 120 may correspond to a left rear (RL) radar sensor, the third detection sensor 130 may correspond to a right front (FR) radar sensor, and the fourth detection sensor 140 may correspond to a left front (FL) radar sensor. Therefore, the detection sensors 110, 120, 130, and 140 may be operated to detect external objects by transmitting radar signals according to a frame having a predefined period and receiving signals reflected from the external objects. In addition, as Figure 1A and Figure 1B As shown, depending on the DAS (Driver Assistance System) (e.g., BSD, LCA, or RCTA) to which the radar sensor is applied, the waveform, frequency, range resolution, angle resolution, maximum sensing distance, and FoV of the radar signal transmitted from the radar sensor may have different characteristics with respect to each frame.

[0048] The control unit 200 is used to monitor the vehicle's surrounding environment by controlling the operation of the vehicle's DAS, and can be implemented as an ECU (Electronic Control Unit), a processor, a CPU (Central Processing Unit), or an SoC (System on a Chip). The control unit 200 can drive an operating system or application program to control multiple hardware components or software components connected to the control unit 200 and perform various data processing operations.

[0049] In this embodiment, the control unit 200 may be operable to extract a stationary object from external objects detected by the sensor unit 100 using the vehicle's behavior information, map the extracted stationary object to a preset grid map, and add occupancy information to each grid constituting the grid map based on whether the stationary object is mapped to the grid map. Furthermore, the control unit 200 may be operable to calculate an occupancy probability parameter indicating the probability that a stationary object will be located at each grid map based on the occupancy information of the grids within the grid map added to the plurality of frames to be monitored, and monitor the vehicle's surrounding environment based on the calculated occupancy probability parameter.

[0050] Hereinafter, a process of monitoring the surrounding environment of the vehicle will be described in detail with respect to each of detailed operations of the control unit 200 .

[0051] 1. Stationary Object Extraction

[0052] First, the control unit 200 can extract a stationary object from the external objects detected by the sensor unit 100 by using the behavior information of the vehicle and the object information acquired based on the result obtained by detecting the external object by the sensor unit 100. That is, the description of this embodiment focuses on a configuration for monitoring stationary objects, not moving objects, among various external objects around the vehicle.

[0053] The vehicle's behavior information may include vehicle speed, yaw rate, speed change information, and steering angle, and the object information may include the number of external objects detected by the sensor unit 100, the longitudinal and lateral distances to each object, the longitudinal and lateral speeds of each object, and the strength of the received signal. The control unit 200 can use the vehicle's behavior information and object information to extract only stationary objects from the external objects. For example, the control unit 200 can distinguish between moving objects and stationary objects by analyzing the relationship between the vehicle's speed and the longitudinal and lateral speeds of the objects, thereby extracting only the stationary objects.

[0054] 2. Stationary Object Mapping

[0055] When extracting a stationary object, the control unit 200 may map the extracted stationary object to a preset grid map. Before the mapping process of the stationary object, the grid map and the update process of the grid map will be described first.

[0056] 2-1. Grid map

[0057] like Figure 3 As shown, a grid map may be pre-set in the control unit 200, and the grid map has a size corresponding to the surrounding area of the vehicle to be monitored. Figure 3 In, X map_max Indicates the maximum distance in the vertical direction (the vertical size of the grid), Y map_max Indicates the maximum distance in the horizontal direction (the horizontal size of the grid), X map_min Indicates the vertical reference position of the grid, Y map_min Indicates the horizontal reference position of the grid map, X map_step Indicates the vertical size of each grid, and Y map_step Indicates the horizontal size of each grid.

[0058] The vertical axis and horizontal axis of the grid map can be set based on the vehicle. If the vertical axis and horizontal axis of the grid map are set based on a specific point rather than the vehicle, more memory resources may be required depending on the mileage of the vehicle. In addition, for the surrounding area of the vehicle, it is effective to set the surrounding environment monitoring area required for outputting warnings to the driver or performing driving control operations of the vehicle. Therefore, the vertical axis and horizontal axis of the grid map can be set based on the vehicle. Therefore, the index (coordinate (i, j)) of the grid constituting the grid map can also be set based on the vehicle, where i and j represent the vertical index and the horizontal index, respectively.

[0059] like Figure 4 As shown, a threshold value for determining whether a stationary object occupies each grid in the grid map can be defined for the corresponding grid in the grid map. As will be described below, the threshold value is used as a value for comparison with the occupancy probability parameter and is used as a reference value for determining whether a stationary object is located at the corresponding grid. The threshold value can be defined for each grid based on a mathematical model according to the strength of the received signal input to the sensor unit 100, and the mathematical model can correspond to the following well-known radar equation, where Pr represents the strength of the received signal, Gt represents the antenna gain, and Rt represents the distance to the object:

[0060]

[0061] Specifically, according to the radar equation, the strength of the received signal can vary depending on the antenna gain and the relative distance to the object. Therefore, the probability of detecting the same object via radar can vary depending on the object's location. For example, when the object is at a short distance, the received signal strength is high, increasing the probability of object detection; whereas, when the object is at a long distance, the received signal strength is low, decreasing the probability of object detection.

[0062] Furthermore, when an object is located in a position with high antenna gain, the received signal strength is high, increasing the probability of object detection. Conversely, when an object is located in a position with low antenna gain, the received signal strength is low, decreasing the probability of object detection. As described above, depending on the DAS (e.g., BSD, LCA, or RCTA) of the vehicle employing the radar sensor, the waveform, frequency, range resolution, angular resolution, maximum sensing range, and Field of View (FoV) of the radar signal transmitted from the radar may have different characteristics across frames. Therefore, each frame may include an area where an object can be repeatedly detected, while only certain frames may include areas where an object can be detected. Consequently, areas that repeat in every frame may have a high probability of object detection, while areas that do not repeat in every frame may have a low probability of object detection. This is because, during two frames, an object can be detected twice in a repeated area, but only once in a non-repeated area.

[0063] Furthermore, for two adjacent radar sensors, such as an RR radar sensor and an RL radar sensor, there may be an area where an object can be redundantly detected by both radar sensors, and an area where the object can be detected by only one radar sensor. Therefore, the area where the object can be redundantly detected by both radar sensors can have a high probability of object detection, while the area where the object can be detected by only one radar sensor can have a low probability of object detection. This is because even if one radar sensor fails to detect an object within the area where the object can be redundantly detected by both radar sensors, the object can still be detected by the other adjacent radar sensor. Conversely, if one radar sensor fails to detect an object within the area where the object can be detected by only one radar sensor, the object cannot be detected by the other adjacent radar sensor.

[0064] From the above, the following two situations can be considered.

[0065] i) Cases where the probability of object detection is highest: "Areas where the distance to the object is short and the antenna gain is high", "Detection areas that are repeated in each frame", and "Detection areas that are redundant between adjacent radar sensors"

[0066] i) Cases with the lowest object detection probability: "Areas with a long distance to the object and low antenna gain," "Detection areas that are not repeated in each frame," and "Detection areas that are not redundant between adjacent radar sensors"

[0067] In both cases, setting the same threshold for all grids within the grid map to determine whether each grid is occupied may be inappropriate. This is because: in case (i), even if the object is not actually present, it may be incorrectly determined to be present (false detection), while in case (ii), even if the object is actually present, it may be incorrectly determined to be absent (missed detection). Therefore, in this embodiment, the threshold for each grid can be set differently based on the object detection probability, thereby preventing false determinations (false detection and missed detection).

[0068] Specifically, the thresholds may be set to different values for independent areas, single overlapping areas, and multiple overlapping areas within the grid map.

[0069] An independent area may be defined as an area sensed by the first detection sensor 110 in the K-th frame within the grid diagram, where K is a natural number, and a single overlap area may be defined as an area within the grid diagram in which the independent area and an area sensed by the first detection sensor 110 in the (K+1)-th frame (i.e., after the K-th frame) different from the K-th frame overlap with each other. That is, an independent area and a single overlap area are distinguished based on whether detection areas by the same detection sensor in each frame overlap with each other. In the case where the first detection sensor is an RR radar Figure 5 In the above example, the grid of the independent region is represented by "0", while the grid of the single overlapping region is represented by "1". The threshold of the grid of the independent region can be set lower than the threshold of the grid of the single overlapping region, which can compensate for the possible false detection and missed detection of objects located in the independent region.

[0070] The multiple overlap area can be defined as an area within the grid diagram in which the area sensed by the second detection sensor 120 adjacent to the first detection sensor 110 overlaps with the single overlap area in the same frame (Kth or (K+1)th frame). That is, the multiple overlap area is determined based on whether the areas detected by two adjacent detection sensors overlap with each other in the same frame. In the case where the first detection sensor is an RR radar and the second detection sensor is an RL radar Figure 6 In FIG, the grid of the area sensed by the first detection sensor 110 is indicated by "0", and the grid of the area where the area sensed by the first detection sensor 110 and the area sensed by the second detection sensor 120 overlap with each other is indicated by "1". Therefore, if the first detection sensor is an RR radar and the second detection sensor is an RL radar Figure 7As shown, the grid diagram can be divided into: an independent area "0" sensed by the first detection sensor 110 in the Kth frame; a single overlapping area "1", which is the overlapping area between the area sensed by the first detection sensor 110 in the Kth frame and the area sensed in the (K+1)th frame; and a multiple overlapping area "2", which is an overlapping area sensed by the first detection sensor 110 and the second detection sensor 120 in the same frame and overlaps with the single overlapping area. When the thresholds of the independent area, the single overlapping area, and the multiple overlapping area are defined as the first threshold, the second threshold, and the third threshold, respectively, a relationship of "first threshold < second threshold < third threshold" can be established in the portion where the thresholds increase linearly, as shown in FIG. Figure 8 shown.

[0071] 2-2. Grid map update

[0072] As described above, since the vertical axis, horizontal index, and grid map index are set based on the vehicle, the grid map index changes according to the vehicle's behavior. Therefore, in order to map a stationary object to the grid map, a process of updating the grid map by changing the grid map index is required. Furthermore, even after a stationary object is mapped to the grid map, the index of the grid to which the stationary object is mapped must be changed according to the vehicle's behavior. When the grid map is updated after a stationary object is mapped to the grid map, the index of the grid to which the stationary object is mapped also changes.

[0073] For this operation, during the period from the (K-1)th frame to the Kth frame, when the longitudinal movement distance of the vehicle is greater than the longitudinal size of the grid or the lateral movement distance of the vehicle is greater than the lateral size of the grid, the control unit 200 may update the grid map. In this case, the control unit 200 may change the index of each grid in the (K-1)th frame according to the index of each grid in the Kth frame based on the longitudinal movement distance, lateral movement distance, and longitudinal angle change of the vehicle.

[0074] Take the index of the grid where the stationary object is located as the changed index as an example, Figure 9A The grid map in the (K-1)th frame is shown with the index of the grid where the stationary object is located. Figure 9B As shown, when the distance traveled by the vehicle in the longitudinal direction is greater than the longitudinal size of the grid, the index of the stationary object on the grid map in the (K-1)th frame needs to be changed based on the Kth frame, because the index of the stationary object on the grid map in the Kth frame is different from the index of the stationary object on the grid map in the (K-1)th frame. Figure 9CWhen the vehicle's longitudinal or lateral movement distance becomes smaller than the longitudinal or lateral size of the grid while turning at the predetermined yaw rate shown, the index of the stationary object in the grid map in the (K-1)th frame needs to be changed based on the Kth frame because the index of the stationary object in the grid map in the Kth frame is different from the index of the stationary object in the grid map in the (K-1)th frame. In this case, the angle change based on the yaw rate can be reflected in the grid map update.

[0075] Will refer to Figure 10 Modeling to describe Figures 9A to 9C The updating process of the grid map.

[0076] First, the control unit 200 calculates the cumulative values of the yaw axis angle change and the movement displacement change of the vehicle during the period from the (K-1)th frame to the Kth frame according to Equation 1 below.

[0077] [Equation 1]

[0078] -Δθ_acc=Δθ_acc+Δθ

[0079]

[0080] -Δγ_acc=Δγ_acc+Δγ

[0081] In Equation 1, Δθ represents the instantaneous angle change of the yaw axis reference of the vehicle, Δθ_acc represents the cumulative angle change of the yaw axis reference during the period from the (K-1)th frame to the Kth frame, Δγ represents the instantaneous displacement of the vehicle, Vs represents the speed of the vehicle, and dt represents the period from the (K-1)th frame to the Kth frame. represents the longitudinal unit vector, represents a lateral unit vector, and Δγ_acc represents the cumulative movement displacement of the vehicle during the period from the (K-1)th frame to the Kth frame.

[0082] The control unit 200 determines whether a grid map update condition is satisfied according to Equation 2 below.

[0083] [Equation 2]

[0084] Δx k = -Δγ·cos(Δθ)

[0085] Δy k =Δγ·sin(Δθ)

[0086] Δx k _acc=Δx k _acc+Δx k

[0087] Δy k_acc=Δy k _acc+Δy k

[0088]

[0089] In Equation 2, Δx k Indicates the instantaneous longitudinal movement distance of the vehicle, Δy k Indicates the instantaneous lateral movement distance of the vehicle, Δx k _acc represents the cumulative longitudinal movement distance of the vehicle, and Δy k _acc represents the cumulative lateral movement distance of the vehicle.

[0090] When the grid map update condition is satisfied according to Equation 2, the control unit 200 updates the grid map according to Equation 3 below.

[0091] [Equation 3]

[0092]

[0093]

[0094]

[0095] In Equation 3, (i, i) represents the index of the grid, (i_update, j_update) represents the index of the updated grid, and floor represents the truncation operator. In Equation 3, the matrix is used as a rotation matrix for rotating the grid map according to the vehicle yaw rate:

[0096] 2-3. Stationary Object Mapping

[0097] According to Equation 4 below, the control unit 200 may convert the position information of the stationary object (ie, the longitudinal distance and the lateral distance to the stationary object) into an index corresponding to the (updated) grid map.

[0098] [Equation 4]

[0099]

[0100]

[0101] In Equation 4, I tgt_n Indicates the vertical index of the target grid, J tgt_n Indicates the horizontal index of the target grid, X tgt_n represents the longitudinal distance to the stationary object, and Y tgt_n Indicates the lateral distance to the stationary object.

[0102] like Figure 11As shown, by specifying the target grid corresponding to the changed index of the grid map, the control unit 200 can map the extracted stationary object to the grid map. In this case, the control unit 200 can add occupancy information having a first value to the target grid to which the stationary object is mapped, and add occupancy information having a second value to other grids. In the present embodiment, the first value can be set to "1" and the second value can be set to "0". Therefore, the value "1" can be added as occupancy information to the target grid to which the stationary object is mapped, and the value "0" can be added as occupancy information to other grids to which the stationary object is not mapped. Hereinafter, the occupancy information added to the index (i, j) in the K-th frame will be represented by Pmap(i, j, k).

[0103] 3. Extended mapping area determination

[0104] As described above, depending on the DAS (e.g., BSD, LCA, or RCTA) of the vehicle to which the radar sensor is applied, the radar signal transmitted from the radar sensor may have different characteristics in terms of waveform, frequency, range resolution, angular resolution, maximum sensing range, and Field of View (FOV) for each frame. Therefore, even if the same stationary object is detected, the index at which the stationary object is detected may change from frame to frame due to the different signal characteristics in each frame. In this case, the occupancy probability parameter (described below) can be reduced by using the number of signal waveforms used. Figure 12 The results obtained when a radar sensor detects the same stationary object by transmitting radar signals with a single waveform and multiple waveforms are shown. Compared to a single waveform, the multiple waveforms disperse the occupied grids across frames, reducing the probability of detecting the stationary object. When the grid map threshold is set to a low value to compensate for the reduced occupancy probability parameter, the stationary object is likely to be falsely detected due to clutter or noise.

[0105] In order to prevent false detection, the control unit 200 according to this embodiment may add occupancy information to the target grid corresponding to the detected stationary object and the surrounding grids. Figure 13 As shown, the control unit 200 can determine that the target grid to which the stationary object is mapped extends the extended mapping area by a preset range, and calculate an occupancy probability parameter by adding occupancy information having a first value to each grid constituting the extended mapping area to monitor the vehicle's surrounding environment. The preset range extended from the target grid can be predefined by the designer, taking into account the similarity between signal waveforms (range resolution and velocity resolution).

[0106] Figure 14The following diagram shows the results obtained when the radar sensor detects the same stationary object by transmitting radar signals with a single waveform and multiple waveforms. By setting an expanded mapping area that extends a preset range from the target grid, the following method for calculating the occupancy probability parameter for each grid cell that constitutes the expanded mapping area can be used to avoid a decrease in the probability of detecting the stationary object, even with multiple waveforms.

[0107] 4. Calculation of occupation probability parameters

[0108] The process of calculating the occupancy probability parameter of the grid map in this embodiment follows the occupancy probability calculation method of a general OGM (Occupancy Grid Map) based on the following equation 5.

[0109] [Equation 5]

[0110]

[0111]

[0112]

[0113] In Equation 5, R 1：k represents the sensing data (the above-mentioned object information) from the first frame to the K-th frame of the sensor unit 100 (radar sensor), and V 1：k represents the behavior data of the vehicle from the first frame to the K-th frame (the above-mentioned behavior information), and 10 represents the prior probability (0 in this embodiment).

[0114] When the occupancy information Pmap(i, j, k) added to each mesh in this embodiment is applied to the occupancy probability calculation method based on the above-mentioned Equation 5, the occupancy probability parameter p is calculated according to the following Equation 6.

[0115] [Equation 6]

[0116]

[0117] In Equation 6, M represents the number of frames to be monitored.

[0118] 5. Grid map update error correction

[0119] The vehicle's speed, movement displacement, and yaw axis angle change are acquired by sensors applied to the vehicle and used as factors to determine whether the grid map update conditions are met. Since such sensed values inevitably contain errors, even if the grid map update conditions are not actually met due to the errors contained in the sensed values, it is determined that the grid map update conditions have been met. In this case, the grid map may be incorrectly updated. As described above, during the grid map update process, the control unit 200 operates to change the index of the target grid to which the stationary object is mapped. Therefore, when the grid map is incorrectly updated, an error may occur between the index corresponding to the actual position of the stationary object and the index of the stationary object mapped to the incorrectly updated grid map. As a result, this error may lead to false detection and missed detection of the stationary object.

[0120] Will refer to Figure 15 (a) to (d) describe the occurrence of errors. Figure 15 (a) shows that a stationary object is mapped to a grid ① in a (K-1)th frame, and then the grid ① is expanded by a preset range to determine a first expanded mapping area; and Figure 15 (b) shows that the above-mentioned grid map update conditions are satisfied at frame K, thereby updating the grid map. Since the grid map has been updated, the index of the grid to which the stationary object is mapped also changes, resulting in the grid being updated to grid ②. Furthermore, the position of the stationary object actually detected by sensor unit 100 remains within grid ①. As a result, an error occurs between the index of the grid corresponding to the actual position of the stationary object and the index of the grid for the stationary object mapped to the updated grid map.

[0121] When the grid map is updated as the (K-1)th frame switches to the Kth frame, the control unit 200 can correct the various occupancy probability parameters of the grid constituting the second extended mapping area by comparing the first extended mapping area in the (K-1)th frame with the second extended mapping area in the Kth frame, thereby correcting the above-mentioned update error.

[0122] refer to Figure 15(c), based on the K-th frame on the (K-1)-th frame, the control unit 200 can specify a first area consisting of grids with increased occupancy probability parameters in the grids of the second extended mapping area. That is, the first area corresponds to the grids that were not occupied in the (K-1)-th frame but were occupied in the K-th frame. In addition, based on the K-th frame on the (K-1)-th frame, the control unit 200 can specify a second area consisting of grids with decreased occupancy probability parameters in the grids of the first extended mapping area. That is, the second area corresponds to the grids that were occupied in the (K-1)-th frame but were not occupied in the K-th frame. In addition, by replacing the occupancy probability parameters of the second area with the occupancy probability parameters of the first area, the control unit 200 can correct the individual occupancy probability parameters of the grids constituting the second extended mapping area in the K-th frame. Therefore, as Figure 15 As shown in (d), the extended mapping area can be constructed while matching the extended mapping area with the position of the stationary object actually detected by the sensor unit 100. When the grids constituting the second area are unoccupied for a preset time, the occupation probability parameter of the grid can be reset to "0".

[0123] Figure 16A FIG. 4 shows an example of an occupancy probability parameter on a grid map before the update error of the grid map is updated. Figure 16A As shown, grid ① corresponds to a position that has a lateral / longitudinal error from the actual position of the stationary object but maintains a predetermined occupancy probability value, while grid ② corresponds to the actual position of the stationary object but has an occupancy probability value lower than the surrounding grids because grid ② is a newly occupied grid.

[0124] Figure 16B FIG. 4 shows an example of an occupancy probability parameter on a grid map after the update error of the grid map is corrected. Figure 16B As shown, grid ① is the previously occupied grid and has a low occupancy probability value through reset, while grid ② corresponds to the actual position of the stationary object and has an occupancy probability value higher than the surrounding grids because grid ② is a newly occupied grid but inherits the predetermined occupancy probability value.

[0125] 6. Correction of shadow areas

[0126] As described above, the detection sensor according to this embodiment can be implemented as a radar sensor. Figure 17 As shown in FIG, due to the FoV and installation characteristics (installation angle and position) of the radar sensor, shadow areas appear where the radar sensor cannot detect external objects.

[0127] To correct the shadow grid corresponding to the shadow area, the control unit 200 may operate to correct the shadow grid by using a first method of receiving occupancy probability parameters in the (K-1)th frame or a second method of receiving occupancy probability parameters of grids surrounding the shadow grid.

[0128] When the speed of the vehicle is equal to or higher than a preset reference value, the first method may be executed. Figure 18 As shown, grid ① in the (K-1)th frame does not correspond to a shadow grid, and therefore retains its occupancy probability parameter. When the vehicle speed is equal to or higher than the reference value, the grid map is updated, and grid ① in the Kth frame belongs to a shadow grid. In this case, the control unit 200 can set the occupancy probability parameter of grid ① in the (K-1)th frame to the occupancy probability parameter of the shadow grid ① in the Kth frame, thereby minimizing the loss caused by missed detection by the radar sensor.

[0129] When the speed of the vehicle is lower than the reference value, the second method can be performed. That is, when the vehicle is traveling at a very low speed or stopped, the grid map is not updated even if the (K-1)th frame is switched to the Kth frame. Therefore, the first method cannot be applied. In this case, the control unit 200 can be operated to set the occupancy probability parameter of the grid surrounding the shadow grid to the occupancy probability parameter of the shadow grid. In this case, as Figure 19 As shown, the control unit 200 may execute the second method from the outermost shadow grid to obtain the occupancy probability parameters of the grids that are not the shadow grids. The control unit 200 may set the highest occupancy probability parameter among the occupancy probability parameters of the grids within a preset range (e.g., one grid) of the shadow grid as the occupancy probability parameter of the corresponding shadow grid. Figure 20 Results obtained by correcting the shadow area to set an occupancy probability parameter having a predetermined value to the shadow grid are shown.

[0130] 7. Stationary object position determination (peak detection)

[0131] When updating the grid map, determining the extended mapping area, correcting the update error, and correcting the shadow area are performed through the above process, the control unit 200 can operate to specify the grid where the stationary object is likely to be located based on the occupancy probability parameters of the grids within the extended mapping area.

[0132] That is, the control unit 200 can determine the peak grid with the highest occupancy probability parameter among the grids within the extended mapping area determined for the multiple frames to be monitored. When the occupancy probability parameter of the peak grid is equal to or greater than the threshold defined for the peak grid, the control unit 200 can determine that the stationary object is located at the peak grid. When the vehicle is driving, the control unit 200 can monitor the surrounding environment of the vehicle by repeatedly executing the stationary object position determination method based on "peak detection". The peak grid where the determined stationary object is located can be composed of multiple peak grids related to the driving of the vehicle. In this embodiment, if the grid corresponding to the index (i, j) corresponds to the peak grid, it is marked as "Dmap (i, j) = 1", and if the grid (i, j) corresponding to the index (i, j) does not correspond to the peak grid, it is marked as "Dmap (i, j) = 0".

[0133] 8. Free space identification

[0134] Based on the plurality of peak grids determined as above, the control unit 200 may be operable to identify an empty space around the vehicle. An empty space may be a space that ensures safety when a vehicle enters the space, such as a parking space in a parking lot that is not occupied by other vehicles.

[0135] Regarding the grid pattern for identifying free spaces, although free spaces can be identified on the above-mentioned grid pattern, this embodiment adopts a grid pattern such as Figure 21 The configuration of identifying free space on the orientation map shown in FIG. 1 can reduce the memory resources required to identify free space. Figure 21 As shown, the azimuth map has a predefined setting radius and has a plurality of azimuth indexes divided at equal radial intervals according to predefined setting angles. The azimuth map can be predefined in the control unit 200 ( Figure 21 An example of an orientation map with 16 orientation indexes formed by a predefined setting angle of 22.5° is shown). The orientation map has a center relative to the vehicle setting and shares the vertical and horizontal coordinates with the above-mentioned grid map. Figure 21 As shown, the azimuth index on the azimuth map may have an index flag (represented by index) having a sequential value (1 to 16) set according to the sequential position of each azimuth index.

[0136] Therefore, if Figure 22 As shown, the control unit 200 may map the stationary objects corresponding to the previously determined peak grid to the azimuth map, and identify the azimuth indexes to which the stationary objects are not mapped as free spaces around the vehicle (e.g., Figure 22 As shown, when multiple stationary objects are mapped to one azimuth index, the position information of the stationary object having the smallest distance from the vehicle can be used to perform the azimuth correction process to be described later). Figure 23The result of free space recognition is shown when other vehicles are parked diagonally around the vehicle. Figure 24 The result of free space recognition is shown when other vehicles are parked at right angles around the vehicle.

[0137] like Figure 25 As shown, depending on the driving conditions of the vehicle, there may be a structure (e.g., a guardrail) where a stationary object and a free space are continuously formed. In this case, the free space is formed discontinuously and cannot be used as a free space that can ensure the safety of the vehicle. Therefore, in terms of vehicle driving safety, it may be desirable to correct this space to a space where a stationary object exists. Therefore, this embodiment adopts a configuration that corrects the discontinuous free space to a space where a stationary object exists.

[0138] Specifically, when the third azimuth index to which the stationary object is not mapped is adjacent to the first azimuth index and the second azimuth index to which the first stationary object and the second stationary object are mapped, respectively, and when the third stationary object exists in the area radially extended by the third azimuth index ( Figure 26 ), when a predefined correction condition is met, the control unit can correct the third-party azimuth index to the azimuth index mapped to the stationary object.

[0139] When the distances from the vehicle to the first stationary object (target 1) to the third stationary object (target 3) are defined as first distance R1 to third distance R3, respectively, the correction conditions may include the following four conditions (each of the following reference values may be defined as a specific numerical value according to the designer's intention and radar detection performance (range resolution)):

[0140] Condition i) the difference between the first distance and the second distance is less than a reference value: |R1-R2|<reference value;

[0141] Condition ii) the third azimuth index is between the first azimuth index and the second azimuth index;

[0142] Condition iii) a difference between the first distance and the third distance is greater than a reference value and a difference between the second distance and the third distance is greater than a reference value: “|R1-R3|>reference value” and “|R2-R3|>reference value”; and

[0143] Condition iv) The third distance is greater than the first distance and the second distance: "R3>R1" and "R3>R2".

[0144] The above condition i) is a condition for determining whether the first and second stationary objects correspond to objects having the same characteristics (e.g., a continuous guardrail). The above condition ii) is a condition for determining when an empty space is identified between the first and second stationary objects. The above conditions iii) and iv) are predetermined conditions for determining whether a third stationary object exists in the region radially extending from the third azimuth index, so as to incorporate the third stationary object into the third azimuth index.

[0145] When the above correction conditions are met, the control unit 200 can incorporate the third stationary object into the third azimuth index (e.g., Figure 27 Specifically, the control unit 200 may use the following equation 7 to correct the third-party azimuth index:

[0146] [Equation 7]

[0147] Xpos(target3)=ΔXpos_unit x(index3-index1)+Xpos(target1)

[0148] Ypos(target3)=ΔXpos_Unit x(index3-index1)+Ypos(target1)

[0149] in

[0150]

[0151]

[0152] Among them, Xpos(target3) and Ypos(target3) are the vertical coordinate and horizontal coordinate of the position where the third stationary object is incorporated into the third azimuth index, Xpos(target1) and Ypos(target1) are the vertical coordinate and horizontal coordinate of the first stationary object, Xpos(target2) and Ypos(target2) are the vertical coordinate and horizontal coordinate of the second stationary object, and index1 to index3 are the index marks of the first azimuth index to the third azimuth index respectively.

[0153] As a specific example, assuming that the index marks of the first azimuth index to the third azimuth index are 1, 3, and 2 respectively (index1=1, index2=3, index3=2), the vertical coordinate and horizontal coordinate of the first stationary object are 5 and 0 respectively (Xpos(target1)=5, Ypos(target1)=0), and the vertical coordinate and horizontal coordinate of the second stationary object are 0 and 3 respectively (Xpos(target2)=0, Ypos(target2)=3). In the above case, the calculations are ΔXpos_unit=-2.5, ΔYpos_unit=1.5, Xpos(target3)=2.5, and Ypos(target3)=1.5, so that the third stationary object is incorporated into the coordinates (2.5, 1.5) in the third azimuth index. Therefore, discontinuous free space can be removed to improve the driving safety of the vehicle. Figure 28 The result of removing the discontinuous free space existing on the guardrail around the vehicle through the above correction process is shown.

[0154] 9. Continuous structural monitoring

[0155] The multiple peak grids determined in the process of "7. Stationary object position determination (peak detection)" can be clustered according to whether they have the same characteristics (that is, whether they correspond to the same physical structure). When the edge grids of the corresponding formed clusters are arranged continuously, these grids can be determined as grids where the continuous structures around the vehicle are located. The determined continuous structures may correspond to other vehicles parked continuously in the parking lot, guardrails, etc. Therefore, the control unit 200 can park or drive the vehicle by avoiding the continuous structures around it. The continuous structure monitoring process can be divided into a cluster creation process and an edge grid extraction process.

[0156] First, the cluster creation process will be described. A predefined clustering algorithm can be applied to the peak grid specified in the process of "7. Stationary object position determination (peak detection)" to create one or more clusters consisting of multiple grids with the same characteristics. The clustering algorithm may correspond to the density-based clustering of applications space with noise (DBSCAN) algorithm. As is well known, the DBSCAN algorithm is an algorithm that clusters high-density parts in a point set, which is known to be a strong model for noise and outlier identification. In this embodiment, the distance between the peak grids where the stationary objects are located is applied as the clustering criterion of the DBSCAN algorithm. That is, the cluster consists of a set of peak grids in which the distance between the peak grids is less than or equal to the reference value, or they are adjacent to each other. Figure 29 An example of the created clusters is shown.

[0157] Thereafter, the control unit 200 may be operated to extract the edge mesh of each cluster and monitor the continuous structure around the vehicle. Specifically, the control unit 200 may monitor the continuous structure around the vehicle by determining the edge mesh of each cluster and extracting the nearest edge mesh therefrom.

[0158] First, in extracting edge meshes of a cluster, when no peak mesh exists in any adjacent meshes among meshes constituting the cluster (Dmap(i,j)=1), the control unit 200 may extract the mesh as an edge mesh of the cluster.

[0159] Figures 30A to 30C An example of extracting the edge grid of the clusters is shown. Figure 30A The grids constituting the cluster are shown, where the following Equation 8 is used to determine whether a peak grid exists in the grid immediately adjacent to the grid.

[0160] [Equation 8]

[0161]

[0162] refer to Figure 30B , in the "test grid", since all the adjacent grids correspond to the peak grid, the edge flag is set to the value 0 in Equation 8, while in the "adjacent grid", since at least one of the adjacent grids does not correspond to the peak grid, the edge flag is set to the value 1 in Equation 8. As a result, as Figure 30C As shown in FIG, the edge grid can be extracted by setting the edge flag to a value of 1. Figures 30A to 30C The edge marker extraction process in Equation 8 is performed to secure memory resources and reduce the amount of calculation and complexity in the process of determining the nearest edge marker to be described later.

[0163] When extracting the edge grids of the cluster, the control unit 200 can determine the nearest edge grid located at the position closest to the vehicle from the extracted edge grids. As described above, the vertical axis, horizontal axis and index of the grid map are set relative to the vehicle. Therefore, if Figure 29 As shown, the grid map can be divided into the first to fourth quadrants relative to the vehicle. The positions of the first to fourth quadrants are the same as the positions of the quadrants in the general XY coordinate system.

[0164] In determining the closest edge mesh of a cluster, the control unit 200 may be operable to determine the closest edge mesh of the cluster in a variable manner depending on in which quadrant of the first to fourth quadrants the cluster exists.

[0165] Specifically, if Figure 31As shown, the control unit 200 can: i) when the cluster is located in the first quadrant, determine the grid with the index of the smallest vertical coordinate and the smallest horizontal coordinate in the edge grid of the cluster as the nearest edge grid; ii) when the cluster is located in the second quadrant, determine the grid with the index of the largest vertical coordinate and the smallest horizontal coordinate in the edge grid of the cluster as the nearest edge grid; iii) when the cluster is located in the third quadrant, determine the grid with the index of the largest vertical coordinate and the largest horizontal coordinate in the edge grid of the cluster as the nearest edge grid; and iv) when the cluster is located in the fourth quadrant, determine the grid with the index of the smallest vertical coordinate and the largest horizontal coordinate in the edge grid of the cluster as the nearest edge grid.

[0166] When the clustered nearest edge grids are determined and the determined nearest edge grids are arranged continuously, the control unit 200 may determine these nearest edge grids as grids where continuous structures around the vehicle are located. Figure 32 The results of clustering and nearest edge mesh extraction are shown when there are other vehicles parked consecutively in a diagonal manner around the vehicle. Figure 33 Shown are the results of clustering and nearest edge mesh extraction when there are other vehicles parked consecutively at right angles around the vehicle. Figure 34 The results of clustering and nearest edge mesh extraction are shown when there are guardrails around the vehicle. Therefore, the control unit 200 can stop or drive the vehicle by avoiding the continuous structures around it.

[0167] Figure 35 FIG. 1 is a flowchart for illustrating a method for monitoring the surrounding environment of a vehicle according to an embodiment of the present disclosure. Figure 35 A method of monitoring the surrounding environment of a vehicle according to the present embodiment will be described. Detailed description of any portion that overlaps with the preceding portion will be omitted, and the following description will focus on its time-series configuration.

[0168] First, in step S100 , the control unit 200 extracts a stationary object from objects outside the vehicle detected by the sensor unit 100 by using behavior information of the vehicle.

[0169] Then, in step S200, the control unit 200 maps the stationary object extracted in step S100 to a preset grid map, adds occupancy information to each grid constituting the grid map according to whether the stationary object is mapped to the grid map, and calculates an occupancy probability parameter based on the occupancy information of the grids within the grid map added to the multiple frames to be monitored, the occupancy probability parameter indicating the probability that the stationary object will be located at the corresponding grid.

[0170] In step S200 , the control unit 200 maps the stationary object to the grid map, and at the same time updates the grid map by changing each index of the grid constituting the grid map according to the behavior information of the vehicle.

[0171] In addition, in step S200, the control unit 200 converts the position information of the stationary object into an index corresponding to the grid map, maps the stationary object to the grid map by specifying a target grid corresponding to the index of the grid map, adds occupancy information having a first value to the target grid to which the stationary object is mapped, and adds occupancy information having a second value to other grids, wherein the second value is smaller than the first value.

[0172] Furthermore, in step S200 , the control unit 200 calculates an occupancy probability parameter by determining an expanded mapping area expanded by a preset range based on the target mesh to which the stationary object is mapped, and adding occupancy information having a first value to each mesh constituting the expanded mapping area.

[0173] In addition, in step S200, when the grid map is updated as the (K-1)th frame switches to the Kth frame, the control unit 200 corrects the occupancy probability parameters of the grids constituting the second extended mapping area by comparing the first extended mapping area in the (K-1)th frame with the second extended mapping area in the Kth frame. Specifically, based on the Kth frame on the (K-1)th frame, the control unit 200 specifies a first area consisting of grids with increased occupancy probability parameters in the grids of the second extended mapping area, and specifies a second area consisting of grids with decreased occupancy probability parameters in the grids of the first extended mapping area. Then, by replacing the occupancy probability parameters of the second area with the occupancy probability parameters of the first area, the control unit 200 corrects the respective occupancy probability parameters of the grids constituting the second extended mapping area in the Kth frame.

[0174] Furthermore, in step S200, the control unit 200 corrects the shadow mesh corresponding to the shadow region in which the sensor unit 100 cannot detect external objects in the K-th frame, using either a first method of receiving the occupancy probability parameter in the (K-1)-th frame or a second method of receiving the occupancy probability parameter of the mesh surrounding the shadow mesh. In this case, when the speed of the vehicle is equal to or higher than a preset reference value, the control unit 200 corrects the shadow mesh according to the first method, and when the speed of the vehicle is lower than the reference value, the control unit 200 corrects the shadow mesh according to the second method.

[0175] Following step S200, in step S300, the control unit 200 monitors the vehicle's surroundings based on the occupancy probability parameters calculated in step S200. Specifically, the control unit 200 determines the peak grid with the highest occupancy probability parameter among the grids within the expanded mapping area determined for the multiple frames to be monitored. When the occupancy probability parameter of the peak grid is equal to or greater than a threshold defined for the peak grid, the control unit 200 determines that the stationary object is located at the peak grid. In step S300, multiple peak grids may be determined based on the vehicle's driving. In step S300, the control unit 200 may use the peak grids to identify free space around the vehicle.

[0176] Next, the control unit 200 applies a predefined clustering algorithm to the grid specified in step 300 to create one or more clusters consisting of multiple grids with the same characteristics (S400). In step S400, the control unit 200 uses the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm as a clustering algorithm to create one or more clusters. The clustering criterion of the DBSCAN algorithm corresponds to the distance between the peak grids where the stationary objects are located.

[0177] Next, the control unit 200 extracts edge meshes of each cluster created in step S400 to monitor the continuous structure around the vehicle (S500). In step S500, the control unit 200 extracts meshes that do not have a peak mesh in the adjacent meshes from the meshes constituting the cluster as edge meshes of the cluster, determines the nearest edge meshes located at a position close to the vehicle among the extracted edge meshes, and when the determined nearest edge meshes are arranged continuously, determines these nearest edge meshes as meshes where the continuous structure around the vehicle is located.

[0178] In this manner, the apparatus and method for monitoring a vehicle's surroundings according to this embodiment can map stationary objects detected by radar onto a preset grid map, add occupancy information to each grid cell constituting the grid map based on whether the stationary object is mapped to the grid map, and then calculate an occupancy probability parameter based on the occupancy information for each grid cell within the grid map over multiple frames to be monitored, thereby monitoring the vehicle's surroundings. The occupancy probability parameter indicates the probability that a stationary object will be located at the corresponding grid cell. Consequently, when monitoring the vehicle's surroundings using radar, the apparatus and method can improve the accuracy of detecting external objects.

[0179] For example, the embodiments described in this specification may be implemented as a method or process, an apparatus, a software program, a data stream, or a signal. Although features are discussed only in a single context (e.g., only in a method), the features discussed may be implemented in another type (e.g., an apparatus or program). The apparatus may be implemented in suitable hardware, software, or firmware. The method may be implemented in a device such as a processor, which generally refers to a processing device including a computer, a microprocessor, an integrated circuit, or a programmable logic device. Processors also include communication devices that facilitate information communication between end users, such as computers, cellular phones, PDAs (personal digital assistants), and other devices.

[0180] Although exemplary embodiments of the present disclosure are disclosed for illustrative purposes, it will be appreciated by those skilled in the art that various modifications, additions, and substitutions are possible without departing from the scope and spirit of the present disclosure as defined in the appended claims. Therefore, the true technical scope of the present disclosure should be defined by the appended claims.

Claims

1. A device for monitoring the surrounding environment of a vehicle, comprising: a sensor unit including a plurality of detection sensors for detecting an object outside the vehicle according to frames at a predetermined period; as well as a control unit configured to: extract a stationary object from external objects detected by the sensor unit, map the extracted stationary object to a preset grid map, calculate an occupancy probability parameter indicating a probability that the stationary object will be located on a grid of the grid map based on a result of the mapping, and monitor the surrounding environment of the vehicle by specifying a grid in the grid map where the stationary object is located based on the calculated occupancy probability parameter, while identifying a free space around the vehicle using the specified grid. The control unit determines a peak grid having a maximum occupancy probability parameter among the grids in the grid map, and determines that the stationary object is located on the peak grid when the occupancy probability parameter of the peak grid is equal to or greater than a threshold defined for the peak grid. 2 . The apparatus according to claim 1 , wherein the peak grid in which the determined stationary object is located consists of a plurality of peak grids related to driving of the vehicle.

3. The device according to claim 2, wherein The control unit is configured to: map a stationary object corresponding to the peak grid to an azimuth map, the azimuth map having a plurality of azimuth indexes, the plurality of azimuth indexes being divided at equal radial intervals according to a set radius and a set angle and sharing coordinates with the grid map; and identify the azimuth index to which the stationary object is not mapped as a free space around the vehicle.

4. The device according to claim 3, wherein When a third azimuth index to which a stationary object is not mapped is adjacent to the first azimuth index and the second azimuth index to which the first stationary object and the second stationary object are mapped respectively, and when a third stationary object exists in an area radially extending from the third azimuth index, the control unit corrects the third azimuth index to the azimuth index to which the stationary object is mapped if a predefined correction condition is met.

5. The device according to claim 4, wherein When the distances from the vehicle to the first stationary object, the second stationary object, and the third stationary object are defined as a first distance, a second distance, and a third distance, respectively, the correction conditions include the following conditions: i) the difference between the first distance and the second distance is less than a reference value; ii) the third azimuth index is between the first azimuth index and the second azimuth index; iii) the difference between the first distance and the third distance is greater than a reference value and the difference between the second distance and the third distance is greater than a reference value; and iv) the third distance is greater than the first distance and the second distance.

6. The device according to claim 5, wherein: The azimuth indexes on the azimuth map have index marks, and the index marks have sequence values set according to the sequence positions of the respective azimuth indexes; as well as The control unit corrects the third azimuth index by incorporating the third stationary object into the third azimuth index using the ordinate and abscissa of the first stationary object, the ordinate and abscissa of the second stationary object, and the first index flag and the second index flag of each of the first azimuth index and the second azimuth index.

7. The device according to claim 6, wherein The control unit corrects the third position angle index using the following equation: [equation] Among them, Xpos(target3) and Ypos(target3) are the vertical coordinate and horizontal coordinate of the position where the third stationary object is incorporated into the third azimuth index, Xpos(target1) and Ypos(target1) are the vertical coordinate and horizontal coordinate of the first stationary object, Xpos(target2) and Ypos(target2) are the vertical coordinate and horizontal coordinate of the second stationary object, and index1, index2 and index3 are the respective index marks of the first azimuth index to the third azimuth index.

8. The device according to claim 1, wherein The control unit is configured to apply a predefined clustering algorithm to the designated grids to create one or more clusters consisting of a plurality of grids having the same characteristics, and extract edge grids of each of the created clusters to monitor continuous structures around the vehicle.

9. The device according to claim 8, wherein The control unit determines a peak grid having a maximum occupancy probability parameter among the grids in the grid map, and determines that the stationary object is located on the peak grid when the occupancy probability parameter of the peak grid is equal to or greater than a threshold value defined for the peak grid, wherein the peak grid where the stationary object is determined to be located is composed of a plurality of peak grids related to driving of the vehicle.

10. The device according to claim 9, wherein The control unit creates one or more clusters using a density-based spatial clustering with noise (DBSCAN) algorithm as the clustering algorithm.

11. The device according to claim 10, wherein The clustering criterion of the DBSCAN algorithm corresponds to the distances between the peak grids where the stationary objects are determined to be located.

12. The device according to claim 8, wherein The control unit extracts a mesh in which no peak mesh exists among adjacent meshes from the meshes constituting the cluster as an edge mesh of the cluster.

13. The device according to claim 12, wherein The control unit determines a nearest edge mesh located at a position closest to the vehicle among the extracted edge meshes, and when the determined nearest edge meshes are arranged continuously, the control unit determines the nearest edge meshes as meshes where continuous structures around the vehicle are located.

14. The apparatus according to claim 13, wherein: The grid diagram has a vertical axis, a horizontal axis, and an index disposed relative to the vehicle; as well as When the nearest edge mesh of the cluster is determined, the control unit determines the nearest edge mesh of the cluster in a variable manner depending on in which quadrant the cluster is located among the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant divided in the grid map for the vehicle.

15. The device according to claim 14, wherein The control unit is configured to: When the cluster is located in the first quadrant, determining a grid with an index of a minimum ordinate and a minimum abscissa among edge grids of the cluster as a nearest edge grid; When the cluster is located in the second quadrant, determining a grid with an index of a maximum ordinate and a minimum abscissa among edge grids of the cluster as a nearest edge grid; When the cluster is located in the third quadrant, determining a grid with an index of a maximum ordinate and a maximum abscissa among edge grids of the cluster as a nearest edge grid; as well as When the cluster is located in the fourth quadrant, a grid with an index of a minimum ordinate and a maximum abscissa among the edge grids of the cluster is determined as a nearest edge grid.

16. The device according to claim 1, wherein The control unit corrects a shadow grid corresponding to a shadow area where the sensor unit cannot detect an external object in a differential manner according to a speed of the vehicle.

17. The device according to claim 1, wherein When the speed of the vehicle is equal to or higher than a preset reference value, the control unit corrects the shadow mesh corresponding to the shadow area in which the sensor unit cannot detect the external object in the K-th frame using a first scheme, wherein the first scheme is a scheme of receiving the occupancy probability parameter in the K-1-th frame.

18. The device according to claim 1, wherein When the speed of the vehicle is less than a preset reference value, the control unit corrects a shadow grid corresponding to a shadow area in which the sensor unit cannot detect an external object in the Kth frame using a second scheme, wherein the second scheme is a scheme of receiving occupancy probability parameters of grids surrounding the shadow grid.

19. The device according to claim 18, wherein The control unit corrects the shadow grid from the outermost shadow grid in the second scheme and sets the highest occupation probability parameter among the occupation probability parameters of the grids located within the setting range from the shadow grid as the occupation probability parameter of the shadow grid.

Images