Ghost object detection

Abstract

The techniques and systems herein enable ghost object detection. Specifically, a reflection line is determined that indicates a potential reflection surface between a first mobile object and a second mobile object. If enough stationary objects are within a region of the reflection line, then it is determined whether one or more of the stationary objects within the region are within a distance of a reflection point. The expected velocity of the second object is then determined and checked against the velocity of the second object. If the expected velocity is close to the velocity, then the second object is determined to be a ghost object. By doing so, the system can effectively identify ghost objects in a variety of environments, allowing downstream operations to function as designed.

Description

Background Technology

[0001] Advanced sensor systems are increasingly being implemented in vehicles to provide context awareness. Technologies such as radio detection and ranging (RADAR) and light detection and ranging (LiDAR) enable vehicles to identify and track objects in their vicinity. However, many of these technologies rely on the propagation of electromagnetic waves. Because these waves can reflect objects, they often travel multiple paths to and from the target of interest. These multiple paths can result in ghost objects—objects that appear to be present to the sensor system but are not actually there. Accurately identifying such ghosts is crucial to ensuring that downstream operations, such as advanced driver assistance systems (ADAS), function as designed. Some conventional systems attempt to identify ghost objects in specific contexts (e.g., using guardrail trackers) or by using multiple sensors. However, such conventional systems often cannot identify ghosts in the broader environment and / or without prior information about the environment. Summary of the Invention

[0002] This document relates to systems, apparatuses, techniques, and methods for implementing ghost object detection. These systems and apparatuses may include components or devices (e.g., processing systems) for performing the techniques and methods described herein.

[0003] Some aspects described below include a system comprising at least one processor configured to receive information about a plurality of objects approaching a main vehicle. The objects may include two or more moving objects and one or more stationary objects. The processor is also configured to identify one or more pairs of moving objects. The processor is further configured to, for each of the moving object pairs, determine a reflection line perpendicular to and between a first and a second object in the respective moving object pair. The processor is also configured to determine whether one or more stationary objects are within a region of the intersection of the connecting line and the reflection line. The processor is further configured to, based on determining that one or more stationary objects are within this region, determine whether one or more of the stationary objects are within a distance of a reflection point located at the intersection of a signal line between a second object and the main vehicle and the reflection line. The processor is further configured to, based on determining that one or more of the stationary objects are within this distance, determine the expected speed of the second object based on the first object and the reflection line, and determine whether the speed of the second object is within a differential speed range. The processor is also configured to determine, based on the rate of difference between the speed of the second object and the expected speed, whether the second object is a ghost object, and output an indication that the second object is a ghost object.

[0004] The techniques and methods can be performed by the aforementioned system, another system or component, or a combination thereof. Some aspects described below include a method that includes receiving information about multiple objects approaching a primary vehicle. The objects may include two or more moving objects and one or more stationary objects. The method also includes identifying one or more pairs of moving objects. The method further includes, for each of the moving object pairs, determining a reflection line perpendicular to and between the first and second objects in the respective moving object pair. The method also includes determining whether one or more stationary objects are within a region at the intersection of the connecting line and the reflection line. The method further includes, based on determining that one or more stationary objects are within this region, determining whether one or more of the one or more stationary objects are within a distance of a reflection point located at the intersection of a signal line between a second object and the primary vehicle and the reflection line. The method further includes, based on determining that one or more of the one or more stationary objects are within this distance, determining the expected speed of the second object based on the first object and the reflection line, and determining whether the speed of the second object is within the expected speed difference rate. The method further includes, based on the second object's speed being within the expected speed difference rate, determining that the second object is a ghost object and outputting an indication that the second object is a ghost object.

[0005] Components may include a computer-readable medium (e.g., a non-transient storage medium) including instructions that, when executed by the system described above, another system or component, or a combination thereof, implement the methods described above and other methods. Some aspects described below include a computer-readable storage medium including instructions that, when executed, cause at least one processor to receive information about a plurality of objects near a main vehicle. The objects may include two or more moving objects and one or more stationary objects. The instructions also cause the processor to identify one or more pairs of moving objects. The instructions further cause the processor to, for each of the pairs of moving objects, determine a reflection line perpendicular to and between a first and a second object in the respective pair of moving objects. The instructions further cause the processor to determine whether one or more of the stationary objects are within a region at the intersection of the connecting line and the reflection line. The instructions further cause the processor to, based on the determination that one or more of the stationary objects are within this region, determine whether one or more of the one or more stationary objects are within a distance of a reflection point located at the intersection of a signal line between a second object and the main vehicle and the reflection line. The instruction also causes the processor to determine, based on the first object and the reflected ray, the expected velocity of the second object within the distance, and whether the velocity of the second object is within the rate of difference of the expected velocity. The instruction further causes the processor to determine, based on the second object's velocity being within the rate of difference of the expected velocity, that the second object is a ghost object and output an indication that the second object is a ghost object.

[0006] This invention provides a simplified concept for implementing ghost object detection, which is further described in the detailed embodiments and accompanying drawings. This invention is not intended to identify essential features of the claimed subject matter, nor is it intended to define the scope of the claimed subject matter. Attached Figure Description

[0007] The following figures illustrate systems and techniques for implementing ghost object detection, and these figures use some of the same numbers throughout to represent examples of analogous or similar functions and components.

[0008] Figure 1A and Figure 1B An example environment in which ghost object detection can be used according to the techniques of this disclosure is shown.

[0009] Figure 2 An example system of a primary vehicle configured to implement ghost object detection according to the technology of this disclosure is shown.

[0010] Figure 3 An example data stream for ghost object detection according to the techniques of this disclosure is shown.

[0011] Figure 4 The technology according to this disclosure is shown Figure 3 Further aspects of data flow.

[0012] Figure 5A and Figure 5B Example connecting lines and reflective lines according to the technology of this disclosure are shown.

[0013] Figure 6A and Figure 6B An example area of ​​a stationary object is shown according to the technology of this disclosure.

[0014] Figure 7A and Figure 7B An example reflection point according to the technique of this disclosure is shown.

[0015] Figure 8A and Figure 8B An example distance determination based on the technique of this disclosure is shown.

[0016] Figure 9A and Figure 9B An example of the expected speed according to the technology disclosed herein is shown.

[0017] Figure 10 An example method for ghost object detection according to the techniques of this disclosure is shown. Detailed Implementation

[0018] Overview

[0019] Environmental detection systems (e.g., RADAR and LiDAR) often struggle to distinguish ghost objects from real objects. For example, when a vehicle is parked near a wall, both the approaching vehicle and its ghost form (due to multipath reflections on the wall) can be detected. Many conventional techniques rely on complex systems (e.g., multiple sensors or fusion), specific algorithms (e.g., guardrail detection), or physically altering the environment (e.g., maneuvering the main vehicle) to identify ghost objects. Doing so reduces functionality while increasing cost.

[0020] The techniques and systems described in this paper implement ghost object detection. Specifically, a reflection line indicating a potential reflective surface between a first moving object and a second moving object is identified. If a sufficient number of stationary objects are within a region of the reflection line, it is determined whether one or more of the stationary objects within that region are within a distance of the reflection point. The expected velocity of the second object is then determined, and the expected velocity of the second object is checked relative to its velocity. If the expected velocity is close to the actual velocity, the second object is determined to be potentially a ghost object. By doing so, the system is able to effectively identify ghost objects in various environments, thereby allowing downstream operations to run as designed.

[0021] Example Environment

[0022] Figure 1A and Figure 1B Example environments 100 and 102 in which ghost object detection can be used are shown. Figure 1A Example environment 100 is shown, while Figure 1B Example environment 102 is shown. Example environments 100 and 102 contain a primary vehicle 104 and objects 106 detected by the primary vehicle 104. The primary vehicle 104 can be any type of system (car, sedan, truck, motorcycle, electric bicycle, boat, air vehicle, etc.). Object 106 can be any type of moving or stationary object (car, sedan, truck, motorcycle, electric bicycle, boat, pedestrian, cyclist, boulder, sign, wall, etc.). Object 106 is classified as a moving object or a stationary object. For example, objects 106-a and 106-b are moving objects (e.g., vehicles), while object 106-c is a stationary object (e.g., a wall).

[0023] In example environment 100, the main vehicle 104 is stationary (e.g., parked). In example environment 102, the main vehicle 104 is moving in reverse at a main speed 108. However, this disclosure is not limited to reverse movement, and the techniques described herein can be applied to the main vehicle 104 moving in any direction. Object 106 has an object speed 110. It can be assumed that the object speed of a stationary object (e.g., object 106-c) is zero.

[0024] To identify ghost objects, the main vehicle 104 includes a detection module 112 configured to identify which of the moving objects 106 are ghost objects 114 (e.g., not real objects). In example environment 100, object 106-b is a ghost of object 106-a created by reflection of object 106-c. In example environment 102, object 106-b is a ghost of the main vehicle 104 created by reflection of object 104-c. Therefore, in both example environments 100 and 102, object 106-b is ghost object 114.

[0025] Object 106, with or without ghost object 114, may be sent to or otherwise received by vehicle component 118. Vehicle component 118 may be any downstream operation, function, or system that uses information about object 106 to perform its function. Depending on the implementation, detection module 112 may output an indication of ghost object 114, remove ghost object 114 from object 106, output an indication of the real object of object 106, or some combination thereof.

[0026] Therefore, the detection module 112 is able to identify ghost objects 114 both when the main vehicle 104 is stationary and when it is moving, without any prior information about the environment and without sensor fusion or complex object trackers. In doing so, the detection module 112 can effectively identify ghost objects 114 in a variety of environments (and potentially filter them out), allowing downstream operations to run as designed.

[0027] Example System

[0028] Figure 2 An example system 200 is shown, configured to be situated within a main vehicle 104 and configured to perform ghost object detection. Components of the example system 200 may be arranged anywhere within or on the main vehicle 104. The example system 200 may include at least one processor 202, a computer-readable storage medium 204 (e.g., a medium, a set of media, or multiple media), and a vehicle component 116. These components are operatively and / or communicatively coupled via a link 208.

[0029] Processor 202 (e.g., application processor, microprocessor, digital signal processor (DSP), controller) is coupled to computer-readable storage medium 204 via link 208 and executes instructions (e.g., code) stored in computer-readable storage medium 204 (e.g., a non-transient storage device such as a hard disk drive, solid-state drive (SSD), flash memory, read-only memory (ROM)) to implement or otherwise cause detection module 112 (or a portion thereof) to perform the techniques described herein. Although shown as residing within computer-readable storage medium 204, detection module 112 may be a separate component (e.g., having a dedicated computer-readable storage medium including instructions and / or executing on dedicated hardware such as a dedicated processor, pre-programmed field-programmable gate array (FPGA), system-on-a-chip (SOC), etc.). The processor 202 and the computer-readable storage medium 204 can be any number of components, including multiple components distributed throughout the main vehicle 104, located remotely from the main vehicle 104, dedicated or shared with other components, modules or systems of the main vehicle 104, and / or configured in a manner different from that shown in the figures without departing from the scope of this disclosure.

[0030] The computer-readable storage medium 204 also contains sensor data 210 generated by one or more sensors or one or more types of sensors (not shown), which may be local or remote relative to the example system 200. The sensor data 210 indicates or otherwise enables the determination of information that can be used to perform the techniques described herein. For example, one or more sensors (e.g., RADAR, LiDAR) may generate sensor data 210 indicating information about object 106. The sensor data 210 can be used to determine other properties, as discussed below.

[0031] In some implementations, sensor data 210 may originate from a remote source (e.g., via link 208). Example system 200 may include a communication system (not shown) that receives sensor data 210 from a remote source.

[0032] Vehicle component 116 includes one or more systems or components communicatively coupled to detection module 112 and configured to perform vehicle functions using information about object 106 (e.g., about the real object, ghost object 114, or some combination thereof). For example, vehicle component 116 may include ADAS with means for accelerating, steering, or braking the main vehicle 104. Vehicle component 116 is communicatively coupled to detection module 112 via link 208. Although shown as a separate component, detection module 112 may be part of vehicle component 116, or vice versa.

[0033] Example data stream

[0034] Figure 3 and Figure 4 This is an example data stream 300 for ghost object detection. Example data stream 300 can be implemented in any of the environments previously described and can be implemented by any of the systems or components previously described. For example, example data stream 300 can be implemented in example environments 100, 102 and / or via example system 200. Example data stream 300 can also be implemented in other environments, by other systems or components, and using other data streams or technologies. Example data stream 300 can be implemented by one or more entities (e.g., detection module 112). The order of operations shown and / or described is not intended to be construed as limiting and can be rearranged without departing from the scope of this disclosure. Furthermore, any number of operations can be combined with any other number of operations to implement the example data stream or alternative data streams.

[0035] Example data stream 300 begins with attribute 302 of the environment (e.g., example environments 100, 102) obtained by detection module 112. As shown, attribute 302 includes objects 106, each including its corresponding object coordinates 304, object velocity 110, and (multiple) sensors 306. Sensors 306 are indications from vehicle sensors that have detected the corresponding objects. For example, often, object 106 can be detected by multiple sensors located at corresponding positions relative to the main vehicle 104. Object coordinates 304 can be absolute coordinates (e.g., relative to the Earth) or relative to the main vehicle 104. For example, object coordinates 304 can be latitude and longitude coordinates, distance and azimuth coordinates, lateral and longitudinal coordinates, or any other information that enables the main vehicle 104 to determine at least two-dimensional position of object 106 relative to the main vehicle 104 (e.g., position within or relative to the vehicle coordinate system (VCS)). The object velocity 110 may have a corresponding speed and direction, or speed vector components (e.g., lateral and longitudinal speeds). Attribute 302 also includes a master velocity 108 and a sensor position 308, which indicates the physical position of the corresponding sensor 306.

[0036] Attribute 302 can be acquired, received, or determined by detection module 112. For example, detection module 112 can determine attribute 302 directly from sensor data 210, from a bus or interface connected to a sensor (e.g., sensor 306) coupled to example system 200, or from another module or system of example system 200. Regardless of how or where attribute 302 is collected, received, exported, or calculated, detection module 112 is configured to use attribute 302 to determine which of the objects 106 (if any) are ghost objects 114.

[0037] For this purpose, attribute 302 is input into object pairing module 310. Object pairing module 310 is configured to generate moving object pairs 312 from objects 106. Object pairing module 310 can first determine which of objects 106 are moving objects and which are stationary objects. Subsequently, object pairing module 310 can generate moving object pairs 312 from each combination of two moving objects in response to determining that the main vehicle 104 is stationary. Object pairing module 310 can also generate moving object pairs 312 from each combination of the main vehicle 104 and moving object 104 in response to determining that the main vehicle 104 is moving in reverse. Moving object pairs 312 contain corresponding first and second objects, with the second object located at a greater distance from the main vehicle 104. In the case where the main vehicle 104 is moving in reverse, the first object is the main vehicle 104. The second object is a potential ghost object.

[0038] As part of generating moving object pairs 312, object pairing module 310 can look for moving objects that share the same sensor 306 (e.g., they are in the same field of view of one or more of the sensors 306). If the main vehicle 104 is moving in reverse, such a determination may not be made (e.g., all moving objects in this scenario would satisfy the constraint). Object pairing module 310 can also look for distance differences between certain values, such as the distance to the second object minus the distance to the first object, between 0.5 and 20 meters. In some implementations, the distance difference can be the distance between the corresponding objects (rather than the difference in distance to the main vehicle 104 or a combination of the difference in distance to the main vehicle 104). If the main vehicle 104 is moving in reverse, the distance difference can be the corresponding distance of the moving object (e.g., because the main vehicle 104 has a position at the origin). Object pairing module 310 can also look for moving objects with an object speed 110 greater than a movement rate threshold (e.g., object speed 110 is greater than 0.4 meters per second). If one or more of these criteria are not met for a pair of objects (e.g., if the main vehicle 104 is stationary, then both objects in object 106, or if the main vehicle 104 is moving in reverse, then one of the main vehicle 104 and object 106), the object pairing module 310 may not consider the pair of objects to be a moving object pair 312. Furthermore, if (e.g., based on one or more of the criteria) a suitable object pair cannot be found, the process may wait until the next loop (e.g., if none of the objects 106 in the current loop can be identified as ghost objects 114).

[0039] Subsequently, for each moving object pair 312, the reflection line module 314 generates a connecting line 316 and a reflection line 318. Connecting line 316 is a line passing through the centroids of the first and second objects in the corresponding moving object pair 312. If the main vehicle 104 is moving in reverse, connecting line 316 can pass through the origin of the VCS to the centroid of the second object in the moving object pair 312. Reflection line 318 is a line that passes through the midpoint of the segment of connecting line 316 between the first and second points and is perpendicular to connecting line 316. Typically, if the second object in the moving object pair 312 is a ghost object 114, then reflection line 318 represents a potential reflective surface. Details of connecting line 316 and reflection line 318 are shown in Figure 5.

[0040] Next, for the moving object pair 312, the region module 320 determines whether any stationary object among the stationary objects is within the region of the connecting line 316 and the reflection line 318. This region may be rectangular and aligned with the connecting line 316 and the reflection line 318. Stationary objects within this region become region stationary objects 322. In some implementations, the region module 320 may, in response to determining that multiple stationary objects among the stationary objects are within this region (e.g., five), only stationary objects within that region are designated as region stationary objects 322. If one or more of the stationary objects (depending on the implementation) are not within this region, region stationary objects 322 are not designated, and the process may move to the next moving object pair 312 (e.g., the second object is not considered ghost object 114). Details of this region are shown in Figure 6.

[0041] Subsequently, reflection point module 324 determines reflection point 326, which indicates a potential reflection point for ghost detection of the second object. Reflection point 326 may be located at the intersection of the signal line and reflection line 318, which passes through the centroid of the second object and the position of the main vehicle 104. The position on the main vehicle 104 may be a virtual sensor position, which is located at the average position (relative to VCS) between the respective sensors 306 of the first and second objects. The virtual sensor position may be the sensor position if only one sensor detects the first and second objects (or if only one sensor detects the second object in the case that the main vehicle 104 is moving in reverse). If reflection point 326 cannot be determined (e.g., the signal line and reflection line 318 are parallel), the process may move to the next moving object pair 312 (e.g., the second object is not considered ghost object 114). Details of reflection point 326 are shown in Figure 7.

[0042] Next, distance module 328 generates distance determination 330 (yes / no) based on whether any of the area stationary objects 322 are within a distance of reflection point 326 (e.g., within 3 meters of reflection point 326). If any of the area stationary objects 322 are within that distance of reflection point 326, distance determination 330 is "yes". If no area stationary object in area stationary objects 322 is within that distance of reflection point 326, distance determination 330 is "no". If distance determination 330 is "no", the process can move to the next moving object pair 312 (e.g., the second object is not considered ghost object 114). Details of this distance are shown in Figure 8.

[0043] Subsequently, the expected speed module 332 generates the expected speed 334 for the second object. If the second object is the ghost of the first object or the main vehicle 104, the expected speed 334 indicates the expected speed of the second object. Details of the expected speed 334 are shown in Figure 9.

[0044] Next, the ghost determination module 336 generates a ghost determination 338 (yes / no) based on the object velocity 110 of the second object and the expected velocity 334. The ghost determination module 336 can determine whether the rate of the object velocity 110 of the second object and / or the two vector rates are within the rate of the expected velocity 334 and / or the difference rate of the two vector rates. Equation 1 shows two example conditions indicating this comparison, one or both of which can be used to generate the ghost determination 338.

[0045] Condition 1:AbsSpeed egv -Speed ov )≤ds

[0046] Condition 2:

[0047]

[0048] Speed egv The expected speed is 334. Speed ov The second object's velocity is 110, ds is the difference rate and can be based on the distance between the first and second objects (e.g., 0.04 * distance), LongSpeed egv This is the expected longitudinal speed of 334, LongSpeed ov The longitudinal speed of the second object is 110, LatSpeed. egv It is the expected lateral velocity of 334, and LatSpeed ov The lateral velocity is the object velocity 110 of the second object. The longitudinal velocity and lateral velocity are examples of two vector velocities. Other vector components of velocity may be used without departing from the scope of this disclosure.

[0049] If the ghost determines 338 as "No" (e.g., one or both conditions are not met), the process can move to the next moving object pair 312 (e.g., the second object is not considered ghost object 114). If the ghost 338 is determined as "Yes," then the second object in the corresponding moving object pair 312 is likely to be ghost object 114. The above process can be repeated for other moving object pairs 312 to generate indications of possible ghost objects for the current frame. It should be noted that the first object in one moving object pair 312 can be the second object in another moving object pair 312.

[0050] Then, the ghost probability module 340 can track the probability 342 of object 106. Probability 342 indicates the probability that object 106 is a ghost object 114. In some implementations, only moving objects can have probability 342. To calculate probability 342, the indications of possible ghost objects in the current frame (e.g., all second objects in the current frame with a ghost determination 338 of "yes") are used to update the current probability 342 of the corresponding object. Object 106 or a moving object can start with a probability of 0.5 (e.g., in the first detection frame). Probability 342 can be updated according to Equation 2.

[0051] p new =(1-lpf) α )*p current +lpf α If the ghost determines 338 as "yes", otherwise

[0052] p new = 0.996 * p current (2)

[0053] Where p new The new probability of the corresponding object 106 is 342, p current The current probability of the corresponding object 106 is 342, and the LPF is... α It is a constant (e.g., 0.3).

[0054] Next, the object classification module 344 can generate an indication of which objects among the objects 106 are ghost objects 114. The object classification module 344 can generate this indication when the probability 342 reaches a probability threshold (e.g., 0.8). By using the probability and tracking the change of ghost determination 338 over time, the object classification module 344 can more reliably determine ghost objects 114. As discussed above, the indication of objects 106 (e.g., real objects, ghost objects 114, or a combination thereof) is sent to or received by the vehicle component 116 for downstream operations.

[0055] By using the aforementioned techniques, detection module 112 is able to identify ghost objects 114 both when the main vehicle 104 is stationary and when it is moving, without any prior information about the environment and without sensor fusion or complex object trackers. In doing so, detection module 112 can effectively identify ghost objects 114 in various environments (and potentially filter them out), allowing downstream operations to run as designed.

[0056] Example calculation

[0057] Figure 5AFigure 9 illustrates an example aspect of the process discussed above. The example illustrations correspond to the example environments of Figure 1. For example, example illustrations 500 and 502 correspond to example environments 100 and 102, respectively, and example illustrations 600 and 602 correspond to example environments 100 and 102, respectively, and so on. However, the following is not limited to these two environments. Object 106 can be in any orientation relative to the main vehicle 104 and have any object speed 110 without departing from the scope of this disclosure.

[0058] Figure 5A and Figure 5B Example connecting line 316 and reflective line 318 are shown at example figures 500 and 502, respectively. The main vehicle 104 has a VCS 504. The VCS may have a longitudinal axis and a lateral axis, and an origin at the center of the front bumper of the main vehicle 104. Different axes and / or origins may be used without departing from the scope of this disclosure. Connecting line 316 is shown passing through the centroids of the first and second objects (e.g., objects 106-a and 106-b) in figure 500. Connecting line 316 is shown passing through the origin of VCS 504 and the centroid of the second object (e.g., object 106-b) in figure 502. Objects 106-a and 106-b have object coordinates 304 (e.g., lateral / longitudinal) relative to VCS 504. Object coordinates 304 are used to determine connecting line 316. The reflection ray 318 is located at the midpoint of a segment of the connecting line 316 between two objects (e.g., the centroids of objects 106-a and 106-b, or the origin of VCS 504 and the centroid of 106-b) and is perpendicular to the connecting line. The reflection ray 318 can be given by Equation 3.

[0059] a1*longitudinal + b1*lateral + c1 = 0 (3)

[0060] Where a1 = 2 (vertical) b -Vertical a b1 = 2 (lateral) b - Lateral a c1 = (vertical) a 2 + Lateral a 2 ) - (vertical) b 2 + Lateral b 2 ).

[0061] Figure 6A and Figure 6BExample static object 322 is shown in example illustrations 600 and 602 respectively. Region 604 (used to define static object 322) can be centered at the intersection of connecting line 316 and reflection line 318. Region 604 can also be rectangular and oriented such that its length is parallel to reflection line 318 and its width is parallel to connecting line 316. The length can be on the order of five times the width. For example, the length could be 20 meters and the width could be 4 meters.

[0062] To determine whether a stationary object is within region 604, each of the stationary objects can be projected onto the connecting line 316 along with the first and second objects. The projected stationary object should be within half the width of region 604, which is the midpoint between the projected first and second objects. Each of the stationary objects can also be projected onto the reflection line 318 along with the first and second objects (this will be a single point because the connecting line 316 and the reflection line 318 are perpendicular). The projected stationary object should be within half the length of region 604, which is the region between the projected first and second objects. Projecting object 106 onto the connecting line 316 and the reflection line 318 is only one technique for determining whether a stationary object is within region 604. Other techniques (e.g., coordinate transformation, axis rotation, graphical analysis) can be used without departing from the scope of this disclosure.

[0063] In example diagrams 600 and 602, object 106-c is within region 604 because object 106-c... c It is a reflection object (such as) Figure 1A or Figure 1B (As described in the text). Although a single point is shown, object 106-c may provide multiple points, any number of which are within region 604. For example, object 106-c may include ten objects 106, five of which are within region 604. Again, any stationary object within region 604 becomes region stationary object 322 (e.g., object 106-c).

[0064] Figure 7A and Figure 7BExample reflection point 326 is shown at example illustrations 700 and 702, respectively. To determine reflection point 326, signal line 704 is generated. Signal line 704 passes through the centroid of the second object (e.g., object 106-b) and a virtual sensor position 706. Virtual sensor position 706 is based on the average position (relative to VCS 504) of the corresponding sensors 306 that detected the second object. Therefore, the lateral coordinate of virtual sensor position 706 can be the sum of the lateral coordinates of the sensor positions 308 of the sensors 306 that detected the second object divided by 2, and the longitudinal coordinate of virtual sensor position 706 can be the sum of the longitudinal coordinates of the sensors 306 that detected the second object divided by 2. In example illustration 700, the second object is detected by sensors 306-a and 306-b, and in example illustration 702, the second object is detected by sensors 306-b and 306-c. If only one sensor 306 detects the second object, the virtual sensor position 706 can be the sensor position 308 of sensor 306.

[0065] Signal line 704 can be determined by Equation 4.

[0066] a2*longitudinal + b2*lateral + c2 = 0 (4)

[0067] Where, a2 = 2 (vertical) b -Vertical vsl b2 = 2 (lateral) b - Lateral vsl c2 = (vertical) vsl 2 + Lateral vsl 2 )-(vertical) b 2 + Lateral b 2 ), longitudinal vsl It is the longitudinal coordinate of the virtual sensor position 706, and the lateral coordinate. vsl It is the lateral coordinate of the virtual sensor position 706.

[0068] Reflection point 326 is located at the intersection of signal line 704 and reflection line 318. If abs(b1*a2-a1*b2)<0.001, reflection point 326 may not be determined and the process may move to the next moving object pair 312 (e.g., the second object may not be ghost object 114). If abs(b1*a2-a1*b2)≥0.001, reflection point 326 can be determined by Equation 5.

[0069]

[0070]

[0071] rp long It is the longitudinal coordinate of reflection point 326, and rp lat These are the lateral coordinates of reflection point 326.

[0072] Figure 8A and Figure 8B Example distance determination 330 is shown in example illustrations 800 and 802, respectively. Distance 804 can be a fixed constant (e.g., three meters). To determine whether any regional stationary object in the regional stationary object 322 is within distance 804, the relative distance between the regional stationary object 322 and the reflection point 326 can be determined. If any regional stationary object in the regional stationary object 322 is within distance 804 of the reflection point 326, then distance determination 330 will be "yes". In example illustrations 800 and 802, object 106-c is within distance 804; therefore, distance determination 330 for object 106-c is "yes".

[0073] Figure 9A and Figure 9B Example expected speed 334 is shown at example illustrations 900 and 902 respectively. Expected speed 334 can be determined by Equation 6.

[0074] egv long =(va long *cosf(θ)+va lat *sinf(θ))*cosf(-θ)-(-va long *sinf(θ)+va lat *cosf(θ))*sinf(-θ)

[0075] egv lat =-(va long *cosf(θ)+va lat *sinf(θ))*sinf(-θ)-(-va long *sinf(θ)+va lat *cosf(θ))*cosf(-θ) (6)

[0076] Among them egv long It is the longitudinal component of the expected velocity 334, egv lat It is the lateral component of the expected velocity 334, va long It is the longitudinal component of the object velocity 110 of the first object (e.g., object velocity 110-a or main velocity 108), va let θ is the lateral component of the object velocity 110 of the first object, and θ is the angle 904 of the reflected ray 318 relative to the lateral axis of VCS 504.

[0077] Because object 106-b is a ghost object 114, the expected velocity 334 is roughly aligned with the object velocity 110-b. Therefore, object 106-b can be designated as ghost object 114 and either sent to vehicle component 116 or filtered out from objects 106 sent to vehicle component 116.

[0078] By using the above techniques, the main vehicle 104 can accurately determine which objects among the objects 106 are ghost objects 114, whether stationary or in motion, without any other knowledge of the environment surrounding the main vehicle 104 (e.g., the presence of guardrails). Furthermore, the main vehicle 104 can perform ghost object detection without unnecessary movement or multiple sensors 306 (although multiple sensors 306 have been described as detecting a second object, the process works similarly if a second object is detected by only a single sensor 306). Therefore, the main vehicle 104 can quickly and efficiently identify ghost objects 114 in a wide range of environments.

[0079] Example Method

[0080] Figure 10 This is an example method 1000 for ghost object detection. Example method 1000 can be implemented in any of the environments previously described, by any of the systems or components previously described, and by utilizing any of the data flows, process flows, or techniques previously described. For example, it can be implemented in... Figure 1A and Figure 1B In the example environment, example method 1000 is implemented by example system 200, by following example data flow 300, and / or as illustrated in the example diagrams of Figures 5-9. Example method 1000 can also be implemented in other environments, by other systems or components, and using other data flows, process flows, or techniques. Example method 1000 can be implemented by one or more entities (e.g., detection module 112). The order of operations shown and / or described is not intended to be construed as limiting, and the order can be rearranged without departing from the scope of this disclosure. Furthermore, any number of operations can be combined with any other number of operations to implement the example process flow or alternative process flows.

[0081] At 1002, information about multiple objects near the main vehicle is received. For example, the object pairing module 310 may receive attributes 302, including object 106 and its corresponding attributes (e.g., object coordinates 304, object speed 110, and sensor 306).

[0082] At 1004, a reflection ray indicating a potential reflective surface between the first moving object and the second moving object is determined. For example, the reflection ray module 314 may determine the reflection ray 318 based on a portion of the connecting line 316 between the first and second moving objects.

[0083] At position 1006, one or more stationary objects are identified within a region of the reflected light that is close to the first and second objects. For example, region module 320 may determine the presence of one or more stationary objects 322 in the region. In some implementations, region module 320 may determine the presence of one or more stationary objects 322 in the region.

[0084] At 1008, one or more of the stationary objects are determined to be within a distance of a potential reflection point on the indicator reflection line. For example, distance module 328 can determine that one or more of the stationary objects 322 in the region are within a distance 804 of the reflection point 326 and provide a distance determination 330 as "yes".

[0085] At 1010, the expected velocity of the second object is determined based on the first object and the reflected ray. For example, the expected velocity module 332 can determine the expected velocity 334 based on the object velocity 110 of the second object and the reflected ray 318.

[0086] At 1012, based on the fact that the velocity of the second object is within the rate of difference of the expected velocity, it is determined that the second object is a ghost object. For example, the ghost determination module 336 may determine that one or more rates of the object velocity 110 of the second object are within the rate of difference of one or more rates of the expected velocity 334, and provide a ghost determination 338 as "yes". In response to the ghost determination 338 as "yes", the second object may be indicated as ghost object 114 (e.g., by the object classification module 344). In some implementations, the ghost determination 338 may be used by the ghost probability module 340 to update the current probability 342 of the second object (e.g., the current probability 342 of the second object from a previous frame). The object classification module 344 may then indicate that the second object is ghost object 114 in response to determining that the probability 342 of the second object exceeds a threshold.

[0087] At 1014, an indication is output that the second object is a ghost object. For example, the object classification module 344 may output an indication of ghost object 114, a real object (e.g., object 106 other than ghost object 114), or a combination thereof.

[0088] By using example method 1000, ghost objects in the environment of the main vehicle can be identified efficiently and effectively. This allows downstream operations to be performed multiple times with increased safety and reliability, as expected.

[0089] Example

[0090] Example 1: A method comprising: receiving information about a plurality of objects approaching a primary vehicle, the objects including two or more moving objects and one or more stationary objects; identifying one or more pairs of moving objects; and for each of the pairs of moving objects: determining a reflection line that is perpendicular to a connecting line connecting a first object and a second object in the respective pair of moving objects; and between the first object and the second object; determining whether one or more of the stationary objects are within a region at the intersection of the connecting line and the reflection line; and based on determining that one or more of the stationary objects are within this region: determining whether one or more of the stationary objects are within a distance of a reflection point located at the intersection of a signal line between a second object and the primary vehicle; and the reflection line; and based on determining that one or more of the stationary objects are within this distance: determining the expected speed of the second object based on the first object and the reflection line; determining whether the speed of the second object is within a rate of difference of the expected speed; and based on the speed of the second object being within a rate of difference of the expected speed: determining that the second object is a ghost object; and outputting an indication that the second object is a ghost object.

[0091] Example 2: The method of Example 1, where the main vehicle is stationary.

[0092] Example 3: The method of Example 1 or 2, where: the main vehicle is moving in reverse; and the first object is the main vehicle.

[0093] Example 4: The method of Example 1, 2 or 3, wherein determining the moving object pair includes determining the moving object pair as follows: having a distance difference within the span; and having a rate greater than the moving rate threshold.

[0094] Example 5: Any of the methods in the preceding examples, wherein the region comprises a rectangular region that is parallel to the reflection line; has a length centered on the connecting line; and has a width centered on the reflection line, the width being less than the length.

[0095] Example 6: Any of the methods in the preceding examples, wherein determining whether one or more of the stationary objects are in the region includes: projecting the stationary objects together with the first and second objects onto the connecting line; and projecting the stationary objects onto the reflection line.

[0096] Example 7: Any of the methods in the preceding examples, where the length is approximately five times the width.

[0097] Example 8: Any of the methods in the preceding examples, wherein determining whether one or more of the stationary objects are in the region includes determining that multiple stationary objects are in the region.

[0098] Example 9: Any of the methods in the preceding examples, wherein the signal line is between a second object and a virtual sensor, the virtual sensor comprising the centroid of one or more sensors that detect the second object.

[0099] Example 10: Any of the methods in the preceding examples, where the information is received from a RADAR or LIDAR system, module, or component.

[0100] Example 11: Any of the methods in the preceding examples, where the expected speed is further based on the speed of the first object and the angle of the reflected ray.

[0101] Example 12: Any of the methods in the preceding examples, wherein determining whether the velocity of the second object is within the rate of difference of the expected velocity includes: determining whether the velocity of the second object is within the rate of difference of the expected velocity; and determining whether the velocity of the second object in both directions is within the rate of difference of the expected velocity in both directions.

[0102] Example 13: Any of the methods in the foregoing examples further includes updating the probability that the second object is a ghost object based on the rate of difference between the speed of the second object and the expected speed, wherein the indication is based on probability.

[0103] Example 14: A method comprising: receiving information about a plurality of objects approaching a primary vehicle, the objects including first and second moving objects and one or more stationary objects; determining a reflection ray indicating a potential reflective surface between the first and second moving objects; determining that one or more of the stationary objects are within a region of the reflection ray near the first and second objects; determining that one or more of the stationary objects are within a distance of a reflection point indicating a potential reflection point on the reflection ray; determining an expected speed of the second object based on the first object and the reflection ray; determining that the second object is a ghost object based on a rate of difference between the speed of the second object and the expected speed; and outputting an indication that the second object is a ghost object.

[0104] Example 15: A system comprising at least one processor configured to: receive information about a plurality of objects approaching a primary vehicle, the objects including two or more moving objects and one or more stationary objects; determine one or more pairs of moving objects; and for each of the pairs of moving objects: determine a reflection line perpendicular to a connecting line connecting a first and a second object in the respective pair of moving objects; and between the first and second objects; determine whether one or more stationary objects are within a region at the intersection of the connecting line and the reflection line; and based on determining that one or more stationary objects are within this region: determine whether one or more of the one or more stationary objects are within a distance of a reflection point located at the intersection of a signal line between a second object and the primary vehicle; and the reflection line; and based on determining that one or more of the one or more stationary objects are within this distance: determine the expected speed of the second object based on the first object and the reflection line; determine whether the speed of the second object is within a rate of difference of the expected speed; and based on the speed of the second object being within a rate of difference of the expected speed: determine that the second object is a ghost object; and output an indication that the second object is a ghost object.

[0105] Example 16: The system of Example 15, where the main vehicle is stationary.

[0106] Example 17: The system of Example 15 or 16, where: the main vehicle is traveling in reverse; and the first object is the main vehicle.

[0107] Example 18: The system of Examples 15, 16 or 17, wherein determining whether one or more of the stationary objects are in the region includes: projecting the stationary objects together with the first and second objects onto the connecting line; and projecting the stationary objects onto the reflection line.

[0108] Example 19: A system of any of Examples 15-18, wherein a signal line is between a second object and a virtual sensor, the virtual sensor comprising the centroid of one or more sensors that detect the second object.

[0109] Example 20: A system of any of Examples 15-19, wherein the expected speed is further based on the speed of the first object and the angle of the reflected line relative to the main vehicle.

[0110] Example 21: A computer-readable storage medium comprising instructions, which, when executed, cause at least one processor to: receive information about a plurality of objects approaching a primary vehicle, the objects including two or more moving objects and one or more stationary objects; determine one or more pairs of moving objects; and for each of the pairs of moving objects: determine a reflection line that is perpendicular to a connecting line connecting a first object and a second object in the respective pair of moving objects; and between the first object and the second object; determine whether one or more stationary objects are within a region at the intersection of the connecting line and the reflection line; and based on determining that one or more stationary objects are within this region: determine whether one or more of the one or more stationary objects are within a distance of a reflection point located at the intersection of a signal line between a second object and the primary vehicle; and the reflection line; and based on determining that one or more of the one or more stationary objects are within this distance: determine the expected speed of the second object based on the first object and the reflection line; determine whether the speed of the second object is within a rate of difference of the expected speed; and based on the speed of the second object being within a rate of difference of the expected speed: determine that the second object is a ghost object; and output an indication that the second object is a ghost object.

[0111] Example 22: A system comprising: at least one processor configured to perform a method of any one of Examples 1-14.

[0112] Example 23: A computer-readable storage medium comprising instructions that, when executed, cause at least one processor to perform the method of any one of Examples 1-14.

[0113] Example 24: A system comprising means for performing the method of any one of Examples 1-14.

[0114] Example 25: A method executed by a system from any of Examples 15-20.

[0115] Example 26: A method included by the instructions in Example 21.

[0116] Conclusion

[0117] While various embodiments of the present disclosure have been described in the foregoing description and illustrated in the accompanying drawings, it should be understood that the present disclosure is not limited thereto, but can be practiced in various ways within the scope of the following claims. It will be apparent from the foregoing description that various modifications can be made without departing from the spirit and scope of the present disclosure as defined by the following claims.

[0118] Unless the context explicitly states otherwise, the use of "or" and grammatically related terms indicates an unrestricted, non-exclusive alternative. As used herein, the phrase referring to "at least one" of a list of items means any combination of those items, including a single member. For example, "at least one of a, b, or c" is intended to cover: a, b, c, ab, ac, bc, and abc, as well as any combination with multiple identical elements (e.g., aa, aaa, aab, aac, abb, acc, bb, bbb, bbb, cc, and ccc, or any other ordering of a, b, and c).

Claims

1. A method for ghost object detection, the method comprising: receiving information about a plurality of objects proximate to a host vehicle, the objects including two or more moving objects and one or more stationary objects; determining one or more pairs of moving objects of the moving objects; and for each of the pairs of moving objects: determining a reflection line that: is perpendicular to a connection line connecting a center of mass of a first object and a center of mass of a second object in the respective pair of moving objects; and passes through a midpoint between the center of mass of the first object and the center of mass of the second object; determining whether one or more stationary objects of the stationary objects are within a region of an intersection of the connection line and the reflection line; and based on determining that one or more stationary objects of the stationary objects are within the region: determining whether one or more of the one or more stationary objects are within a distance of a reflection point at an intersection of: a signal line between the center of mass of the second object and the host vehicle; and the reflection line; and based on determining that one or more of the one or more stationary objects are within the distance: determining an expected velocity of the second object based on a velocity of the first object and an angle of the reflection line relative to a lateral axis of a vehicle coordinate system (VCS) of the host vehicle; determining whether a velocity of the second object is within a disparity rate of the expected velocity; and based on the velocity of the second object being within the disparity rate of the expected velocity: determining that the second object is a ghost object; and outputting an indication that the second object is a ghost object.

2. The method of claim 1, wherein, the host vehicle is stationary.

3. The method of claim 1, wherein: the host vehicle is traveling in reverse; and the first object is the host vehicle.

4. The method of claim 1, wherein, determining the pairs of moving objects includes determining pairs of moving objects that: have a distance difference that is within a span; and have a velocity that is greater than a velocity threshold.

5. The method of claim 1, wherein, the region includes a rectangular region that: is parallel to the reflection line; has a length centered on the connection line; and has a width centered on the reflection line that is less than the length.

6. The method of claim 5, wherein, determining whether one or more stationary objects of the stationary objects are within the region includes: projecting the stationary objects onto the connection line with the first object and the second object; and projecting the stationary objects onto the reflection line.

7. The method of claim 5, wherein, the length is five times the width.

8. The method of claim 1, wherein, determining whether one or more stationary objects of the stationary objects are within the region includes determining that multiple stationary objects of the stationary objects are within the region.

9. The method of claim 1, wherein, the signal line is between the second object and a virtual sensor, the virtual sensor including a center of mass of one or more sensors that detect the second object.

10. The method of claim 1, wherein, the information is received from a RADAR or LIDAR system, module, or component.

11. The method of claim 1, wherein, determining whether the velocity of the second object is within the disparity rate of the expected velocity includes: determining whether a velocity of the second object is within the disparity rate of a velocity of the expected velocity; and determining whether a velocity of the second object in two directions is within a difference velocity of a velocity of the expected velocity in the two directions.

12. The method of claim 1, wherein, further comprising, based on the velocity of the second object being within the difference velocity of the expected velocity, updating a probability that the second object is a ghost object, wherein the indication is based on the probability.

13. A system for ghost object detection, the system comprising at least one processor configured to: receive information about a plurality of objects proximate to a host vehicle, the objects including two or more moving objects and one or more stationary objects; determine one or more pairs of moving objects of the moving objects; and for each of the pairs of moving objects: determine a reflection line: perpendicular to a connection line connecting a center of mass of a first object and a center of mass of a second object in a respective pair of moving objects; and passing through a midpoint between the center of mass of the first object and the center of mass of the second object; determine whether one or more stationary objects of the stationary objects are within a region of an intersection of the connection line and the reflection line; and based on determining that one or more stationary objects of the stationary objects are within the region: determine whether one or more of the one or more stationary objects are within a distance of a reflection point at an intersection of: a signal line between the center of mass of the second object and the host vehicle; and the reflection line; and based on determining that one or more of the one or more stationary objects are within the distance: determine an expected velocity of the second object based on a velocity of the first object and an angle of the reflection line relative to a lateral axis of a vehicle coordinate system (VCS) of the host vehicle; determine whether a velocity of the second object is within a difference velocity of the expected velocity; and based on the velocity of the second object being within the difference velocity of the expected velocity: determine that the second object is a ghost object; and output an indication that the second object is a ghost object.

14. The system of claim 13, wherein, the host vehicle is stationary.

15. The system of claim 13, wherein: the host vehicle is traveling in reverse; and the first object is the host vehicle.

16. The system of claim 13, wherein, determining whether one or more stationary objects of the stationary objects are within the region includes: projecting the stationary objects to the connection line with the first object and the second object; and projecting the stationary objects to the reflection line.

17. The system of claim 13, wherein, the signal line is between the second object and a virtual sensor, the virtual sensor including a center of mass of one or more sensors that detect the second object.

18. The system of claim 13, wherein, the expected velocity is further based on a velocity of the first object and an angle of the reflection line relative to the host vehicle.

19. A computer-readable storage medium comprising instructions that, when executed, cause at least one processor to: receive information about a plurality of objects proximate to a host vehicle, the objects including two or more moving objects and one or more stationary objects; determining one or more pairs of moving objects of the moving objects; and for each of the pairs of moving objects: determining a reflection line that: is perpendicular to a connection line connecting a center of mass of a first object and a center of mass of a second object in the respective pair of moving objects; and passes through a midpoint between the center of mass of the first object and the center of mass of the second object; determining whether one or more stationary objects of the stationary objects are within a region of an intersection of the connection line and the reflection line; and based on determining that one or more stationary objects of the stationary objects are within the region: determining whether one or more of the one or more stationary objects are within a distance of a reflection point at an intersection of: a signal line between the center of mass of the second object and the host vehicle; and the reflection line; and based on determining that one or more of the one or more stationary objects are within the distance: determining an expected velocity of the second object based on a velocity of the first object and an angle of the reflection line relative to a lateral axis of a vehicle coordinate system (VCS) of the host vehicle; determining whether a velocity of the second object is within a difference rate of the expected velocity; and based on the velocity of the second object being within the difference rate of the expected velocity: determining that the second object is a ghost object; and outputting an indication that the second object is a ghost object.

Images