Dynamic occupancy grid map building apparatus and method
The dynamic occupancy grid map generation method addresses radar sensor limitations by using extended object tracking and elliptical joining to improve ADAS performance in detecting cut-ins and cluttered environments.
Patent Information
- Application Number
- KR1020240093862
- Authority / Receiving Office
- KR · KR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2024-07-16
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2044-07-16
AI Technical Summary
Radar sensors in vehicles often struggle to accurately detect cut-in situations due to sparse point clouds and cluttered environments, leading to misrecognition and ineffective operation of Advanced Driver Assistance Systems (ADAS) functions.
A dynamic occupancy grid map generation method that utilizes extended object tracking and elliptical joining to update the occupancy probability of a target vehicle based on point clouds and tracks, compensating the vehicle's position over time.
Enhances perception performance by generating precise and robust dynamic occupancy grid maps, enabling accurate detection of cut-in situations even under adverse conditions, and maintaining reliable ADAS functions.
Smart Images

Figure 112024077188045-PAT00006_ABST
Abstract
Description
Technology Field
[0001] The present embodiments relate to a dynamic occupancy grid map creation device and method. Background Technology
[0002] Recently, the number of vehicles equipped with radar sensors has been increasing. Based on information output from radar sensors mounted on the vehicle, the vehicle's electronic control unit can calculate the distance, relative speed, and angle between the vehicle and surrounding objects.
[0003] Vehicles equipped with radar sensors in this way can provide various safety or convenience functions by utilizing the distance, relative speed, and angle between the vehicle and surrounding objects.
[0004] For example, ADAS (Advanced Driver Assistance Systems) can refer to advanced driver assistance systems designed to help the driver drive and enhance safety.
[0005] With the information from the aforementioned radar sensor, ADAS can be utilized for Adaptive Cruise Control (ACC) to maintain distance from the vehicle ahead and control speed, Lane Keeping Assist (LKA) to help keep the lane, and Blind Spot Detection (BSD) to monitor blind spots.
[0006] These functions require accurate detection by radar sensors. Cut-ins, where surrounding vehicles enter the vehicle's path, occur frequently in congested situations with many vehicles nearby, and misrecognizing them can make it difficult to provide ADAS functions.
[0007] Therefore, technology to detect the cut-in of surrounding vehicles is required. The problem to be solved
[0008] Against this background, the present disclosure aims to provide a dynamic occupancy grid map generation apparatus and method that updates the occupancy probability of a target vehicle within an occupancy grid map by performing extended object tracking and elliptical joining of a target vehicle based on the point cloud and track of the target vehicle. means of solving the problem
[0009] In order to solve the aforementioned problem, in one aspect, the present disclosure provides a dynamic occupancy grid map generation device comprising: an object detection unit that calculates a measurement based on a received signal from a radar sensor and detects objects around a self-vehicle; a shape estimation unit that estimates the shape of a target vehicle when a target vehicle is detected in the measurement; an occupancy probability update unit that performs elliptical fitting of the shape of the target vehicle to a point cloud and updates the occupancy probability of a grid for the target vehicle in a dynamic occupancy grid map (DOGM) based on the fitted shape; and a compensation unit that compensates the position of the self-vehicle over time to the dynamic occupancy grid map.
[0010] In another aspect, the present disclosure provides a method for creating a dynamic occupancy grid map, comprising: an object detection step of calculating a measurement based on a received signal from a radar sensor and detecting an object around an ego vehicle; a shape estimation step of estimating the shape of a target vehicle when a target vehicle is detected in the measurement; an occupancy probability update step of ellipsizing the shape of the target vehicle to a point cloud and updating the occupancy probability of a grid for the target vehicle in a dynamic occupancy grid map (DOGM) based on the fitted shape; and a compensation step of compensating the position of the ego vehicle over time to the dynamic occupancy grid map. Effects of the invention
[0011] According to the present disclosure, a dynamic occupancy grid map generation apparatus and method can improve perception performance by generating a precise and robust DOGM by recognizing the cut-in situation of a target vehicle even under adverse conditions where point clouds are sparse using only a radar sensor.
[0012] In addition, the present disclosure can create a reliable dynamic occupancy grid map even in environments where a large number of clutters, such as ground signals or superstructures, are detected by using the track of a target vehicle instead of a point cloud. Brief explanation of the drawing
[0013] FIG. 1a is a drawing showing a cut-in situation in which another vehicle enters in the direction of travel from the right rear of a self-vehicle according to one embodiment. FIG. 1b is a diagram showing the detection results of a radar sensor mounted on a vehicle in the situation of FIG. 1a according to one embodiment. FIG. 2 is a diagram showing a dynamic occupancy grid map created based on the results of FIG. 1b according to one embodiment. FIG. 3 is a block diagram for briefly explaining a dynamic occupancy grid map creation device (10) according to one embodiment of the present disclosure. FIG. 4 is a drawing for explaining that the least squares method according to one embodiment is applied to form an elliptical joint. FIG. 5 is a drawing for explaining the joining of the shape of a target vehicle to a track and point cloud according to one embodiment. FIG. 6a is a diagram showing the situation on a dynamic occupancy grid map before a target vehicle according to one embodiment cuts into the vehicle's driving path. FIG. 6b is a diagram showing a situation in which a target vehicle cuts into the vehicle's driving path according to one embodiment on a dynamic occupancy grid map. FIG. 7a is a diagram showing the point in time when a target vehicle cuts into the vehicle's driving path according to one embodiment on a dynamic occupancy grid map. FIG. 7b is a diagram showing the point in time when a point cloud is detected on the vehicle's path of travel according to one embodiment on a dynamic occupancy grid map. FIG. 8 is a flowchart illustrating a method for creating a dynamic occupancy grid map according to one embodiment of the present disclosure. FIG. 9 is a drawing for explaining step S830 according to one embodiment in more detail. Specific details for implementing the invention
[0014] Hereinafter, some embodiments of the present disclosure will be described in detail with reference to the exemplary drawings. In assigning reference numerals to the components of each drawing, the same components may have the same reference numeral as much as possible, even if they are shown in different drawings. Furthermore, in describing the embodiments, if it is determined that a detailed description of related known components or functions may obscure the essence of the technical concept, such detailed description may be omitted. Where terms such as "comprising," "having," or "consisting of" are used in this specification, other parts may be added unless "only" is used. Where a component is expressed in the singular, it may include a plural unless otherwise specified.
[0015] Additionally, terms such as first, second, A, B, (a), (b), etc., may be used to describe the components of the present disclosure. These terms are used merely to distinguish the components from other components, and the nature, order, sequence, or number of the components are not limited by such terms.
[0016] In describing the positional relationship of components, where it is stated that two or more components are "connected," "combined," or "joined," it should be understood that while the two or more components may be directly "connected," "combined," or "joined," they may also be "connected," "combined," or "joined" with other components "intervened." Here, the other components may be included in one or more of the two or more components that are "connected," "combined," or "joined" with one another.
[0017] In describing the temporal flow relationship regarding components, methods of operation, or methods of production, for example, when the temporal or sequential relationship is described using "after," "following," "next," or "before," it may include cases where the relationship is not continuous unless "immediately" or "directly" is used.
[0018] Meanwhile, where numerical values or corresponding information regarding a component (e.g., levels, etc.) are mentioned, even without separate explicit notation, the numerical values or corresponding information may be interpreted as including a range of error that may occur due to various factors (e.g., process factors, internal or external shocks, noise, etc.).
[0019] There are various reasons why the observed point cloud of a conventional radar sensor may not be dense, even when a target is located at close range.
[0020] The aforementioned causes may include cases where the target is located across a side lobe area where the signal strength is relatively weak, or where there is a lot of noise in the Range-Doppler (RD) map generated in an environment with high Radar Cross Section (RCS), such as inside a steel tunnel or a congested situation where many vehicles are distributed, or cases where valid peak information is not detected by the peak detection logic, such as the Constant False Alarm Rate (CFAR) algorithm.
[0021] FIG. 1a is a diagram showing a cut-in situation in which another vehicle enters in the direction of travel from the right rear of a self-vehicle according to one embodiment, and FIG. 1b is a diagram showing the detection result of a radar sensor mounted on the self-vehicle in the situation of FIG. 1a according to one embodiment.
[0022] FIG. 2 is a diagram showing a dynamic occupancy grid map created based on the results of FIG. 1b according to one embodiment.
[0023] Referring to Figures 1a and 1b, the number of point clouds detected in a of Figure 1b indicates that the vehicle is positioned across the side lobe area, and the signal strength of other vehicles is relatively low.
[0024] Referring to Figure 2, grid cells occupied according to the lacking cloud points in the dynamic occupancy grid map do not occupy the path of the vehicle's movement, so even if other vehicles are actually blocking the path of the vehicle's movement, it can be determined that they are not blocking it on the dynamic occupancy grid map.
[0025] In such cases, misrecognition by the radar sensor may make it difficult to provide ADAS (Advance Driver Assistance Systems) functions that utilize radar information.
[0026] In order to overcome the above, the following describes a reliable dynamic occupancy grid map generation device (10) that overcomes the adverse conditions where measurements are scarce using only radar sensors.
[0027] FIG. 3 is a block diagram for briefly explaining a dynamic occupancy grid map creation device (10) according to one embodiment of the present disclosure.
[0028] Referring to FIG. 3, the dynamic occupancy grid map creation device (10) of the present disclosure may include an object detection unit (110), a shape estimation unit (120), an occupancy probability update unit (130), and a compensation unit (140), etc.
[0029] The dynamic occupancy grid map generation device (10) of the present disclosure calculates a measurement based on a received signal from a radar sensor, detects objects around the vehicle, and when a target vehicle is detected in the measurement, estimates the shape of the target vehicle, fits the shape of the target vehicle into an ellipse on a point cloud, updates the grid occupancy probability for the target vehicle in the dynamic occupancy grid map (DOGM) based on the fitted shape, and compensates the position of the vehicle over time in the dynamic occupancy grid map.
[0030] In one embodiment, the dynamic occupancy grid map creation device (10) may be an ADAS that provides information to help drive a vehicle or helps the driver control the vehicle.
[0031] The dynamic occupancy grid map creation device (10) according to the present disclosure can be mounted on a manned vehicle and an autonomous vehicle that is controlled by a driver.
[0032] The object detection unit (110) can calculate a measurement value based on a received signal from a radar sensor and detect objects around the vehicle.
[0033] Specifically, the object detection unit (110) can receive information from a plurality of radar sensors mounted on the vehicle, for example, a front radar sensor that detects the front, a side rear radar sensor that detects the side rear, etc.
[0034] In one embodiment, the radar sensor mounted on the vehicle may include an antenna unit, a radar transmitter unit, and a radar receiver unit.
[0035] Specifically, the antenna section includes one or more transmitting antennas and one or more receiving antennas, and each transmitting and receiving antenna may be an array antenna in which one or more radiating elements are connected in series by a feed line, but is not limited thereto.
[0036] This antenna section includes a plurality of transmitting antennas and a plurality of receiving antennas, and may have various types of antenna array structures depending on the arrangement order and spacing thereof.
[0037] The radar transmitter can switch to one of the multiple transmitting antennas included in the antenna section and transmit a transmission signal through the switched transmitting antenna or transmit a transmission signal through a multi-transmission channel assigned to the multiple transmitting antennas.
[0038] This radar transmitter includes an oscillator that generates a transmission signal for a single transmission channel assigned to a switched transmitting antenna or for multiple transmission channels assigned to multiple transmitting antennas. This oscillator may include, for example, a voltage-controlled oscillator (VCO) and an oscillator.
[0039] The radar receiver can receive the received signal reflected from an object through the receiving antenna.
[0040] In addition, the radar receiver can switch to one of a plurality of receiving antennas to receive a receiving signal, which is a reflected signal of a transmission signal transmitted through the switched receiving antenna and reflected by a target, or receive a receiving signal through a multi-receiving channel assigned to a plurality of receiving antennas.
[0041] Such a radar receiver may include a low-noise amplifier (LNA) that low-noise amplifies a received signal received through a single receiving channel assigned to a switched receiving antenna or through a multiple receiving channel assigned to a transmitting antenna, a mixer that mixes the low-noise amplified received signal, an amplifier that amplifies the mixed received signal, and an analog-digital converter (ADC) that converts the amplified received signal into digital data.
[0042] The object detection unit (110) can calculate a measurement value by performing a Fast Fourier Transform (FFT) on the received signal. Specifically, the calculation unit can calculate a measurement value by performing a first FFT to convert the frequency into a Range-Time Index and performing a second FFT on the time to convert it into a Range-Doppler Index.
[0043] The object detection unit (110) can detect objects around the vehicle based on the calculation location of each measurement and the relative speed between each measurement and the vehicle.
[0044] The shape estimation unit (120) can estimate the shape of the target vehicle when the target vehicle is detected in the measurement. The target vehicle may be set during the process in which the present disclosure provides ADAS, and for example, the target vehicle may be a vehicle that blocks the vehicle's path of travel when performing ACC.
[0045] The shape estimation unit (120) can generate a track using measurements to estimate the shape of a target vehicle. The shape estimation unit (120) can generate a track that includes adjacent measurements, and the center of the track can be calculated as the average value of the included measurements. The shape estimation unit (120) can detect a target vehicle by comparing the track generated in the previous scan with the track generated in the current scan.
[0046] In one embodiment, the shape estimation unit (120) can estimate the shape of a target vehicle by utilizing an Extended Object Tracking (EOT) technique. An Extended Object Tracking technique may refer to a tracking technique that considers the shape of an object, such as direction angle, width, and length, in addition to a point target tracking technique that estimates the position and velocity of an object.
[0047] The shape estimation unit (120) may use one of the following elliptical extension object tracking techniques: a random matrix method (a method of estimating a random matrix containing information on the elliptical orientation angle and major / minor axis) and an explicit parameterization method (a method of estimating a random vector by explicitly modeling the elliptical orientation angle and major / minor axis). In some cases, two or more methods may be adopted and combined in the form of an Interacting Multiple Model (IMM) to use a hybrid measurement equation.
[0048] The shape estimation unit (120) can estimate the shape of the target vehicle by applying an extended object tracking technique to the detected target vehicle.
[0049] As described above, the present disclosure can estimate information of a track having a certain size by applying extended object tracking.
[0050] The occupancy probability update unit (130) can ellipse-fit the shape of the target vehicle to the point cloud and update the occupancy probability of the grid for the target vehicle in the dynamic occupancy grid map (DOGM) based on the fitted shape.
[0051] A point cloud can refer to a data set formed by the aggregation of multiple measurements. For example, a point cloud can be a set of clustered measurements during the process of creating a track.
[0052] The occupancy probability update unit (130) can join the shape of the target vehicle into an ellipse. The ellipse joining technique is a method used in computer vision and pattern recognition.
[0053] FIG. 4 is a drawing for explaining that the least squares method according to one embodiment is applied to form an elliptical joint.
[0054] Referring to FIG. 4, the occupancy probability update unit (130) can calculate the center of the ellipse based on the least squares method. The occupancy probability update unit (130) can calculate the most suitable optimal ellipse shape by applying the least squares method to the point cloud detected in two dimensions.
[0055] The occupancy probability update unit (130) can update the grid occupancy probability for the target vehicle only when the target vehicle's track exists within the dynamic occupancy grid map.
[0056] FIG. 5 is a drawing for explaining the joining of the shape of a target vehicle to a track and point cloud according to one embodiment.
[0057] Referring to FIG. 5, the occupancy probability update unit (130) can attach the shape of the target vehicle to the point clouds (p1, p2, p3, p4, p5) used to update the track. For example, the occupancy probability update unit (130) can attach the shape of the target vehicle so that the closest ellipse is formed for a plurality of point clouds.
[0058] The occupancy probability update unit (130) can perform elliptical fitting using the following mathematical formula 1.
[0059] [Mathematical Formula 1]
[0060]
[0061]
[0062] Here, x c , y c θ represents the center position of the ellipse, l1 and l2 represent the major and minor axes of the joined ellipse, and θ represents the angle formed between the x-axis and the major axis (α in FIG. 5). The occupancy probability update unit (130) can calculate the center position of the ellipse by utilizing the least squares method to join the optimal ellipse to the point cloud. Additionally, the occupancy probability update unit (130) can determine the center position of the track as the center of the ellipse (x c , y c It can be joined to a point cloud using ).
[0063] As described above, the present disclosure can estimate the shape of a target vehicle more accurately by performing elliptical fitting based on a detected point cloud or track.
[0064] The present disclosure may utilize the ellipse position calculated according to the aforementioned mathematical formula 1 only when updating the dynamic occupancy grid map, and may not reflect it in the elliptical extension object tracking technique that estimates the shape of the target vehicle. Accordingly, the present disclosure can ensure that the expected performance in Bayesian filtering is maintained and the shape of the target object is accurately estimated, as there is a risk that the filter will diverge if the estimate is forcibly corrected.
[0065] In one embodiment, the occupancy probability update unit (130) can be joined so that the center of the ellipse is located within a predetermined distance from the center of the target vehicle updated at the previous time. Accordingly, the present disclosure can prevent the result by least squares method from overfitting in an unintended direction when a small number of points are detected.
[0066] The occupancy probability update unit (130) can predict the occupancy probability over time. Based on the occupancy probability of the updated grid, the occupancy probability update unit (130) can predict movement of the moving object and update the occupancy probability. In one embodiment, the occupancy probability update unit (130) can input the occupancy probability of the updated grid into a prediction model to which a Kalman filter, an extended Kalman filter, an unscented Kalman filter, a Recurrent Neural Network (RNN), and a Convolutional Neural Network (CNN) are applied, and obtain the occupancy probability over time as output data.
[0067] FIG. 6a is a diagram showing the situation before a target vehicle cuts into the vehicle's driving path according to one embodiment on a dynamic occupancy grid map, and FIG. 6b is a diagram showing the situation when a target vehicle cuts into the vehicle's driving path according to one embodiment on a dynamic occupancy grid map.
[0068] Referring to FIGS. 6a and FIGS. 6b, the target vehicle (track 19) is before the vehicle's path cut-in in FIGS. 6a, and is in a situation where it has cut-in into the vehicle's path in FIGS. 6b.
[0069] In the case where a general dynamic occupancy grid map is updated, the vehicle is mistakenly judged to have free space up to the vehicle ahead (track 7) because the point cloud in Fig. 6b is not detected on the vehicle's path of travel, and thus it is impossible to respond accordingly.
[0070] On the other hand, in a situation where the dynamic occupancy grid map of the present disclosure is updated, even though the point cloud is not detected on the vehicle's path of travel as in FIG. 6b, the shape of the target vehicle can be estimated using an extended object tracking technique and, by ellipsojunction, it can be determined that the target vehicle is blocking the path of the ego vehicle, and a response accordingly is possible (ego vehicle deceleration control when ACC is in operation).
[0071] FIG. 7a is a diagram showing the point in time when a target vehicle cuts into the vehicle's path of travel according to one embodiment on a dynamic occupancy grid map, and FIG. 7b is a diagram showing the point in time when a point cloud is detected on the vehicle's path of travel according to one embodiment on a dynamic occupancy grid map.
[0072] Referring to FIGS. 7a and 7b, with the update of a typical dynamic occupancy grid map, a cut-in situation of the target vehicle occurs at the 358th scan, and a point cloud is not detected on the vehicle's driving path until the 368th scan, so there is a possibility that ADAS functions such as ACC may malfunction.
[0073] On the other hand, according to the present disclosure, the shape of the target vehicle is estimated based on the track or point cloud of the target vehicle in the 358th scan, and by updating the occupancy probability, it is possible to recognize that the target vehicle is in a cut-in situation.
[0074] The compensation unit (140) can compensate the position of the vehicle over time in a dynamic occupancy grid map.
[0075] The compensation unit (140) continuously receives a reception signal while the vehicle is moving, and the position of the vehicle before the update and the position of the vehicle after the update may differ on the dynamic occupancy grid map. Therefore, the compensation unit (140) corrects the position of the vehicle to the time when the reception signal is received so that all data on the dynamic occupancy grid map can be consistently represented at the same reference point.
[0076] In one embodiment, the dynamic occupancy grid map creation device (10) may be implemented as an Electronic Control Unit (ECU). The Electronic Control Unit may include at least one element among one or more processors, memory, storage, user interface inputs, and user interface outputs, and these may communicate with each other via a bus. Additionally, the computer system may also include a network interface for connecting to a network. The processor may be a CPU or a semiconductor device that executes processing instructions stored in memory and / or storage. The memory and storage may include various types of volatile / non-volatile storage media. For example, the memory may include ROM and RAM.
[0077] Hereinafter, a steering control method using a dynamic occupancy grid map creation device (10) capable of performing all of the above-described disclosures will be described.
[0078] FIG. 8 is a flowchart illustrating a method for creating a dynamic occupancy grid map according to one embodiment of the present disclosure.
[0079] Referring to FIG. 8, the method for creating a dynamic occupancy grid map according to the present disclosure may include an object detection step (S810) that calculates a measurement based on a received signal from a radar sensor and detects an object around the vehicle; a shape estimation step (S820) that estimates the shape of the target vehicle when a target vehicle is detected in the measurement; an occupancy probability update step (S830) that ellipse-fits the shape of the target vehicle to a point cloud and updates the occupancy probability of the grid for the target vehicle in the dynamic occupancy grid map (DOGM) based on the fitted shape; and a compensation step (S840) that compensates the position of the vehicle over time in the dynamic occupancy grid map.
[0080] The shape estimation step (S820) can estimate the shape of the target vehicle by utilizing Extended Object Tracking (EOT). The dynamic occupancy grid map creation method of the present disclosure can initialize the dynamic occupancy grid map before performing shape estimation.
[0081] The occupancy probability update step (S830) can update the grid occupancy probability for the target vehicle only when the target vehicle's track exists within the dynamic occupancy grid map.
[0082] The occupancy probability update step (S830) can fit the shape of the target vehicle to the point cloud used to update the track.
[0083] The occupancy probability update step (S830) can fit the center position of the track to the center of the ellipse in the point cloud.
[0084] The occupancy probability update step (S830) can be joined so that the center of the ellipse is located within a predetermined distance from the center of the target vehicle updated at the previous time point. Here, the center of the target vehicle may refer to the center position of the ellipse joined to the point cloud detected in the previous scan.
[0085] The occupancy probability update step (S830) can calculate the center of the ellipse based on the least squares method.
[0086] FIG. 9 is a drawing for explaining step S830 according to one embodiment in more detail.
[0087] Referring to FIG. 9, the dynamic occupancy grid map creation device (10) can determine whether a track of a target vehicle exists within the dynamic occupancy grid map (S910).
[0088] If the track of the target vehicle exists within the dynamic occupancy grid map (Yes of S910), the dynamic occupancy grid map creation device (10) can elliptically fit the shape of the target vehicle to the point cloud used to update the track (S920).
[0089] The dynamic occupancy grid map creation device (10) can update the occupancy probability of a corresponding grid within the dynamic occupancy grid map using an elliptical shape (S930).
[0090] The dynamic occupancy grid map creation device (10) can predict the movement direction of updated moving objects and predict the occupancy probability of the corresponding grid (S940).
[0091] As described above, according to the present disclosure, the device and method for creating a dynamic occupancy grid map can improve perception performance by creating a precise and robust dynamic occupancy grid map even in adverse conditions where it is difficult for the radar sensor to detect using only the radar sensor.
[0092] In addition, the present disclosure can provide robust ADAS functions by more accurately estimating the shape of a target vehicle using only a small number of detected point clouds.
[0093] In addition, since the present disclosure updates the occupancy probability of the grid using tracks, it is possible to create a dynamic occupancy grid map that is effective even in environments where a large number of clutters, such as ground signals or superstructures, are detected.
[0094] The foregoing description is merely an illustrative explanation of the technical concept of the present disclosure, and those skilled in the art to which the present disclosure pertains may make various modifications and variations within the scope of the essential characteristics of the technical concept. Furthermore, since these embodiments are intended to explain, not limit, the scope of the technical concept is not limited by these embodiments. The scope of protection of the present disclosure shall be interpreted by the claims below, and all technical concepts within an equivalent scope shall be interpreted as being included within the scope of rights of the present disclosure.
Claims
Claim 1 A dynamic occupancy grid map creation device comprising: an object detection unit that calculates a measurement based on a received signal from a radar sensor and detects objects around a self-vehicle; a shape estimation unit that estimates the shape of a target vehicle when a target vehicle is detected in the measurement; an occupancy probability update unit that ellipse-fits the shape of the target vehicle to a point cloud only when a track of the target vehicle exists within a dynamic occupancy grid map (DOGM) and updates the occupancy probability of the grid for the target vehicle in the dynamic occupancy grid map based on the fitted shape; and a compensation unit that compensates the position of the self-vehicle over time to the dynamic occupancy grid map. Claim 2 delete Claim 3 In claim 1, the occupancy probability update unit is a dynamic occupancy grid map creation device that combines the shape of the target vehicle with the point cloud used to update the track. Claim 4 In claim 1, the occupancy probability update unit is a dynamic occupancy grid map creation device that fits the center position of the track to the center of the ellipse in the point cloud. Claim 5 In claim 1, the occupancy probability update unit is a dynamic occupancy grid map creation device that joins the center of the ellipse to be located within a predetermined distance from the center of the target vehicle updated at the previous time point. Claim 6 In claim 1, the shape estimation unit is a dynamic occupancy grid map creation device that estimates the shape of the target vehicle using Extended Object Tracking (EOT). Claim 7 In claim 1, the occupancy probability update unit is a dynamic occupancy grid map creation device that calculates the center of an ellipse based on the least squares method. Claim 8 A method for creating a dynamic occupancy grid map comprising: an object detection step for calculating a measurement based on a received signal from a radar sensor and detecting objects around the self vehicle; a shape estimation step for estimating the shape of the target vehicle when a target vehicle is detected in the measurement; an occupancy probability update step for ellipse fitting the shape of the target vehicle to a point cloud only when a track of the target vehicle exists within a dynamic occupancy grid map (DOGM), and updating the occupancy probability of the grid for the target vehicle in the dynamic occupancy grid map based on the fitted shape; and a compensation step for compensating the position of the self vehicle over time to the dynamic occupancy grid map. Claim 9 delete Claim 10 In claim 8, the occupancy probability update step is a dynamic occupancy grid map creation method that joins the shape of the target vehicle to the point cloud used to update the track. Claim 11 In claim 8, the occupancy probability update step is a dynamic occupancy grid map creation method that fits the center position of the track to the point cloud as the center of the ellipse. Claim 12 In claim 8, the occupancy probability update step is a dynamic occupancy grid map creation method in which the center of the ellipse is positioned within a predetermined distance from the center of the target vehicle updated at a previous time point. Claim 13 In claim 8, the shape estimation step is a dynamic occupancy grid map generation method that estimates the shape of the target vehicle using Extended Object Tracking (EOT). Claim 14 In claim 8, the above-mentioned occupancy probability update step is a dynamic occupancy grid map creation method that calculates the center of an ellipse based on the least squares method.
Citation Information
Patent Citations
Image processing apparatus, image processing method,data processing apparatus, data processing method, andstorage medium
KR1020010060219A
Autonomous driving apparatus and method of collision avoidance controlling the same
KR1020230117936A
Sensor fusion system based on radar and camera and method of calculating the location of nearby vehicles
KR102456151B1
Feature evaluation method by using occupancy probability of grids
KR100787242B1
Detection of position overlap (PO) between objects
US20230408642A1
