Point cloud processing device and point cloud processing method

Abstract

A point cloud processing apparatus and a point cloud processing method are provided. The point cloud processing method includes: receiving a vertically scanned point cloud; performing a random sampling consensus algorithm on the vertically scanned point cloud to obtain a fitting plane; filtering a first ground point in the vertically scanned point cloud according to the fitting plane to update the vertically scanned point cloud; and generating a processed point cloud according to the updated vertically scanned point cloud and outputting the processed point cloud.

Description

[Technical Field]

[0001] This invention relates to a data processing technology, and more particularly to a point cloud processing apparatus and a point cloud processing method. [Previous Technology]

[0002] LiDAR systems have broad application prospects in fields such as autonomous driving. If a vehicle is equipped with only a single LiDAR (e.g., a top LiDAR), the LiDAR system may have many blind spots. To reduce blind spots and increase safety, the LiDAR system can use multiple LiDARs to capture multiple point clouds and perform multi-LiDAR sensor fusion to fuse the multiple point clouds. However, fusing multiple point clouds often results in an excessively large amount of data, which is not conducive to real-time data processing. In addition, different LiDARs in the LiDAR system may be set up on the vehicle in different ways, resulting in point cloud data with different characteristics from different LiDARs. If the same filtering algorithm is used for all point cloud data, the filtering algorithm may be inefficient. [Summary of the Invention]

[0003] The present invention provides a point cloud processing device and a point cloud processing method, which can filter ground points for point clouds with different characteristics.

[0004] A point cloud processing apparatus of the present invention includes a processor and a transceiver. The transceiver receives a vertically scanned point cloud. The processor is coupled to the transceiver and configured to perform: performing a random sampling consensus algorithm on the vertically scanned point cloud to obtain a fitting plane; filtering first ground points in the vertically scanned point cloud according to the fitting plane to update the vertically scanned point cloud; and generating a processed point cloud according to the updated vertically scanned point cloud, and outputting the processed point cloud through the transceiver.

[0005] In one embodiment of the present invention, the processor described above is configured to further perform: receiving a horizontal scan point cloud via a transceiver; performing ray ground filtering on the horizontal scan point cloud to filter out second ground points in the horizontal scan point cloud to update the horizontal scan point cloud; and performing data fusion on the updated vertical scan point cloud and the updated horizontal scan point cloud to generate a processed point cloud.

[0006] In one embodiment of the present invention, the above-mentioned ray ground filtering includes: obtaining a first point corresponding to a first distance and a second point corresponding to a second distance from a horizontally scanned point cloud, wherein the second distance is greater than the first distance; determining a first angle between the first point and a reference plane and a second angle between the second point and the reference plane; calculating the difference between the first angle and the second angle; and determining that the second point is a second ground point in response to the difference being less than a first threshold and the second angle being less than a second threshold.

[0007] In one embodiment of the present invention, the above-described horizontal scan point cloud corresponds to the field of view of the light source, wherein the processor is configured to further perform: obtaining a region in the field of view according to a single scanning direction of the light source; and obtaining a first point and a second point from the region.

[0008] In one embodiment of the present invention, the above-mentioned light source for generating horizontal scanning point clouds is a rotating light source, wherein the rotation axis of the rotating light source is perpendicular to the ground.

[0009] In one embodiment of the present invention, the processor described above is configured to further perform: performing occupancy estimation on the processed point cloud to generate an occupancy map; and generating a vehicle control signal based on the occupancy map, and outputting the control signal via a transceiver.

[0010] In one embodiment of the present invention, the processor described above is configured to further perform: concatenating vertically scanned point clouds and horizontally scanned point clouds to generate concatenated point clouds; performing object detection on the concatenated point clouds to generate bounding boxes; and generating control signals based on the bounding boxes and the occupancy map.

[0011] In one embodiment of the present invention, the above-described random sampling consensus algorithm includes: generating multiple fitting planes based on a vertically scanned point cloud, wherein the multiple fitting planes correspond to multiple point numbers respectively; and selecting a fitting plane from the multiple fitting planes in response to the first point number of the fitting plane being the maximum point number among the multiple point numbers.

[0012] In one embodiment of the present invention, the processor described above is configured to further perform: determining whether the distance between a first point in the vertically scanned point cloud and the fitted plane is less than a threshold; and in response to the distance being less than the threshold, determining that the first point is a first ground point.

[0013] In one embodiment of the present invention, the above-mentioned light source for generating vertical scanning point clouds is a rotating light source, wherein the rotation axis of the rotating light source is parallel to the ground.

[0014] A point cloud processing method of the present invention includes: receiving a vertically scanned point cloud; performing a random sampling consensus algorithm on the vertically scanned point cloud to obtain a fitting plane; filtering a first ground point in the vertically scanned point cloud according to the fitting plane to update the vertically scanned point cloud; and generating a processed point cloud according to the updated vertically scanned point cloud and outputting the processed point cloud.

[0015] In one embodiment of the present invention, the above-described point cloud processing method further includes: receiving a horizontally scanned point cloud; performing ray ground filtering on the horizontally scanned point cloud to filter out second ground points in the horizontally scanned point cloud to update the horizontally scanned point cloud; and performing data fusion on the updated vertically scanned point cloud and the updated horizontally scanned point cloud to generate a processed point cloud.

[0016] In one embodiment of the present invention, the step of performing ray ground filtering on a horizontally scanned point cloud to filter out a second ground point in the horizontally scanned point cloud to update the horizontally scanned point cloud includes: obtaining a first point corresponding to a first distance and a second point corresponding to a second distance from the horizontally scanned point cloud, wherein the second distance is greater than the first distance; determining a first angle between the first point and a reference plane and a second angle between the second point and the reference plane; calculating the difference between the first angle and the second angle; and determining that the second point is a second ground point in response to the difference being less than a first threshold and the second angle being less than a second threshold.

[0017] In one embodiment of the present invention, the above-mentioned horizontal scanning point cloud corresponds to the field of view of the light source, wherein the step of obtaining a first point corresponding to a first distance and a second point corresponding to a second distance from the horizontal scanning point cloud includes: obtaining a region in the field of view according to a single scanning direction of the light source; and obtaining the first point and the second point from the region.

[0018] In one embodiment of the present invention, the above-mentioned light source for generating horizontal scanning point clouds is a rotating light source, wherein the rotation axis of the rotating light source is perpendicular to the ground.

[0019] In one embodiment of the present invention, the above-described point cloud processing method further includes: performing occupancy estimation on the processed point cloud to generate an occupancy map; and generating a vehicle control signal based on the occupancy map, and outputting the control signal.

[0020] In one embodiment of the present invention, the step of generating a vehicle control signal based on an occupancy map includes: concatenating a vertically scanned point cloud and a horizontally scanned point cloud to generate a concatenated point cloud; performing object detection on the concatenated point cloud to generate a bounding box; and generating a control signal based on the bounding box and the occupancy map.

[0021] In one embodiment of the present invention, the step of performing a random sampling consensus algorithm on a vertically scanned point cloud to obtain a fitting plane includes: generating multiple fitting planes based on the vertically scanned point cloud, wherein the multiple fitting planes correspond to multiple point numbers respectively; and selecting a fitting plane from the multiple fitting planes in response to the first point number of the fitting plane being the maximum point number among the multiple point numbers.

[0022] In one embodiment of the present invention, the step of updating the vertical scan point cloud by filtering the first ground point in the vertical scan point cloud according to the fitting plane includes: determining whether the distance between the first point in the vertical scan point cloud and the fitting plane is less than a threshold; and determining the first point as the first ground point in response to the distance being less than the threshold.

[0023] In one embodiment of the present invention, the above-mentioned light source for generating vertical scanning point clouds is a rotating light source, wherein the rotation axis of the rotating light source is parallel to the ground.

[0024] Based on the above, the point cloud processing device of the present invention can use different algorithms to filter ground points in horizontal and vertical scan point clouds, thereby reducing the amount of point cloud data that the autonomous driving system needs to process.

Implementation Method

[0026] In order to make the contents of the present invention more readily apparent, the following specific embodiments are provided as examples on which the present invention can indeed be implemented. Furthermore, wherever possible, elements / components / steps using the same reference numerals in the drawings and embodiments represent the same or similar parts.

[0027] FIG1 illustrates a schematic diagram of a point cloud processing apparatus 100 according to an embodiment of the present invention. The point cloud processing apparatus 100 may include a processor 110, a storage medium 120, and a transceiver 130.

[0028] The processor 110 is, for example, a central processing unit (CPU), or other programmable general-purpose or special-purpose microcontroller (MCU), microprocessor, digital signal processor (DSP), programmable controller, application-specific integrated circuit (ASIC), graphics processing unit (GPU), image signal processor (ISP), image processing unit (IPU), arithmetic logic unit (ALU), complex programmable logic device (CPLD), field programmable gate array (FPGA), or other similar elements or combinations thereof. The processor 110 may be coupled to storage medium 120 and transceiver 130, and access and execute multiple modules and various applications stored in storage medium 120.

[0029] The storage medium 120 is, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk drive (HDD), solid state drive (SSD) or similar element or combination of the above elements, for storing multiple modules or various applications that can be executed by the processor 110.

[0030] Transceiver 130 transmits or receives signals wirelessly or via a wired connection. Transceiver 130 may also perform operations such as low-noise amplification, impedance matching, mixing, up or down frequency conversion, filtering, amplification, and similar operations. Processor 110 can communicate with external electronic devices via transceiver 130.

[0031] In one embodiment, the processor 110 can communicate with the light-transmitter via the transceiver 130 and receive point clouds captured by the light-transmitter. The point cloud processing device 100 can be a local device. For example, if the light-transmitter is mounted on a vehicle, the point cloud processing device 100 can be mounted in the vehicle. On the other hand, the point cloud processing device 100 can be a remote device. For example, the point cloud processing device 100 can communicate with the light-transmitter via a wireless network to receive information from the light-transmitter.

[0032] Depending on the arrangement of the LiDAR, the point cloud captured by the LiDAR may have different characteristics. Figure 2 illustrates a schematic diagram of a vehicle 200 and LiDAR according to an embodiment of the present invention. To implement an autonomous driving system or a collision avoidance system, the vehicle 200 may be equipped with multiple LiDARs. For example, a LiDAR 310 for detecting information about a large area of ​​the environment may be provided on the top of the vehicle 200. The LiDAR 310 may have blind spots for areas closer to the vehicle 200. Accordingly, one or more LiDARs 320 for blind filling may be provided on the sides of the vehicle 200. The LiDARs 320 can be used to detect information about the environment closer to the vehicle 200.

[0033] The LiDAR 310 may include a rotating LiDAR. The detection beam 311 of the LiDAR 310 may rotate around a rotation axis 312. Assume that the ground 500 where the vehicle 200 is located is parallel to the XY plane of the Cartesian coordinate system. The rotation axis 312 of the LiDAR 310 may be parallel to the Z-axis. That is, the rotation axis 312 may be perpendicular to the ground 500. The point cloud captured by the LiDAR 310 configured in the above manner may be a horizontally scanned point cloud.

[0034] The lidar 320 may include a rotating lidar. The detection beam 321 of the lidar 320 may rotate around a rotation axis 322. The rotation axis 322 of the lidar 320 is parallel to the XY plane. That is, the rotation axis 322 may be parallel to the ground 500. The point cloud captured by the lidar 320 configured in the above manner may be a vertically scanned point cloud.

[0035] FIG3 illustrates a flowchart of point cloud processing according to an embodiment of the present invention, wherein the point cloud processing steps can be implemented by the point cloud processing apparatus 100 shown in FIG1. ​​In step S301, the processor 110 can receive the complete horizontal scan point cloud from the light sensor 310 via the transceiver 130. The processor 110 can crop a portion of the horizontal scan point cloud for signal processing.

[0036] In step S302, the processor 110 can receive the complete vertical scan point cloud from the light-emitting diode 320 via the transceiver 130. The processor 110 can crop a portion of the vertical scan point cloud for signal processing.

[0037] In step S303, the processor 110 may perform time axis correction on the horizontal and vertical scan point clouds. The processor 110 aligns the time axes of the horizontal and vertical scan point clouds to ensure that the horizontal and vertical scan point clouds processed in subsequent steps correspond to the same time point.

[0038] In step S304, processor 110 may concatenate the vertical scan point cloud and the horizontal scan point cloud to generate a concatenated point cloud. Processor 110 may convert the vertical scan point cloud and the horizontal scan point cloud to the same coordinate system. Then, processor 110 may take the union of the vertical scan point cloud and the horizontal scan point cloud to generate the concatenated point cloud. In one embodiment, each point in the concatenated point cloud may have a corresponding label, wherein the label may indicate that the point belongs to the vertical scan point cloud or the horizontal scan point cloud.

[0039] In step S305, the processor 110 may obtain the horizontal scan point cloud from step S303 or from the concatenated point cloud. The processor 110 may perform ray ground filtering on the horizontal scan point cloud to filter ground points in the horizontal scan point cloud (e.g., points generated when scanning to the ground 500), thereby updating the horizontal scan point cloud to reduce the amount of data.

[0040] Specifically, the processor 110 can obtain a region in the field of view (FoV) of the light-emitting diode (LED) according to a single scanning direction of the LED. FIG4 illustrates a top view of the carrier 200 and the LED 310 according to an embodiment of the present invention. The LED 310 can capture a horizontal scan point cloud from the field of view 400. The processor 110 can obtain a region 410 in the field of view 400 according to a single scanning direction 40 of the LED 310, wherein the region 410 may be a fan-shaped region.

[0041] Next, the processor 110 can obtain points 41 and 42 from the horizontally scanned point cloud within region 410, wherein the distance between point 41 and the light source 310 can be less than the distance between point 42 and the light source 310. The processor 110 can determine the angle θ1 between point 41 and a reference plane (e.g., a fitted plane corresponding to ground 500 generated from the point cloud) (e.g., the angle formed by the reference plane, point 41, and light source 310), and can determine the angle θ2 between point 42 and the reference plane (e.g., the angle formed by the reference plane, point 42, and light source 310). The processor 110 can calculate the difference between angle θ1 and angle θ2.

[0042] If the difference between angles θ1 and θ2 is less than a first threshold and angle θ2 is less than a second threshold, it means that point 41 and point 42 may be on the same plane and point 42 may be very close to the ground 50°. Based on this, processor 110 can determine that point 42 is a ground point. Processor 110 can filter point 42 out of the horizontal scan point cloud to reduce the amount of data in the horizontal scan point cloud (or concatenated point cloud). If the difference between angles θ1 and θ2 is greater than or equal to the first threshold, or angle θ2 is greater than or equal to the second threshold, then processor 110 can determine that point 42 is not a ground point. Based on this, processor 110 may not filter point 42.

[0043] Returning to Figure 3, in step S306, the processor 110 can obtain the vertical scan point cloud from step S303 or from the concatenated point cloud. The processor 110 can perform a random sample consensus (RANSAC) algorithm on the vertical scan point cloud to filter ground points in the vertical scan point cloud, thereby updating the vertical scan point cloud to reduce the amount of data.

[0044] The processor 110 can obtain a fitting plane corresponding to the ground 500 based on the vertically scanned point cloud. Specifically, the processor 110 can generate multiple candidate fitting planes based on the vertically scanned point cloud, wherein each candidate fitting plane corresponds to a multiple number of points. The processor 110 can select the fitting plane with the maximum number of points from the multiple candidate fitting planes. The selected fitting plane is the fitting plane corresponding to the ground 500.

[0045] After selecting the fitting plane, the processor 110 can determine the distance between a point in the fitting plane and the fitting plane itself. If the distance between a point and the fitting plane is less than a threshold, it means that the point is very close to the ground. Accordingly, the processor 110 can determine that the point is a ground point and can filter it out from the vertical scan point cloud to reduce the amount of data in the vertical scan point cloud (or concatenated point cloud). If the distance between a point and the fitting plane is greater than or equal to the threshold, the processor 110 can determine that the point is not a ground point. Accordingly, the processor 110 can choose not to filter the point.

[0046] After obtaining the updated horizontal scan point cloud and the updated vertical scan point cloud, in step S307, the processor 110 may perform data fusion on the updated horizontal scan point cloud and the updated vertical scan point cloud to generate a processed point cloud. The processor 110 may output the processed point cloud through the transceiver 130. For example, the processor 110 may output the processed point cloud to an autonomous driving system or a collision avoidance system.

[0047] In step S308, the processor 110 may perform object detection on the concatenated point cloud to generate bounding boxes. The bounding boxes can be used to indicate objects in the environment (e.g., dynamic objects or non-environmental objects). In addition to the bounding boxes, the results of object detection may further include the object's type, location, size, or orientation.

[0048] In step S309, the processor 110 may perform occupancy estimation on the processed point cloud to generate an occupancy map. The occupancy map may contain point clouds used to indicate static objects or environmental objects (e.g., ground, walls, safety islands, or streetlights).

[0049] In step S310, the processor 110 may generate a control signal based on the object detection result (e.g., a bounding box) or the occupancy map, and output the control signal through the transceiver 130. For example, the processor 110 may output the control signal to the autonomous driving system of the vehicle 200 to manipulate the movement of the vehicle 200 through the control signal.

[0050] FIG. 5 illustrates a flowchart of a point cloud processing method according to an embodiment of the present invention, wherein the point cloud processing method can be implemented by the point cloud processing apparatus 100 shown in FIG. 1. In step S501, a vertically scanned point cloud is received. In step S502, a random sampling consensus algorithm is performed on the vertically scanned point cloud to obtain a fitting plane. In step S503, a first ground point in the vertically scanned point cloud is filtered according to the fitting plane to update the vertically scanned point cloud. In step S504, a processed point cloud is generated based on the updated vertically scanned point cloud, and the processed point cloud is output.

[0051] In summary, the point cloud processing device of the present invention can use different algorithms to filter ground points in horizontal and vertical scan point clouds. The point cloud processing device can use a random sampling consensus algorithm to perform ground point filtering on the vertical scan point cloud, and can use ray ground filtering to perform ground point filtering on the horizontal scan point cloud. The point cloud processing device can perform data fusion on the filtered horizontal and vertical scan point clouds to generate a processed point cloud. Compared with traditional methods, the point cloud processing device of the present invention can correctly filter ground points and significantly reduce the amount of data in the processed point cloud, thereby facilitating real-time processing of point cloud data by autonomous driving systems. [Simplified Explanation of the Diagram]

[0025] Figure 1 is a schematic diagram of a point cloud processing apparatus according to an embodiment of the present invention. Figure 2 is a schematic diagram of a vehicle and a light source according to an embodiment of the present invention. Figure 3 is a flowchart of point cloud processing according to an embodiment of the present invention. Figure 4 is a top view of a vehicle and a light source according to an embodiment of the present invention. Figure 5 is a flowchart of a point cloud processing method according to an embodiment of the present invention.

Claims

1. A point cloud processing device, comprising: Transceiver, receiving vertically scanned point clouds and horizontally scanned point clouds; The transceiver is coupled to a processor and configured to perform: performing a random sampling consensus algorithm on the vertical scan point cloud to obtain a fitting plane; filtering a first ground point in the vertical scan point cloud based on the fitting plane to update the vertical scan point cloud; performing ray ground filtering on the horizontal scan point cloud to filter a second ground point in the horizontal scan point cloud to update the horizontal scan point cloud; and performing data fusion on the updated vertical scan point cloud and the updated horizontal scan point cloud to generate a processed point cloud, and outputting the processed point cloud through the transceiver.

2. The point cloud processing apparatus as described in claim 1, wherein the ray ground filtering comprises: A first point corresponding to a first distance and a second point corresponding to a second distance are obtained from the horizontally scanned point cloud, wherein the second distance is greater than the first distance; Determine the first angle between the first point and the reference plane and the second angle between the second point and the reference plane; calculate the difference between the first angle and the second angle; And in response to the difference being less than a first threshold and the second included angle being less than a second threshold, the second point is determined to be the second ground point.

3. The point cloud processing apparatus of claim 2, wherein the horizontally scanned point cloud corresponds to the field of view of a light source, wherein the processor is configured to further perform: obtaining a region in the field of view according to a single scanning direction of the light source; and obtaining the first point and the second point from the region.

4. The point cloud processing apparatus as claimed in claim 1, wherein the light source for generating the horizontally scanned point cloud is a rotating light source, wherein the rotation axis of the rotating light source is perpendicular to the ground.

5. The point cloud processing apparatus of claim 1, wherein the processor is configured to further perform: performing occupancy estimation on the processed point cloud to generate an occupancy map; and generating a control signal for a vehicle based on the occupancy map, and outputting the control signal via the transceiver.

6. The point cloud processing apparatus of claim 5, wherein the processor is configured to further perform: concatenating the vertically scanned point cloud and the horizontally scanned point cloud to generate a concatenated point cloud; performing object detection on the concatenated point cloud to generate bounding boxes; and generating the control signal based on the bounding boxes and the occupancy map.

7. The point cloud processing apparatus as described in claim 1, wherein the random sampling consensus algorithm comprises: Multiple fitting planes are generated based on the vertically scanned point cloud, wherein the multiple fitting planes correspond to multiple numbers of points; And in response to the first number of points on the fitting plane being the maximum number of points among the plurality of point numbers, the fitting plane is selected from the plurality of fitting planes.

8. The point cloud processing apparatus of claim 7, wherein the processor is configured to further perform: determining whether the distance between a first point in the vertically scanned point cloud and the fitted plane is less than a threshold; and determining, in response to the distance being less than the threshold, that the first point is the first ground point.

9. The point cloud processing apparatus of claim 1, wherein the light source for generating the vertically scanned point cloud is a rotating light source, wherein the rotation axis of the rotating light source is parallel to the ground.

10. A point cloud processing method, comprising: The processor receives both the vertical and horizontal scan point clouds. The processor performs a random sampling consensus algorithm on the vertical scan point cloud to obtain a fitting plane; the processor filters a first ground point in the vertical scan point cloud based on the fitting plane to update the vertical scan point cloud; the processor performs ray ground filtering on the horizontal scan point cloud to filter a second ground point in the horizontal scan point cloud to update the horizontal scan point cloud; and the processor performs data fusion on the updated vertical scan point cloud and the updated horizontal scan point cloud to generate a processed point cloud, and outputs the processed point cloud.

11. The point cloud processing method as described in claim 10, wherein the step of performing the ray ground filtering on the horizontally scanned point cloud by the processor to filter the second ground points in the horizontally scanned point cloud to update the horizontally scanned point cloud includes: The processor obtains a first point corresponding to a first distance and a second point corresponding to a second distance from the horizontally scanned point cloud, wherein the second distance is greater than the first distance; the processor determines a first angle between the first point and a reference plane and a second angle between the second point and the reference plane; the processor calculates the difference between the first angle and the second angle. And in response to the difference being less than a first threshold and the second included angle being less than a second threshold, the processor determines that the second point is the second ground point.

12. The point cloud processing method of claim 11, wherein the horizontally scanned point cloud corresponds to a field of view of a light source, wherein the step of the processor obtaining from the horizontally scanned point cloud a first point corresponding to a first distance and a second point corresponding to a second distance includes: The processor obtains the region in the field of view according to the single scanning direction of the light source; The processor obtains the first point and the second point from the region.

13. The point cloud processing method as described in claim 10, wherein the light source used to generate the horizontally scanned point cloud is a rotating light source, wherein the rotation axis of the rotating light source is perpendicular to the ground.

14. The point cloud processing method as described in claim 10, further comprising: The processor performs occupancy estimation on the processed point cloud to generate an occupancy map; The processor generates control signals for the vehicle based on the occupancy map and outputs the control signals.

15. The point cloud processing method as described in claim 14, wherein the step of the processor generating the control signal of the vehicle based on the occupancy map includes: The processor connects the vertically scanned point cloud and the horizontally scanned point cloud in series to generate a concatenated point cloud. The processor performs object detection on the concatenated point cloud to generate bounding boxes; and the processor generates the control signal based on the bounding boxes and the occupancy map.

16. The point cloud processing method as described in claim 10, wherein the step of the processor performing the random sampling consensus algorithm on the vertically scanned point cloud to obtain the fitting plane includes: The processor generates multiple fitting planes based on the vertically scanned point cloud, wherein the multiple fitting planes correspond to multiple numbers of points; And in response to the first number of points on the fitting plane being the maximum number of points among the plurality of point numbers, the processor selects the fitting plane from the plurality of fitting planes.

17. The point cloud processing method as described in claim 16, wherein the step of the processor filtering the first ground points in the vertically scanned point cloud according to the fitted plane to update the vertically scanned point cloud includes: The processor determines whether the distance between the first point in the vertically scanned point cloud and the fitting plane is less than a threshold. And in response to the distance being less than the threshold, the processor determines that the first point is the first ground point.

18. The point cloud processing method as described in claim 10, wherein the light source used to generate the vertically scanned point cloud is a rotating light source, wherein the rotation axis of the rotating light source is parallel to the ground.

Images