Laser radar and detection method thereof

Abstract

The invention provides a laser radar which comprises a transmitting device and a detecting device, the transmitting device is configured to be capable of transmitting a detecting light beam for detecting an obstacle, the detecting device comprises a pixel array, an echo light beam after the detecting light beam is reflected on the obstacle can form a detecting area on the pixel array, and the pixel array comprises a plurality of pixels, each pixel can respond to an echo light beam and generate detection data; the storage device comprises a plurality of storage units which are distributed in an array mode, each storage unit can store detection data of a plurality of pixels, and the number of pixels corresponding to each storage unit is smaller than the number of pixels covered by the detection area. The detection data blocks of the pixels covered by the detection area are stored in a plurality of storage units of the storage device; and the data processing device is coupled with the detection device and the storage device and is configured to perform block processing on the detection data stored in the storage device. According to the laser radar disclosed by the invention, the detection data of the pixels covered by the detection area is subjected to block storage and block processing, so that the complexity and the calculation amount of data processing can be reduced, the calculation power consumption can be reduced, the data processing speed can be improved, the size of a data processing circuit can be reduced, and the miniaturization of the whole laser radar is facilitated.

Description

Technical Field

[0001] The present disclosure generally relates to the field of laser radar, and more particularly to a laser radar and a detection method of a laser radar. Background Art

[0002] LiDAR is a radar system that emits laser beams to detect the position, speed and other characteristic quantities of a target. It is an advanced detection method that combines laser technology with photoelectric detection technology. LiDAR is widely used in autonomous driving, transportation communications, drones, intelligent robots, resource exploration and other fields due to its advantages such as high resolution, good concealment, strong anti-active interference ability, good low-altitude detection performance, small size and light weight.

[0003] In order to improve the detection performance of LiDAR, LiDAR needs to be configured with a larger detection area. The traditional data processing circuit adapted to the small detection area will become larger as the detection area expands. The traditional data processing circuit only processes a single type of detection area, has a low degree of personalization, and becomes larger accordingly, which is not conducive to the miniaturization of the entire LiDAR. It will also lead to problems such as high data processing complexity, large amount of calculation, high system power consumption, and slow processing speed.

[0004] Therefore, how to design data processing circuits with better performance that are suitable for larger detection areas is a need for continuous improvement in the lidar field.

[0005] The contents of the background technology section are merely the technologies known to the inventors and do not necessarily represent the prior art in the field. Summary of the invention

[0006] In view of one or more problems existing in the prior art, the present disclosure provides a laser radar, which can reduce the complexity and amount of data processing, reduce computing power consumption, improve data processing speed, and reduce the size of the data processing circuit by storing and processing the detection data of pixels covered by the detection area in blocks, which is conducive to the miniaturization of the entire laser radar. The laser radar includes:

[0007] An emitting device and a detecting device, wherein the emitting device is configured to emit a detection beam for detecting an obstacle, the detecting device comprises a pixel array, an echo beam after the detection beam is reflected on the obstacle can form a detection area on the pixel array, the pixel array comprises a plurality of pixels, each pixel can respond to the echo beam and generate detection data;

[0008] A storage device, comprising a plurality of storage units distributed in an array, wherein each storage unit can store detection data of a plurality of pixels, and the number of pixels corresponding to each storage unit is less than the number of pixels covered by the detection area, so that the detection data of the pixels covered by the detection area are stored in blocks in the plurality of storage units of the storage device; and

[0009] The data processing device is coupled to the detection device and the storage device, and is configured to perform block processing on the detection data stored in the storage device.

[0010] According to one aspect of the present disclosure, the transmitting device is configured to emit different types of detection light beams for detecting obstacles, wherein the echo light beams of the different types of detection light beams after being reflected on the obstacles can form detection areas with different properties on the pixel array.

[0011] According to one aspect of the present disclosure, the number of pixels corresponding to the storage device is not less than the number of pixels covered by any of the detection areas.

[0012] According to one aspect of the present disclosure, the data processing device is configured to: store the detection data of the pixels covered by the detection area in blocks into corresponding multiple storage units according to the spatial correspondence between the pixel array and the storage unit array.

[0013] According to one aspect of the present disclosure, the data processing device comprises one or more convolution modules, and the convolution modules are configured to:

[0014] Performing block convolution processing on the detection data stored in the storage device, specifically including performing convolution processing block by block based on the storage unit to obtain a reconstructed pixel array; and

[0015] A laser radar point cloud is generated based on the recombined pixel array.

[0016] According to one aspect of the present disclosure, the data processing device is configured to: determine the boundary attributes of the first direction and the second direction of each storage unit, wherein the boundary attributes include convolution boundaries or non-convolution boundaries; and determine the detection data to be transmitted to the convolution module based on the boundary attributes of the first direction and the second direction of each storage unit.

[0017] According to one aspect of the present disclosure, when the boundary attribute of the first direction and / or the second direction of the storage unit is a non-convolution boundary, the data processing device is configured to: transmit the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction to the convolution module.

[0018] According to one aspect of the present disclosure, when the boundary attributes of the first direction and the second direction of the storage unit are both convolution boundaries, the data processing device is configured to: transmit the detection data stored in the storage unit to the convolution module.

[0019] According to one aspect of the present disclosure, the data processing device further includes an accumulator, which is configured to accumulate the detection data generated by the multiple echo light beams on each pixel pixel by pixel and store them in a storage unit corresponding to the pixel.

[0020] According to one aspect of the present disclosure, the storage unit is provided with an additional storage location for storing part of the detection data in other storage units adjacent to the storage unit.

[0021] According to one aspect of the present disclosure, the data processing device further includes: a control module and a cache module, wherein the control module is coupled to the storage device and is configured to: at least transmit the detection data stored in each storage unit to the cache module block by block; and transmit the detection data stored in the cache module to the convolution module.

[0022] According to one aspect of the present disclosure, the control module is configured to: at least transmit the detection data stored in each storage unit to the cache module block by block according to the boundary attributes of the first direction and the second direction of each storage unit;

[0023] When the boundary attribute of the storage unit in the first direction and / or the second direction is a non-convolution boundary, the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction are transmitted to the cache module; or

[0024] When the boundary attributes of the first direction and the second direction of the storage unit are both convolution boundaries, the detection data stored in the storage unit is transmitted to the cache module.

[0025] According to one aspect of the present disclosure, the detection areas with different properties include any one or more of the following: rectangular detection areas, discrete detection areas and strip detection areas.

[0026] The present disclosure also relates to a detection method of a laser radar, wherein the laser radar comprises a transmitting device, a detecting device, a storage device and a data processing device, wherein the detecting device comprises a pixel array, the transmitting device can emit a detection beam, and an echo beam after the detection beam is reflected on an obstacle can form a detection area on the pixel array, the pixel array comprises a plurality of pixels, each pixel can respond to the echo beam and generate detection data; the data processing device is coupled to the detecting device and the storage device; the detection method comprises:

[0027] S11: emits a detection beam to detect obstacles;

[0028] S12: responding to the echo light beam and generating detection data;

[0029] S13: storing the detection data of the pixels covered by the detection area in blocks into a plurality of storage units of the storage device, wherein the storage device comprises a plurality of storage units distributed in an array, each storage unit can store the detection data of a plurality of pixels, and the number of pixels corresponding to each storage unit is less than the number of pixels covered by the detection area, so as to store the detection data of the pixels covered by the detection area in blocks into the plurality of storage units of the storage device; and

[0030] S14: performing block processing on the detection data stored in the storage device.

[0031] According to one aspect of the present disclosure, step S11 includes: emitting different types of detection light beams for detecting obstacles, wherein the echo light beams of the different types of detection light beams after being reflected on the obstacles can form detection areas with different properties on the pixel array.

[0032] According to one aspect of the present disclosure, the number of pixels corresponding to the storage device is not less than the number of pixels covered by any of the detection areas.

[0033] According to one aspect of the present disclosure, the step S13 includes: storing the detection data of the pixels covered by the detection area in corresponding multiple storage units respectively according to the spatial correspondence between the pixel array and the storage unit array.

[0034] According to one aspect of the present disclosure, wherein the data processing device includes one or more convolution modules, the step S14 includes:

[0035] The detection data stored in the storage device is subjected to block-by-block convolution processing by the convolution module, specifically comprising performing convolution processing block by block in units of storage units to obtain a reconstructed pixel array; and

[0036] A laser radar point cloud is generated based on the recombined pixel array.

[0037] According to one aspect of the present disclosure, step S14 includes: determining boundary attributes of the first direction and the second direction of each storage unit, wherein the boundary attributes include convolution boundaries or non-convolution boundaries; and determining the detection data to be transmitted to the convolution module based on the boundary attributes of the first direction and the second direction of each storage unit.

[0038] According to one aspect of the present disclosure, the step of determining the detection data to be transmitted to the convolution module includes: when the boundary attribute of the first direction and / or the second direction of the storage unit is a non-convolution boundary, the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction are transmitted to the convolution module.

[0039] According to one aspect of the present disclosure, the step of determining the detection data to be transmitted to the convolution module further includes: when the boundary attributes of the first direction and the second direction of the storage unit are both convolution boundaries, transmitting the detection data stored in the storage unit to the convolution module.

[0040] According to one aspect of the present disclosure, the data processing device further includes an accumulator, and step S13 includes: accumulating the detection data generated by the multiple echo light beams on each pixel pixel by pixel, and storing them in a storage unit corresponding to the pixel.

[0041] According to one aspect of the present disclosure, the storage unit is provided with an additional storage location for storing part of the detection data in other storage units adjacent to the storage unit.

[0042] According to one aspect of the present disclosure, the data processing device further includes: a control module and a cache module, and the control module is coupled to the storage device; wherein the step S14 further includes: at least transmitting the detection data stored in each storage unit to the cache module block by block; and transmitting the detection data stored in the cache module to the convolution module.

[0043] According to one aspect of the present disclosure, the step of delivering at least the detection data stored in each storage unit to the cache module block by block includes: delivering at least the detection data stored in each storage unit to the cache module block by block according to the boundary attributes of the first direction and the second direction of each storage unit;

[0044] When the boundary attribute of the storage unit in the first direction and / or the second direction is a non-convolution boundary, the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction are transmitted to the cache module through the control module; or

[0045] When the boundary attributes of the first direction and the second direction of the storage unit are both convolution boundaries, the detection data stored in the storage unit is transmitted to the cache module through the control module.

[0046] According to one aspect of the present disclosure, the detection areas with different properties include any one or more of the following: rectangular detection areas, discrete detection areas and strip detection areas.

[0047] In summary, the laser radar and detection method of the present invention have been introduced in detail. The laser radar of the present invention can reduce the complexity and amount of data processing, reduce computing power consumption, improve data processing speed, and reduce the size of the data processing circuit by storing and processing the detection data of pixels covered by the detection area in blocks. This is conducive to the miniaturization of the entire laser radar.

[0048] The laser radar disclosed in the present invention can perform block-by-block storage and block-by-block processing on the detection data of pixels covered by detection areas of different properties, and therefore has a wide range of applications.

[0049] The laser radar disclosed in the present invention uses a convolution module to perform block convolution processing on the detection data stored in the storage device, which can improve the data processing speed, improve the signal-to-noise ratio, and improve the laser radar's ranging capability and the accuracy of the detection results while reducing the amount of calculation and computing power consumption.

[0050] Furthermore, the laser radar disclosed in the present invention can further reduce the amount of calculation and power consumption while improving the data processing speed by performing convolution processing block by block in units of storage units.

[0051] The laser radar disclosed in the present invention determines the detection data transmitted to the convolution module according to the boundary attributes of the first direction and the second direction of each storage unit, overcomes or alleviates the adverse effect of the storage unit boundary on the convolution processing, further reduces the complexity of data processing, and further improves the data processing speed. In addition, based on the boundary attributes of the storage unit, the convolution processing operation can be operated within the range of the detection area, which is conducive to obtaining more accurate detection results. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the embodiments of the present disclosure, they are used to explain the present disclosure and do not constitute a limitation of the present disclosure. In the accompanying drawings:

[0053] Figure 1 A schematic diagram of a laser radar according to an embodiment of the present disclosure is shown;

[0054] Figure 2a-2c A schematic diagram showing detection areas of different properties according to the present disclosure;

[0055] Figure 3 A schematic diagram of a storage device according to a preferred embodiment of the present disclosure is shown;

[0056] Figure 4a-4c A schematic diagram showing the detection data of pixels covered by the detection area according to some preferred embodiments of the present disclosure is stored in a storage device in blocks;

[0057] Figure 5 A schematic diagram of a laser radar according to a preferred embodiment of the present disclosure is shown;

[0058] Figure 6 A schematic diagram showing a block-wise convolution process for detection data stored in a storage device according to a preferred embodiment of the present disclosure is shown;

[0059] Figure 7a-7d A schematic diagram showing determination of detection data to be transmitted to a convolution module according to some preferred embodiments of the present disclosure is shown;

[0060] Figure 8 A schematic diagram of a storage unit provided with additional storage locations according to a preferred embodiment of the present disclosure is shown;

[0061] Fig. 9 A schematic diagram of a laser radar according to another preferred embodiment of the present disclosure is shown;

[0062] Fig.10 A schematic diagram of a laser radar according to another preferred embodiment of the present disclosure is shown;

[0063] Fig.11 A schematic diagram showing a laser radar according to another preferred embodiment of the present disclosure; and

[0064] Fig.12 A flow chart of a detection method according to an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0065] In the following, only some exemplary embodiments are briefly described. As those skilled in the art will appreciate, the described embodiments may be modified in various ways without departing from the spirit or scope of the present disclosure. Therefore, the drawings and descriptions are considered to be exemplary and non-restrictive in nature.

[0066] In the description of the present disclosure, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, which are only for the convenience of describing the present disclosure and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present disclosure. In addition, the terms "first" and "second" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present disclosure, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0067] In the description of the present disclosure, it should be noted that, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, an electrical connection, or can communicate with each other; it can be directly connected, or indirectly connected through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present disclosure can be understood according to specific circumstances.

[0068] In the present disclosure, unless otherwise clearly specified and limited, a first feature being "above" or "below" a second feature may include that the first and second features are in direct contact, or may include that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, a first feature being "above", "above" and "above" a second feature includes that the first feature is directly above and obliquely above the second feature, or simply means that the first feature is higher in level than the second feature. A first feature being "below", "below" and "below" a second feature includes that the first feature is directly below and obliquely below the second feature, or simply means that the first feature is lower in level than the second feature.

[0069] The disclosure below provides many different embodiments or examples to realize different structures of the present disclosure. In order to simplify the disclosure of the present disclosure, the parts and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the present disclosure. In addition, the present disclosure can repeat reference numbers and / or reference letters in different examples, and this repetition is for the purpose of simplicity and clarity, and does not itself indicate the relationship between the various embodiments and / or settings discussed. In addition, the present disclosure provides various specific examples of processes and materials, but those of ordinary skill in the art can be aware of the application of other processes and / or the use of other materials.

[0070] The preferred embodiments of the present disclosure are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present disclosure, and are not used to limit the present disclosure.

[0071] The present invention provides a laser radar. Figure 1 A schematic diagram of a laser radar 100 according to an embodiment of the present disclosure is shown. Figure 1 As shown, the laser radar 100 includes a transmitting device 110, a detecting device 120, a storage device 130 and a data processing device 140. The transmitting device 110 is configured to transmit a detection beam L for detecting an obstacle OB, and the detecting device 120 includes a pixel array. The echo beam L' after the detection beam L is reflected on the obstacle OB can form a detection area on the pixel array (refer to Figure 2a-2c The pixel array includes a plurality of pixels 121, each pixel 121 can respond to the echo light beam L' and generate detection data. The storage device 130 includes a plurality of storage units 131 distributed in an array, wherein each storage unit 131 can store the detection data of a plurality of pixels 121, and the number of pixels corresponding to each storage unit 131 is less than the number of pixels covered by the detection area, so that the detection data of the pixels covered by the detection area are stored in blocks in the plurality of storage units 131 of the storage device 130. The data processing device 140 is coupled to the detection device 120 and the storage device 130, and is configured to perform block processing on the detection data stored in the storage device 130. It should be noted that the specific number and arrangement of the plurality of storage units and the plurality of pixels are not limited to Figure 1 The situation shown can be flexibly set according to needs in specific applications. The laser radar disclosed in the present invention can optimize the data processing circuit adapted to the large-size detection area by storing and processing the detection data of the pixels covered by the detection area in blocks, reduce the complexity and calculation amount of data processing, reduce the computing power consumption, improve the data processing speed, and reduce the size of the data processing circuit, which is conducive to the miniaturization of the entire laser radar. The laser radar disclosed in the present invention is introduced in detail below.

[0072] In a laser radar, the transmitting device includes one or more lasers, and the multiple lasers can be arranged as a one-dimensional laser array or a two-dimensional laser array. Preferably, the laser can be a vertical cavity surface emitting laser (Vertical-Cavity Surface-Emitting Laser, VCSEL), an edge emitting laser (Edge Emitting Laser, EEL), etc. In addition, the transmitting device also includes one or more control circuits corresponding to the number of lasers, and each control circuit can be electrically connected to the corresponding laser through a separate switching device to form a corresponding transmission channel, so that each laser can be independently controlled. In addition, the transmitting device also includes a driver chip, and the driver chip is preferably a multi-channel driver chip. The driver chip includes a control logic unit, and the control logic unit can be electrically connected to the control circuit, so that the driver chip can drive one or more lasers to emit light through the control circuit to form a detection beam.

[0073] During the detection process, in order to make the laser radar have different resolutions, long-distance measurement, scanning rates and other performances in different fields of view or scenes, so as to adapt to more flexible and personalized needs, the transmitting device can be configured to emit different types of detection beams, such as detection beams with different sizes of spots, detection beams with different numbers of spots, detection beams with different energies, detection beams with different distribution modes, etc., for detecting obstacles. For example, when the laser radar needs to measure nearby (or the required resolution is low), the energy of the detection beam required to be emitted is low, and the transmitting device can emit a detection beam with a larger spot, which can cover a larger field of view to improve the field of view scanning rate. On the contrary, when the laser radar needs to measure far away (or the required resolution is high), the energy of the detection beam required to be emitted is high. Considering the safety of human eyes, the transmitting device can emit a detection beam with a smaller spot to avoid excessive laser energy in the field of view irradiated by the detection beam. Preferably, the transmitting device can also emit multiple detection beams with smaller spots in time-sharing or simultaneously to improve the field of view scanning rate. In addition, in order to speed up the scanning rate, when the laser radar needs to measure a certain horizontal or vertical field of view, the transmitting device can also emit a detection beam of a strip light spot. By moving the scanning field of view (FOV) in the horizontal and / or vertical direction, scanning detection of a one-dimensional or two-dimensional field of view can be achieved. For example, the entire horizontal field of view of a certain vertical field of view angle is first irradiated, and then the entire vertical field of view is gradually scanned in the horizontal direction, thereby achieving scanning detection of a two-dimensional field of view. As a result, the laser radar can emit different types of detection beams to flexibly adapt to different scene scans, so that during the detection process, the corresponding detection beam can be switched or selected as needed to achieve better detection performance in the current scene or field of view.

[0074] During the detection process, the detection beam emitted by the transmitting device can be deflected by a scanning device (such as a rotating mirror, a galvanometer, etc.), or the transmitting device can be driven by the rotor to patrol and emit the detection beam at a certain horizontal angle (such as 0.05°, etc.), and the pixel array can respond to the echo beam reflected by the detection beam on the obstacle and convert it into an electrical signal. Through the movement of the field of view, the laser radar can complete the scanning of the entire scene of the surrounding environment, thereby realizing the detection within a certain horizontal field of view. The data processing device can obtain the point cloud of the surrounding environment by splicing the point cloud of each field of view area. It should be noted that the laser radar disclosed in the present invention is not limited to the mechanical rotating laser radar or scanning laser radar introduced above, and can also be a solid-state laser radar, that is, it does not require any mechanical rotating structure. Through the array setting of the transmitting device and the detection device, it can also cover a certain spatial field of view. The laser radar disclosed in the present invention can switch different types of detection beams and / or scanning methods according to different detection needs, so as to be applied to different field of view ranges or different scenes, and has a strong adaptive adjustment ability.

[0075] After different types of detection beams are reflected on obstacles, the echo beams can form detection areas with different properties on the pixel array (i.e., the area composed of pixels covered by the echo beams). The different properties can be, for example, different numbers of covered pixels, different distribution characteristics of covered pixels (such as different shapes and densities), different sizes of detection areas, etc. The laser radar disclosed in the present invention can perform block storage and block processing (described in detail later) for the detection data of pixels covered by detection areas of different properties, so it has a wide range of applications, fast data processing speed, low system power consumption, and is conducive to reducing the size of the data processing circuit and realizing the miniaturization of the entire laser radar.

[0076] Preferably, the detection areas with different properties include any one or more of the following: rectangular detection areas, discrete detection areas, and strip detection areas. The detection area can be understood as the area formed by the pixels covered by the detection device during each scanning process when the echo light beam generated by the detection light beam reflected by the obstacle is incident; or it can also be understood as the area formed by the activated pixels during each scanning process. Therefore, during the entire detection process, the detection area is constantly moving and changing to cover the entire detection field of view. The detection areas of different properties of the present disclosure are introduced below.

[0077] Reference Figure 2a-2c , Figure 2a The shaded portion in the middle shows a rectangular detection area of ​​a property of the present disclosure, taking the case where the number of pixels covered by the rectangular detection area is 84*44=3696 as an example. Figure 2bThe shaded part in the middle shows a discrete detection area of ​​a property of the present invention, which includes multiple (for example, 8) discretely distributed sub-detection areas. The multiple sub-detection areas constitute the discrete detection area, and the number of pixels covered by each sub-detection area is, for example, 24*24=576. The number of pixels covered by the discrete detection area is, for example, 8*24*24=4608. Figure 2c The shaded portion in the middle shows a strip-shaped detection area of ​​a nature of the present disclosure, taking the number of pixels covered by the strip-shaped detection area as 12*384=4608 as an example. Figure 2a and Figure 2b (or Figure 2c ), the number of pixels covered by the rectangular detection area is different from the number of pixels covered by the discrete detection area (or the strip detection area), and the distribution characteristics of the pixels covered by the detection area are also different. Figure 2a and Figure 2b (or Figure 2c ), the distribution characteristics of the pixels covered by the discrete detection area are different from the distribution characteristics of the pixels covered by the rectangular detection area (or the strip detection area). The pixels covered by the discrete detection area are distributed discretely, while the pixels covered by the rectangular detection area (or the strip detection area) are distributed concentratedly. In addition, Figure 2a-2c In the embodiment of the present invention, the field of view covered by each detection area is also different. It should be noted that: Figure 2a-2c The embodiments are only exemplary and do not constitute a limitation to the present disclosure. It is understandable that the detection area is not necessarily strictly the same as the shape of the light spot of the echo beam. For example, the light spot of the echo beam corresponding to the rectangular detection area may be elliptical.

[0078] In the existing scheme, the detection area that the data processing circuit can adapt to is limited. For example, the data processing circuit of some laser radars is only adapted to a 10*10 detection area. Therefore, for a larger detection area, such as an 84*44 detection area, since the number of pixels in the detection area increases significantly, if a traditional data processing circuit is used, it will face the problems of large size and high complexity of the required data processing circuit, high system power consumption, and being unfavorable for the miniaturization of the entire laser radar. The laser radar disclosed in the present invention can adapt to detection areas of different natures, including but not limited to the above-mentioned rectangular detection area, discrete detection area, and strip detection area. By storing and processing the detection data of the pixels covered by the detection area in blocks, the complexity, calculation amount, and calculation power consumption of data processing are greatly reduced, and the data processing speed is greatly improved. On the basis of reducing the size of the data processing circuit, it is conducive to the miniaturization of the laser radar.

[0079] Before introducing how the laser radar of the present disclosure performs block storage, the storage device of the present disclosure is first introduced. In the present disclosure, the number of pixels corresponding to (or that can be stored in) the storage device as a whole can be no less than the number of pixels covered by any detection area, so as to realize the storage of detection data of pixels covered by any detection area at the same time. Figure 2a-2c , the number of pixels covered by the rectangular detection area is 3696, and the number of pixels covered by the discrete detection area and the strip detection area is 4608. In this case, the number of pixels corresponding to the storage device is not less than 4608. For detection areas covering other numbers of pixels, it is only necessary to ensure that the number of pixels corresponding to the storage device is not less than the number of pixels covered by any detection area. In other words, the number of pixels corresponding to the storage device is not less than the number of pixels covered by the maximum detection area. Preferably, each pixel may include one or more single photon avalanche diodes (SPAD), each of which can be independently selected and addressed, and the data of each single photon avalanche diode can be independently output and read, and the data in one or more SPADs in each pixel can be accumulated to form the detection data of the pixel.

[0080] Figure 3 FIG. 1 is a schematic diagram of a storage device 130 according to a preferred embodiment of the present disclosure. Figure 3 As shown, the storage device 130 includes a plurality of storage units 131 distributed in an array, for example, 4 rows and 8 columns with a total of 32 storage units, each storage unit may include a plurality of storage locations to store detection data of a plurality of pixels, for example, each storage unit may include 144 storage locations to store detection data of 12*12=144 pixels (one storage unit may be regarded as including 12*12=144 storage pixels), so the entire storage array can store detection data of 8*4*144=4608 pixels, and can cover the pixels of the entire detection area. It should be noted that the present disclosure does not limit the specific number of storage units in the storage device 130, their arrangement, the number of pixels corresponding to them, and the number of pixels corresponding to each storage unit. The array distribution in the storage device may be a physical distribution of the storage units or a logical distribution, and the present disclosure does not limit this. Preferably, each storage unit has a unique address, for example Figure 3 Address-11, Address-12, ... Address-48 are shown. Preferably, each storage location of each storage unit may also have a unique address (not shown in the figure), which can be independently selected and addressed. Preferably, each storage unit also has a boundary attribute, and the boundary attribute includes a convolution boundary or a non-convolution boundary (described in detail later).

[0081] In the present disclosure, the number of pixels corresponding to each storage unit is less than the number of pixels covered by the detection area, and the detection data of the pixels covered by the entire detection area can be stored in blocks in multiple storage units of the storage device. Figure 2a-2c The detection data of the detection area shown can be obtained through Figure 3 The storage array shown is used to implement storage. Preferably, the data processing device 140 can store the detection data of the pixels covered by the detection area in blocks into corresponding multiple storage units according to the spatial correspondence between the pixel array and the storage unit array (in this disclosure, "spatial correspondence" can be understood as the physical spatial correspondence or logical spatial correspondence between the pixels in the pixel array and the storage positions in the storage unit array.). The following is a detailed introduction.

[0082] Figure 4a-4c FIG. 1 is a schematic diagram showing how the detection data of pixels covered by the detection area according to some preferred embodiments of the present disclosure are stored in a storage device in blocks, referring to FIG. Figure 4a ,for Figure 2a When storing data in the 84*44 rectangular detection area shown in FIG. 1 , the detection data of 3696 pixels covered by the rectangular detection area can be sequentially stored in the storage units of the 1st to 3rd rows and the 1st to 7th columns and a part of the storage units of the 4th row and the 1st to 7th columns in the 4*8 storage array according to the spatial correspondence between the pixel array and the storage pixels in the storage unit array. Figure 4a As shown in the shaded filled area. That is, it is equivalent to splitting the 84*44 rectangular detection area into blocks, for example, into 21 12*12 block detection areas and 7 12*8 block detection areas, and storing the detection data of the pixels covered by the 21 12*12 block detection areas and the 7 12*8 block detection areas in the corresponding storage units in sequence, and each storage unit corresponds to one 12*12 block detection area.

[0083] Reference Figure 4b ,for Figure 2b When storing data in the discrete detection area shown in FIG. 1 , the detection data of 4608 pixels covered by the 8 24*24 sub-detection areas in the discrete detection area can be sequentially stored in the storage cells of the 1st to 4th rows and the 1st to 8th columns in the 4*8 storage array according to the spatial correspondence between the pixel array and the storage unit array. Figure 4bAs shown in the shaded filled area, every four storage units adjacent to each other in the upper, lower, left, and right directions correspond to a 24*24 sub-detection area, and each storage unit corresponds to a 12*12 block detection area. That is, it is equivalent to splitting each 24*24 sub-detection area in the discrete detection area into four 12*12 block detection areas, and storing the detection data of the pixels covered by each 12*12 block detection area in the corresponding storage unit in sequence.

[0084] Reference Figure 4c ,for Figure 2c When storing data in the 12*384 strip detection area shown in FIG. 1 , the detection data of 4608 pixels covered by the strip detection area can be stored in a certain order (e.g., a continuous serpentine order, see FIG. 1 ) according to the spatial correspondence between the pixel array and the storage unit array. Figure 4c The direction indicated by the arrow in the middle) is stored in a 4*8 storage array in sequence, such as Figure 4c As shown in the shaded filled area. That is, it is equivalent to splitting the 12*384 strip detection area into 32 12*12 block detection areas, and storing the detection data of the pixels covered by the 32 12*12 block detection areas in the corresponding storage units in sequence, and each storage unit corresponds to one 12*12 block detection area. Since the 12*384 detection area is a strip detection area, in order to correspond to the spatial relationship, the detection data of the 32 12*12 block detection areas can be stored in a continuous serpentine order. For details, refer to Figure 4c , it can be stored from the 4 storage units in the first column on the leftmost side from top to bottom. After the storage of the first column is completed, according to the spatial correspondence and the continuity of the data, the 4 storage units in the second column are stored from bottom to top. After the storage of the second column is completed, the 4 storage units in the third column are stored from top to bottom, and so on, until the 32 storage units in the 8 columns store the detection data of the 32 12*12 block detection areas. It should be understood that the storage in a serpentine order can ensure the continuity of the detection area storage, and can realize the interaction between the data stored in the storage units, which is convenient for subsequent processing.

[0085] Figure 4a-4cThe embodiment of the present invention introduces the specific process of how the laser radar of the present invention performs block storage, that is, according to the spatial correspondence between the pixel array and the storage unit array, the detection data of the pixels covered by the detection area are stored in blocks in the corresponding multiple storage units. In the present invention, the detection area can be divided into multiple block detection areas, the number of pixels included in each block detection area can be the same or different, each block detection area corresponds to a storage unit, and the detection data of the pixels included in each block detection area can be stored in the corresponding storage unit. Therefore, even if the detection area is of different nature, it can be divided into multiple block detection areas, and then the detection data of the pixels included in the multiple block detection areas are stored in the corresponding storage unit, so that it can correspond to the detection areas of different nature, without allocating a unique storage location for each detection area of ​​the nature, and the detection data of the pixels covered by the detection areas of different nature can be stored using the same storage device, as long as the number of pixels corresponding to (or storable) the storage device as a whole is not less than the number of pixels covered by any detection area, so it can be compatible with detection areas of different nature and has a wide range of applications, thereby reducing the demand and consumption of storage resources, reducing the complexity, amount of calculation and power consumption of data processing, and improving the data processing speed, which is conducive to the miniaturization and low cost of the entire laser radar. In addition, in the block storage process, the detection data of the pixels covered by the multiple block detection areas can be read and stored in time-sharing mode to reduce the instantaneous power consumption of the laser radar; in addition, the detection data of the pixels covered by the multiple block detection areas can also reuse at least a part of the reading circuit to reduce the complexity of the data processing circuit. In the present disclosure, the reading circuit can be arranged in the data processing device (not shown in the figure).

[0086] The laser radar disclosed in the present invention can also perform block processing on the detection data stored in the storage device, which is described in detail below.

[0087] Figure 5 A schematic diagram of a laser radar 200 according to a preferred embodiment of the present disclosure is shown. Compared with the laser radar 100, the data processing device 140 of the laser radar 200 may include one or more convolution modules 141 (one convolution module 141 is shown in the figure as an example). The convolution module 141 is configured to perform block convolution processing on the detection data stored in the storage device 130, specifically including performing convolution processing block by block in units of storage units to obtain a reconstructed pixel array, and generating a laser radar point cloud based on the reconstructed pixel array. It should be noted that in this embodiment, referring to Figure 5, the convolution module 141 can be arranged in the data processing device 140 and coupled with the storage device 130, but the present disclosure is not limited thereto. The convolution module 141 may also not be arranged in the data processing device 140 but be separately coupled with the storage device (not shown in the figure), which are all within the protection scope of the present invention. In addition, it should be noted that the convolution processing refers to the convolution kernel moving step by step in a preset direction in units of convolution step length. Each time it moves, the detection data of all pixels covered by the convolution kernel (such as echo waveform, number of SPAD triggers, etc.) are accumulated once to obtain a recombined pixel. The detection data of the entire pixel array is traversed to obtain a recombined pixel array. Since the recombined pixel obtained by the convolution processing accumulates the detection data of multiple pixels, the signal strength of the recombined pixel is stronger, which can improve the signal-to-noise ratio of the detection data. Preferably, the spatial resolution of the laser radar can also be flexibly adjusted by adjusting the size of the convolution kernel and the convolution step length.

[0088] Before performing convolution processing, the data processing device 140 needs to read the detection data stored in the storage device 130 and transmit it to the convolution module 141. Since the detection data of the pixels covered by the detection area are stored in blocks during the storage process, in order to implement convolution processing on the detection data of the pixels covered by the detection area, certain rules are required to read the detection data in the storage device. Preferably, the data processing device 140 can send control instructions (such as the address, boundary attributes, read and write instruction types of the storage unit) to the storage unit array, and the storage unit array transmits the corresponding detection data to the convolution module 141 according to the control instructions, so that the convolution module 141 can perform convolution processing. Specifically, refer to Figure 3For example, when it is necessary to read the detection data stored in the storage unit with address Address-11 and transmit it to the convolution module 141, the data processing device 140 can send a control instruction such as "read-Address-11-trans-to-141" to the storage unit array, and the storage unit array receives the control instruction and transmits the detection data stored in the storage unit with address Address-11 to the convolution processing module 141; for another example, when it is necessary to read the detection data stored in the storage unit with address Address-12 and transmit it to the convolution module 141, the data processing device 140 can send a control instruction such as "read-Address-12-trans-to-141" to the storage unit array, and the storage unit array receives the control instruction and transmits the detection data stored in the storage unit with address Address-12 to the convolution processing module 141; when it is necessary to read and transmit the detection data in other storage units, the situation is similar and will not be repeated here. Preferably, each storage location in each storage unit can be independently selected and addressed, so the processing device 140 can also read the detection data stored in each storage location in each storage unit to transmit the corresponding detection data to the convolution module 141.

[0089] Since the detection data of the pixels covered by the detection area are stored in blocks during the storage process, in order to realize the convolution processing of the detection data of the pixels covered by the detection area, based on the continuity of the detection data in the detection area and the certain size of the convolution kernel, before the convolution processing, the data processing device 140 may not only need to read and transmit the detection data stored in a certain storage unit itself, but may also need to read and transmit part of the detection data stored in other storage units adjacent to the storage unit to the convolution module, which is related to the boundary properties of the storage unit. The boundary properties of the convolution unit are related to the boundary of the detection area, the direction of the convolution processing, the size of the convolution kernel, and the position of the convolution kernel (which will be introduced in detail later).

[0090] In the present disclosure, each storage unit has a boundary attribute, and the boundary attribute includes a convolution boundary or a non-convolution boundary. Preferably, each storage unit has boundary attributes in multiple directions, such as a boundary attribute in a first direction and a boundary attribute in a second direction. The data processing device 140 can determine the boundary attributes in the first direction and the second direction of each storage unit, and determine the detection data transmitted to the convolution module 141 according to the boundary attributes in the first direction and the second direction of each storage unit. Preferably, the first direction and the second direction can be determined by the convolution processing direction (the moving direction of the convolution kernel). Assuming that the convolution processing is performed from left to right and from top to bottom, the right side of the storage unit can be specified as the first direction, and the bottom side of the storage unit can be specified as the second direction. For details, please refer to Figure 6The first direction N1 and the second direction N2 are shown by the dotted arrows. It should be understood that the right side and the bottom side of the storage unit can also be specified as the second direction and the first direction respectively. If the convolution process is performed in other directions, such as from right to left or from bottom to top, the left side of the storage unit can be specified as the first direction (or the second direction), and the top side of the storage unit can be specified as the second direction (or the first direction), depending on the actual situation.

[0091] In some preferred embodiments of the present disclosure, when the boundary attribute of the storage unit in the first direction and / or the second direction is a non-convolution boundary, the data processing device 140 is configured to transmit the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction to the convolution module 141; when the boundary attribute of the storage unit in the first direction and the second direction is both a convolution boundary, the data processing device 140 is configured to transmit the detection data stored in the storage unit to the convolution module 141. Figure 6 , Figure 7a-7d , taking the detection data of pixels covered by the 24*24 detection area stored in the storage device 130 as an example (corresponding to Figure 2b and Figure 4b ), the specific process of the data processing device 140 determining the detection data to be transmitted to the convolution module 141 according to the boundary attributes of each storage unit in the first direction and the second direction is introduced. It should be noted that the specific process of transmitting the detection data of pixels covered by other detection areas stored in the storage device 130 to the convolution module 141 is similar.

[0092] Reference Figure 6 , the detection data of 24*24 pixels in the detection area can be stored in blocks in 4 storage units in the storage device 130. For the convenience of explanation, the numbers 1, 2, 3, and 4 are used as examples, where each storage unit corresponds to 12*12 pixels. Each small square in the figure represents 1 pixel. The 5*5 dark shaded part represents the convolution kernel. The figure shows a plurality of different positions (not all) to which the convolution kernel can move. The following is an example of the convolution processing direction from left to right and from top to bottom, the convolution kernel conv=5*5 (dark shaded part), and the convolution step stride=1.

[0093] First, convolution processing starts from storage unit 1. Figure 6 and Figure 7a , storage unit 1 is adjacent to storage unit 2 along the first direction N1, and adjacent to storage unit 3 along the second direction N2. When the convolution kernel moves to the common boundary of storage unit 1 and storage unit 2 (refer to Figure 7aWhen the convolution kernel is at position 1, i.e., P1), since the convolution kernel has a certain size, the detection data of the 5*5 pixels covered by the convolution kernel includes not only the detection data of a part of the pixels of the storage unit 1, but also the detection data of a part of the pixels of the storage unit 2. Based on the continuity of the detection data, the detection data of the 5*5 pixels covered by the convolution kernel need to be transmitted to the convolution module 141. Therefore, the boundary attribute of the storage unit 1 along the first direction N1 is a non-convolution boundary (refer to Figure 7a NIB0 in FIG); when the convolution kernel moves to the common boundary of the storage unit 1 and the storage unit 3 (refer to position 2 in FIG, that is, P2), the detection data of the pixels covered by the convolution kernel includes not only the detection data of a part of the pixels of the storage unit 1, but also the detection data of a part of the pixels of the storage unit 3. Since the convolution kernel has a certain size, based on the continuity of the detection data, the detection data of the 5*5 pixels covered by the convolution kernel need to be transmitted to the convolution module 141. Therefore, the boundary attribute of the storage unit 1 along the second direction N2 is a non-convolution boundary (refer to FIG. Figure 7a N2B0 in the figure); when the convolution kernel moves to the common boundary of storage unit 1, storage unit 2, storage unit 3 and storage unit 4 (refer to position 3 in the figure, i.e., P3), since the convolution kernel has a certain size, the detection data of the 5*5 pixels covered by the convolution kernel includes not only the detection data of a part of the pixels of storage unit 1, but also the detection data of a part of the pixels of storage unit 2, storage unit 3 and storage unit 4. Based on the continuity of the detection data, the detection data of the 5*5 pixels covered by the convolution kernel need to be transmitted to the convolution module 141. Therefore, the boundary attributes of the first direction N1 and the second direction N2 of the storage unit 1 are both non-convolution boundaries (refer to Figure 7a In order to implement convolution processing on the detection data of the pixels covered by the detection area (for example, the detection area in the example includes 24*24 pixels, and its detection data is divided into blocks and stored in storage units 1, 2, 3, and 4), since the convolution kernel has a certain size, based on the continuity of the detection data in the detection area, before performing the convolution processing, the data processing device 140 needs to transmit the detection data stored in the storage unit 1, the partial detection data stored in the storage unit 2 adjacent to the storage unit 1 along the first direction N1, the partial detection data stored in the storage unit 3 adjacent to the storage unit 1 along the second direction N2, and the partial detection data of the storage unit 4 to the convolution module 141. More specifically, refer to Figure 7a, that is, the detection data of 12*12 pixels stored in storage unit 1, the detection data of 4 columns of pixels stored in storage unit 2 adjacent to the right of storage unit 1, the detection data of 4 rows of pixels stored in storage unit 3 adjacent to the bottom of storage unit 1, and the detection data of 4*4 pixels stored in storage unit 4 are transmitted to the convolution module 141. That is, it is equivalent to transmitting the detection data of 16*16 pixels to the convolution module 141. It should be noted that the number of the partial detection data is related to the size of the convolution kernel conv and the convolution step length stride. For the case where the convolution kernel conv=5*5 and the convolution step length stride=1, as Figure 7a Although not shown in the figure, for other convolution kernels and convolution strides, such as conv=3*3 and convolution stride=2, similarly, the detection data of 12*12 pixels stored in storage unit 1, the detection data of 2 columns of pixels stored in the storage unit 2 adjacent to its right, the detection data of 2 rows of pixels stored in the storage unit 3 adjacent to its bottom, and the detection data of 2*2 pixels stored in storage unit 4 are transmitted to the convolution module 141 together.

[0094] Reference Figure 6 and Figure 7b After the convolution processing of storage unit 1 is completed, the convolution processing is performed on storage unit 2. In the same detection area, storage unit 2 is not adjacent to other storage units corresponding to the detection area along the first direction N1, and is adjacent to storage unit 4 along the second direction N2. When the convolution kernel moves to the boundary of storage unit 2 (refer to position 4, i.e., P4 in the figure), the detection data of the 5*5 pixels covered by the convolution kernel only includes the detection data of the pixels of storage unit 2, and does not involve the detection data of the pixels of other storage units. Therefore, the boundary attribute of storage unit 2 along the first direction N1 is the convolution boundary (refer to Figure 7b N1B1 in the figure); when the convolution kernel moves to the common boundary of the storage unit 2 and the storage unit 4 (refer to the position 5 in the figure, i.e., P5), the detection data of the 5*5 pixels covered by the convolution kernel includes not only the detection data of a part of the pixels of the storage unit 2, but also the detection data of a part of the pixels of the storage unit 4. Based on the continuity of the detection data, the detection data of the 5*5 pixels covered by the convolution kernel need to be transmitted to the convolution module 141. Therefore, the boundary attribute of the storage unit 4 along the second direction N2 is a non-convolution boundary (refer to Figure 7bN2B0 in the first direction N1). Therefore, the boundary attribute of the storage unit 2 in the first direction N1 is a convolution boundary, and the boundary attribute of the second direction N2 is a non-convolution boundary. Therefore, in order to implement convolution processing on the detection data of the pixels (24*24 pixels) covered by the detection area, since the convolution kernel has a certain size, based on the continuity of the detection data in the detection area, before performing the convolution processing, the data processing device 140 needs to transmit the detection data stored in the storage unit 2 and the partial detection data stored in the storage unit 4 adjacent to the storage unit 2 along the second direction N2 to the convolution module 141. More specifically, refer to Figure 7b For the case where the convolution kernel conv=5*5 and the convolution step length stride=1, the detection data of the 12*12 pixels stored in the storage unit 2 and the detection data of the 4 rows of pixels stored in the storage unit 4 adjacent to the lower side need to be transmitted to the convolution module 141. That is, it is equivalent to transmitting the detection data of the 12*16 pixels to the convolution module 141. For other convolution kernels and convolution step lengths, it is similar and will not be repeated here.

[0095] Reference Figure 6 and Figure 7c After the convolution processing of storage unit 2 is completed, the convolution processing is performed on storage unit 3. In the same detection area, storage unit 3 is adjacent to storage unit 4 along the first direction N1, and is not adjacent to other storage units along the second direction N2. When the convolution kernel moves to the common boundary of storage unit 3 and storage unit 4 (refer to position 6 in the figure, i.e., P6), the detection data of 5*5 pixels covered by the convolution kernel includes not only the detection data of a part of the pixels of storage unit 3, but also the detection data of a part of the pixels of storage unit 4. Since the convolution kernel has a certain size, based on the continuity of the detection data, the detection data of 5*5 pixels covered by the convolution kernel needs to be transmitted to the convolution module 141. Therefore, the boundary attribute of storage unit 3 along the first direction N1 is a non-convolution boundary (refer to Figure 7c When the convolution kernel moves to the boundary of the storage unit 3 (refer to position 7, i.e., P7 in the figure), the detection data of the 5*5 pixels covered by the convolution kernel only includes the detection data of the pixels of the storage unit 3, and does not involve the detection data of the pixels of other storage units. Therefore, the boundary attribute of the storage unit 3 along the second direction N2 is the convolution boundary (refer to Figure 7cN2B1 in the first direction N1). Therefore, the boundary attribute of the storage unit 3 in the first direction N1 is a non-convolution boundary, and the boundary attribute of the storage unit 3 in the second direction N2 is a convolution boundary. Therefore, in order to implement convolution processing on the detection data of the pixels (24*24 pixels) covered by the detection area, since the convolution kernel has a certain size, based on the continuity of the detection data in the detection area, the data processing device 140 needs to transmit the detection data stored in the storage unit 3 and the partial detection data stored in the storage unit 4 adjacent to the storage unit 3 along the first direction N1 to the convolution module 141. More specifically, refer to Figure 7c For the case where the convolution kernel conv=5*5 and the convolution step length stride=1, the detection data of the 12*12 pixels stored in the storage unit 3 and the detection data of the 4 columns of pixels stored in the storage unit 4 adjacent to the right need to be transmitted to the convolution module 141 together. That is, it is equivalent to transmitting the detection data of the 16*12 pixels to the convolution module 141. For other convolution kernels and convolution step lengths, it is similar and will not be repeated here.

[0096] Reference Figure 6 and Figure 7d After the convolution processing of storage unit 3 is completed, the convolution processing is performed on storage unit 4. In the same detection area, storage unit 4 is not adjacent to other storage units 4 along the first direction N1, and is not adjacent to other storage units along the second direction N2. When the convolution kernel moves to the boundary of storage unit 4 (refer to position 8 and position 9, i.e. P8 and P9 in the figure), the convolution kernel moves to the boundary of the detection area in both the first direction and the second direction, and then the detection data of the 5*5 pixels covered by the convolution kernel only includes the detection data of the pixels of storage unit 4, and does not involve the detection data of the pixels of other storage units. Therefore, the boundary attributes of the first direction N1 and the second direction N2 of storage unit 4 are both convolution boundaries (refer to Figure 7d Therefore, in order to implement convolution processing on the detection data of the pixels covered by the detection area (24*24 pixels in the example), since the convolution kernel has a certain size, based on the continuity of the detection data in the detection area, the data processing device 140 only needs to transmit the detection data stored in the storage unit 4 to the convolution module 141. More specifically, the detection data of 12*12 pixels stored in the storage unit 4 is transmitted to the convolution module 141.

[0097] It should be noted that although the convolution processing is explained above in the order of storage units 1-4, the present disclosure does not limit the order in which the data processing of each storage unit is performed, and it can be performed simultaneously or in a time-sharing manner.

[0098] Figure 6-7dThe embodiment introduces a specific process of transmitting the detection data stored in the storage device 130 to the convolution module 141, that is, according to the boundary attributes of the first direction and the second direction of the storage unit, the detection data transmitted to the convolution module 141 is determined. When the boundary attribute of the first direction and / or the second direction of the storage unit is a non-convolution boundary, the data processing device 140 is configured to transmit the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction to the convolution module 141; when the boundary attributes of the first direction and the second direction of the storage unit are both convolution boundaries, the data processing device 140 is configured to transmit the detection data stored in the storage unit to the convolution module 141, thereby the corresponding data can be transmitted to the convolution module 141 through simple judgment, so that the convolution processing operation can be operated within the range of the detection area, which is conducive to obtaining more accurate detection results, that is, after the detection area is stored in the storage unit in blocks, the block convolution data processing is performed on each storage unit, which further reduces the complexity of data processing and further improves the data processing speed.

[0099] Reference Figure 3-7d In some preferred embodiments of the present disclosure, each storage unit may only have a standard storage location for storing the detection data of the storage unit. In other preferred embodiments of the present disclosure, in addition to the standard storage location, the storage unit may also have an additional storage location for storing part of the detection data in other storage units adjacent to the storage unit. Figure 8 , Figure 8 A schematic diagram of a storage unit 132 provided with additional storage positions according to a preferred embodiment of the present disclosure is shown. In this embodiment, the storage unit 132 is not only provided with standard storage positions pos1 (e.g., 144) for storing the detection data of 12*12=144 pixels in the storage unit, but also may be provided with additional storage positions pos2 (e.g., 112) for storing part of the detection data in other storage units adjacent to the storage unit. More specifically, refer to Figure 6 and Figure 7a , Figure 8The standard storage position pos1 in the storage unit 132 can be used to store the detection data of 12*12=144 pixels of the storage unit 1, and the additional storage position pos2 is used to store the detection data of 4 columns of pixels in the storage unit 2 adjacent to the right side of the storage unit 1, the detection data of 4 rows of pixels in the storage unit 3 adjacent to the bottom side of the storage unit 1, and the detection data of 4*4 pixels in the storage unit 4. It should be understood that this embodiment is only used for exemplary description and does not constitute a limitation of the present disclosure. In practical applications, the specific number and distribution of the standard storage position pos1 and the additional storage position pos2 can be flexibly adjusted according to actual conditions (such as the size of the convolution kernel, the convolution step size and other parameters), which are all within the protection scope of the present disclosure.

[0100] When the detection data is transmitted to the convolution module 141, the convolution module 141 can perform convolution processing on the detection data transmitted thereto block by block in units of storage units. It should be noted that in the present disclosure, "convolution processing block by block" should be understood in a broad sense, which can be understood as time-sharing convolution processing (i.e. serial processing) of the detection data stored in each storage unit in units of storage units, or can be understood as convolution processing of the detection data stored in multiple storage units in parallel in units of storage units. In order to reduce the peak power consumption and the complexity of the circuit, when the time margin is sufficient, a time-sharing convolution processing method can be adopted, that is, only one convolution module needs to be configured to realize time-sharing processing of the data of multiple storage units. In order to increase the data processing speed, a parallel processing method can be adopted. For example, a convolution module can be set separately for each storage unit, so that the data of multiple storage units can be processed in parallel to realize fast data processing. In order to balance the computing speed and power consumption, a combination of serial and parallel processing can also be used. For example, the storage units can be grouped, each group of storage units corresponds to a convolution module, and the storage units in each group are time-sharing convolution processed by the corresponding convolution module, and different groups can perform convolution processing in parallel. Furthermore, the storage units can be grouped according to the size of the detection area and the scanning order of the detection area. For example, multiple storage units corresponding to the same detection area can correspond to different groups respectively, and multiple storage units corresponding to different detection areas can be divided into the same group, so that multiple storage units corresponding to the same detection area can perform convolution processing in parallel, thereby quickly processing the convolution operation of the entire data in the detection area, while taking into account reducing the complexity of the overall circuit.

[0101] Reference Figure 6-7dTaking the block convolution processing of the 24*24 detection data stored in the storage device as an example, the convolution module can perform convolution processing on the detection data of 16*16, 12*16, 16*12 and 12*12 pixels in the order of storage units 1, 2, 3, 4, traverse the 24*24 pixel detection data stored in the entire storage array, obtain a reorganized pixel array, and generate a lidar point cloud based on the reorganized pixel array. Alternatively, the convolution module can perform convolution on storage units 1, 2, 3, 4 in parallel, that is, in parallel Figures 7a-7d The detection data of the 16*16, 12*16, 16*12 and 12*12 pixels shown are convolved, and the detection data of the 24*24 pixels stored in the entire storage array are traversed to obtain a reorganized pixel array, and a lidar point cloud is generated based on the reorganized pixel array.

[0102] In addition, in addition to performing convolution processing on all storage units in a serial or parallel manner, the convolution module may also perform convolution processing only on the detection data stored in one or more storage units among the storage units 1 , 2 , 3 and 4 .

[0103] It should be understood that the block convolution processing process for other detection data such as 84*44, 12*384 stored in the storage device is similar, and will not be repeated here. The laser radar disclosed in the present invention can improve the data processing speed, improve the signal-to-noise ratio, improve the laser radar's distance measurement capability and the accuracy of the detection results while reducing the amount of calculation and computing power consumption by performing convolution processing block by block based on the storage unit, and at the same time can reduce the size of the data processing circuit, which is conducive to the miniaturization of the entire laser radar.

[0104] Fig. 9 A schematic diagram of a laser radar 300 according to another preferred embodiment of the present disclosure is shown. Fig. 9As shown, in this embodiment, the data processing device 140 may include a convolution module 141, a control module 143 and a cache module 144. The control module 143 is coupled to the storage device 130 and the cache module 144. The cache module 144 is coupled to the convolution module 141. The control module 143 is configured to at least transmit the detection data stored in each storage unit to the cache module 144 block by block, and transmit the detection data stored in the cache module 144 to the convolution module 141. Specifically, the control module 143 may send a control instruction (such as the address, boundary attribute, read and write instruction type of the storage unit) to the storage unit array. The storage unit array transmits the corresponding detection data to the cache module 144 according to the control instruction, and transmits the detection data stored in the cache module 144 to the convolution module 141 for the convolution module 141 to perform convolution processing. Preferably, the control module 143 may at least transmit the detection data stored in each storage unit to the cache module 144 block by block according to the boundary attributes of the first direction and the second direction of each storage unit. Specifically, when the boundary attributes of the storage unit in the first direction and / or the second direction are non-convolution boundaries, in order to realize convolution processing on the detection data of the pixels (24*24 pixels) covered by the detection area, since the convolution kernel has a certain size, based on the continuity of the detection data in the detection area, the control module 143 transmits the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction to the cache module 144; or when the boundary attributes of the storage unit in the first direction and the second direction are both convolution boundaries, the control module 143 transmits the detection data stored in the storage unit to the cache module 144.

[0105] More specifically, taking the detection data of pixels covered by the 24*24 detection area stored in the storage device 130 as an example, refer to Figure 6 and Figure 7aIn the same detection area, the boundary attributes of the first direction N1 and the second direction N2 of the storage unit 1 are both non-convolution boundaries. Since the convolution kernel has a certain size, based on the continuity of the detection data in the detection area, the control module 143 transmits the detection data stored in the storage unit 1, part of the detection data stored in the storage unit 2 adjacent to the storage unit 1 along the first direction N1, part of the detection data stored in the storage unit 3 adjacent to the storage unit 1 along the second direction N2, and part of the detection data of the storage unit 4 to the cache module 144, and transmits the detection data stored in the cache module 144 to the convolution module 141. More specifically, the detection data of 12*12 pixels stored in the storage unit 1, the detection data of 4 columns of pixels stored in the storage unit 2 adjacent to the right side of the storage unit 1, the detection data of 4 rows of pixels stored in the storage unit 3 adjacent to the bottom side of the storage unit 1, and the detection data of 4*4 pixels stored in the storage unit 4 are transmitted to the cache module 144, and the detection data stored in the cache module 144 is transmitted to the convolution module 141.

[0106] Reference Figure 6 and Figure 7b In the same detection area, the boundary attribute of the first direction N1 of the storage unit 2 is a convolution boundary, and the boundary attribute of the second direction N2 is a non-convolution boundary. Since the convolution kernel has a certain size, based on the continuity of the detection data in the detection area, the control module 143 transmits the detection data stored in the storage unit 2 and part of the detection data stored in the storage unit 4 adjacent to the storage unit 2 along the second direction N2 to the cache module 144, and transmits the detection data stored in the cache module 144 to the convolution module 141. More specifically, the detection data of 12*12 pixels stored in the storage unit 2 and the detection data of 4 rows of pixels stored in the storage unit 4 adjacent to the lower side are transmitted to the cache module 144 together, and the detection data stored in the cache module 144 is transmitted to the convolution module 141.

[0107] Reference Figure 6 and Figure 7c In the same detection area, the boundary attribute of the first direction N1 of the storage unit 3 is a non-convolution boundary, and the boundary attribute of the second direction N2 is a convolution boundary. Based on the continuity of the detection data in the detection area, the control module 143 transmits the detection data stored in the storage unit 3 and part of the detection data stored in the storage unit 4 adjacent to the storage unit 3 along the first direction N1 to the cache module 144, and transmits the detection data stored in the cache module 144 to the convolution module 141. More specifically, the detection data of 12*12 pixels stored in the storage unit 3 and the detection data of 4 columns of pixels stored in the storage unit 4 adjacent to the right are transmitted to the cache module 144 together, and the detection data stored in the cache module 144 is transmitted to the convolution module 141.

[0108] Reference Figure 6 and Figure 7d In the same detection area, the boundary attributes of the first direction N1 and the second direction N2 of the storage unit 4 are both convolution boundaries. Since the convolution kernel has a certain size, based on the continuity of the detection data in the detection area, the control module 143 only transmits the detection data stored in the storage unit 4 to the cache module 144, and transmits the detection data stored in the cache module 144 to the convolution module 141. More specifically, the detection data of 12*12 pixels stored in the storage unit 4 is transmitted to the cache module 144, and the detection data stored in the cache module 144 is transmitted to the convolution module 141.

[0109] It should be understood that the process of transmitting the detection data of pixels covered by other detection areas stored in the storage device 130 to the cache module 144 and the convolution module 141 is similar and will not be described in detail here. In addition, it should be noted that the above embodiment is introduced by taking the convolution kernel of conv=5*5 and the convolution step length of stride=1 as an example. For other convolution kernels and convolution step lengths, they are similar and will not be described in detail here. In addition, Fig.10 In the laser radar, the convolution module 141, the control module 143, and the cache module 144 are all arranged in the data processing device 140, but the present invention is not limited to this. The convolution module 141 and / or the control module 143 and / or the cache module 144 can also be arranged in the detection device 120, or in the storage device 130, or between the detection device 120 and the storage device 130, and can even be arranged as a separate module at any configurable position in the laser radar, all of which are within the protection scope of the present invention.

[0110] Fig.10 A schematic diagram of a laser radar 400 according to another preferred embodiment of the present disclosure is shown. In this embodiment, in addition to the convolution module 141, the data processing device 140 also includes an accumulator 142 (ACC), and the accumulator 142 is coupled to the detection device 120 and the storage device 130 respectively. In some preferred embodiments of the present disclosure, during a detection process of the laser radar, the transmitting device can transmit the detection beam multiple times to perform multiple scanning detections (for example, 200-800 times, or even more than 1000 times, etc.), and the pixel array of the detection device can respond to multiple echo beams. The accumulator 142 is configured to accumulate the detection data generated by the multiple echo beams on each pixel pixel by pixel, and store them in the storage unit corresponding to the pixel. In the subsequent convolution processing process, the accumulated detection data stored in the storage device can be subjected to block convolution processing, thereby further improving the signal-to-noise ratio and the distance measurement capability of the laser, thereby further improving the accuracy of the laser radar detection result. It should be noted that Fig.10In the laser radar, the convolution module 141 and the accumulator 142 are both arranged in the data processing device 140, but the present invention is not limited thereto. The convolution module 141 and / or the accumulator 142 may also be arranged in the detection device 120, or in the storage device 130, or may even be arranged separately between the detection device 120 and the storage device 130, all of which are within the protection scope of the present invention.

[0111] Fig.11 A schematic diagram of a laser radar 500 according to another preferred embodiment of the present disclosure is shown. In this embodiment, the laser radar 500 includes a convolution module 141, an accumulator 142, a control module 143 and a cache module 144. In this embodiment, the control module 143 is coupled to the storage device 130 and the cache module 144. The cache module 144 is coupled to the convolution module 141. The accumulator is coupled to the detection device 120 and the storage device 130. It should be noted that Fig.11 In the laser radar, the convolution module 141, the accumulator 142, the control module 143 and the cache module 144 are all arranged in the data processing device 140, but the present invention is not limited thereto. One or more of the convolution module 141, the accumulator 142, the control module 143 and the cache module 144 may also be arranged in the detection device 120, or in the storage device 130, or may even be arranged separately between the detection device 120 and the storage device 130, all of which are within the protection scope of the present invention. In addition, the connection relationship between the modules is not limited to the situation shown in the figure. In practical applications, it can be flexibly adjusted according to actual conditions. Although not shown in the figure, in a specific embodiment of the present disclosure, the control module 143 may be coupled with the storage device 130, the cache module 144 may be coupled with the storage device 130 and the convolution module 141, and the accumulator 142 may be arranged between the detection device 120 and the storage device 130.

[0112] The above is a detailed introduction to the laser radar of the present disclosure. The present disclosure also relates to a detection method 10 of the laser radar, wherein the laser radar includes a transmitting device 110, a detection device 120, a storage device 130 and a data processing device 140, wherein the detection device 120 includes a pixel array, the transmitting device 110 can emit a detection light beam L, and an echo light beam L' after the detection light beam L is reflected on an obstacle OB can form a detection area on the pixel array, and the pixel array includes a plurality of pixels 121, each pixel 121 can respond to the echo light beam L' and generate detection data; the data processing device 140 is coupled to the detection device 120 and the storage device 130.

[0113] Fig.12 A flow chart of a detection method 10 according to an embodiment of the present disclosure is shown. Fig.12 As shown, the detection method 10 includes steps S11-S14. In step S11, a detection beam L is emitted to detect an obstacle OB. In step S12, a response is made to the echo beam L' and detection data is generated. In step S13, the detection data of the pixels covered by the detection area are stored in blocks in multiple storage units 131 of the storage device 130, wherein the storage device 130 includes multiple storage units 131 distributed in an array, each storage unit 131 can store the detection data of multiple pixels 121, and the number of pixels corresponding to each storage unit 131 is less than the number of pixels covered by the detection area, so that the detection data of the pixels 121 covered by the detection area are stored in blocks in the multiple storage units 131 of the storage device 130. In step S14, the detection data stored in the storage device 130 is processed in blocks.

[0114] According to a preferred embodiment of the present disclosure, step S11 includes: emitting different types of detection light beams L for detecting obstacles OB, wherein the echo light beams L' of the different types of detection light beams L after being reflected on the obstacle OB can form detection areas with different properties on the pixel array.

[0115] According to a preferred embodiment of the present disclosure, the number of pixels corresponding to the storage device 130 is not less than the number of pixels covered by any of the detection areas.

[0116] According to a preferred embodiment of the present disclosure, the step S13 includes: storing the detection data of the pixels covered by the detection area in corresponding multiple storage units respectively according to the spatial correspondence between the pixel array and the storage unit array.

[0117] According to a preferred embodiment of the present disclosure, the data processing device 140 includes one or more convolution modules 141, and the step S14 includes:

[0118] The detection data stored in the storage device 130 is subjected to block-by-block convolution processing by the convolution module 141, specifically including performing convolution processing block by block with the storage unit 131 as a unit to obtain a reconstructed pixel array; and

[0119] A laser radar point cloud is generated based on the recombined pixel array.

[0120] According to a preferred embodiment of the present disclosure, step S14 includes: determining boundary attributes of the first direction and the second direction of each storage unit 131, wherein the boundary attributes include convolution boundaries or non-convolution boundaries; and determining the detection data to be transmitted to the convolution module 141 according to the boundary attributes of the first direction and the second direction of each storage unit 131.

[0121] According to a preferred embodiment of the present disclosure, the step of determining the detection data to be transmitted to the convolution module 141 includes: when the boundary attribute of the storage unit 131 in the first direction and / or the second direction is a non-convolution boundary, the detection data stored in the storage unit 131 and part of the detection data stored in other storage units 131 adjacent to the storage unit 131 along the first direction and / or the second direction are transmitted to the convolution module 141.

[0122] According to a preferred embodiment of the present disclosure, the step of determining the detection data to be transmitted to the convolution module 141 further includes: when the boundary attributes of the first direction and the second direction of the storage unit 131 are both convolution boundaries, transmitting the detection data stored in the storage unit 131 to the convolution module 141.

[0123] According to a preferred embodiment of the present disclosure, wherein the data processing device 140 further comprises an accumulator 142, the step S13 comprises: accumulating the detection data generated by the plurality of echo beams L' on each pixel 121 pixel by pixel, and storing the accumulated data in the storage unit 131 corresponding to the pixel.

[0124] According to a preferred embodiment of the present disclosure, the storage unit 131 is provided with an additional storage location for storing part of the detection data in other storage units 131 adjacent to the storage unit 131 .

[0125] According to a preferred embodiment of the present disclosure, the data processing device 140 further includes: a control module 143 and a cache module 144, wherein the control module 143 is coupled to the storage device 130; wherein the step S14 further includes: at least transmitting the detection data stored in each storage unit 131 to the cache module 144 block by block; and transmitting the detection data stored in the cache module 144 to the convolution module 141.

[0126] According to a preferred embodiment of the present disclosure, the step of delivering at least the detection data stored in each storage unit 131 to the cache module 144 block by block includes: delivering at least the detection data stored in each storage unit 131 to the cache module 144 block by block according to the boundary attributes of the first direction and the second direction of each storage unit;

[0127] When the boundary attribute of the first direction and / or the second direction of the storage unit 131 is a non-convolution boundary, the detection data stored in the storage unit 133 and part of the detection data stored in other storage units 131 adjacent to the storage unit 131 along the first direction and / or the second direction are transmitted to the cache module 144 through the control module 143; or

[0128] When the boundary attributes of the first direction and the second direction of the storage unit 131 are both convolution boundaries, the detection data stored in the storage unit 131 is transmitted to the cache module 144 through the control module 143 .

[0129] According to a preferred embodiment of the present disclosure, the detection areas with different properties include any one or more of the following: rectangular detection areas, discrete detection areas and strip detection areas.

[0130] In summary, the laser radar and detection method of the present invention have been introduced in detail. The laser radar of the present invention can reduce the complexity and amount of data processing, reduce computing power consumption, improve data processing speed, and reduce the size of the data processing circuit by storing and processing the detection data of pixels covered by the detection area in blocks. This is conducive to the miniaturization of the entire laser radar.

[0131] The laser radar disclosed in the present invention can perform block-by-block storage and block-by-block processing on the detection data of pixels covered by detection areas of different properties, and therefore has a wide range of applications.

[0132] The laser radar disclosed in the present invention uses a convolution module to perform block convolution processing on the detection data stored in the storage device, which can improve the data processing speed, improve the signal-to-noise ratio, and improve the laser radar's ranging capability and the accuracy of the detection results while reducing the amount of calculation and computing power consumption.

[0133] Furthermore, the laser radar disclosed in the present invention can further reduce the amount of calculation and power consumption while improving the data processing speed by performing convolution processing block by block in units of storage units.

[0134] The laser radar disclosed in the present invention determines the detection data transmitted to the convolution module according to the boundary attributes of the first direction and the second direction of each storage unit, overcomes or alleviates the adverse effect of the storage unit boundary on the convolution processing, further reduces the complexity of data processing, and further improves the data processing speed. In addition, based on the boundary attributes of the storage unit, the convolution processing operation can be operated within the range of the detection area, which is conducive to obtaining more accurate detection results.

[0135] The present disclosure also provides a computer-readable storage medium, including computer-executable instructions stored thereon, and when the executable instructions are executed by a processor, the laser radar detection method 10 as described above is implemented.

[0136] In some preferred embodiments, the computer-readable storage medium / storage device may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device, or device. The computer-readable storage medium / storage device may be any combination of one or more computer-readable media. The computer-readable storage medium / storage device may be, for example, but not limited to, an electrical, magnetic, optical, or semiconductor form or device, and more specific examples (a non-exhaustive list) include: an electrical connection with one or more wires, a portable computer hard disk, a hard disk, a random access memory (RAM), a non-volatile random access memory (NVRAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0137] In addition, the present disclosure also provides a computer device (not shown), including a memory and a processor, wherein the memory is configured to store computer executable instructions, and the executable instructions implement the detection method 10 described above when executed by the processor.

[0138] In some preferred embodiments, the processor / data processing device may include a central processing unit (CPU), and may also include other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The present disclosure does not impose any restrictions on this, depending on the specific circumstances.

[0139] It should be noted that although several modules of the laser radar are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules described above can be implemented in one module. Conversely, the features and functions of one module described above can be further divided into multiple modules to be embodied.

[0140] It should be noted that this specification provides method operation steps as described in the embodiments or schematic diagrams, but more or fewer operation steps may be included based on conventional or non-creative labor. The order of steps listed in the embodiments is only one way of executing the order of many steps and does not represent the only execution order. When the actual system or device product is executed, it can be executed in the order or in parallel according to the method shown in the embodiments or flowcharts.

[0141] Finally, it should be noted that the above is only a preferred embodiment of the present disclosure and is not intended to limit the present disclosure. Although the present disclosure is described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

Claims

1. A laser radar, comprising: An emitting device and a detecting device, wherein the emitting device is configured to emit a detection beam for detecting an obstacle, the detecting device comprises a pixel array, an echo beam after the detection beam is reflected on the obstacle can form a detection area on the pixel array, the pixel array comprises a plurality of pixels, each pixel can respond to the echo beam and generate detection data; A storage device, comprising a plurality of storage units distributed in an array, wherein each storage unit can store detection data of a plurality of pixels, and the number of pixels corresponding to each storage unit is less than the number of pixels covered by the detection area, so that the detection data of the pixels covered by the detection area are stored in blocks in the plurality of storage units of the storage device; and The data processing device is coupled to the detection device and the storage device, and is configured to perform block processing on the detection data stored in the storage device.

2. The laser radar according to claim 1, wherein the transmitting device is configured to emit different types of detection beams for detecting obstacles, wherein the echo beams of the different types of detection beams after being reflected on obstacles can form detection areas with different properties on the pixel array.

3. The laser radar according to claim 1, wherein the number of pixels corresponding to the storage device is not less than the number of pixels covered by any of the detection areas.

4. The laser radar according to claim 1, wherein the data processing device is configured to: store the detection data of the pixels covered by the detection area in blocks into corresponding multiple storage units according to the spatial correspondence between the pixel array and the storage unit array.

5. The laser radar according to claim 1, wherein the data processing device comprises one or more convolution modules, wherein the convolution modules are configured to: Performing block convolution processing on the detection data stored in the storage device, specifically including performing convolution processing block by block based on the storage unit to obtain a reconstructed pixel array; and A laser radar point cloud is generated based on the recombined pixel array.

6. The laser radar according to claim 5, wherein the data processing device is configured to: determine the boundary attributes of the first direction and the second direction of each storage unit, wherein the boundary attributes include convolution boundaries or non-convolution boundaries; and determine the detection data to be transmitted to the convolution module based on the boundary attributes of the first direction and the second direction of each storage unit.

7. The laser radar according to claim 6, wherein when the boundary attribute of the first direction and / or the second direction of the storage unit is a non-convolution boundary, the data processing device is configured to: transmit the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction to the convolution module.

8. The laser radar according to claim 6, wherein when the boundary attributes of the first direction and the second direction of the storage unit are both convolution boundaries, the data processing device is configured to: transmit the detection data stored in the storage unit to the convolution module.

9. The laser radar according to any one of claims 1-8, wherein the data processing device further comprises an accumulator, and the accumulator is configured to accumulate the detection data generated by the plurality of echo light beams on each pixel pixel by pixel and store them in a storage unit corresponding to the pixel.

10. The laser radar according to any one of claims 1-8, wherein the storage unit is provided with an additional storage location for storing part of the detection data in other storage units adjacent to the storage unit.

11. The laser radar according to any one of claims 1 to 8, wherein the data processing device further comprises: A control module and a cache module, wherein the control module is coupled to the storage device and configured to: at least transmit the detection data stored in each storage unit to the cache module block by block; and transmit the detection data stored in the cache module to the convolution module.

12. The laser radar according to claim 11, wherein the control module is configured to: at least transmit the detection data stored in each storage unit to the cache module block by block according to the boundary attributes of the first direction and the second direction of each storage unit; When the boundary attribute of the storage unit in the first direction and / or the second direction is a non-convolution boundary, the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction are transmitted to the cache module; or When the boundary attributes of the first direction and the second direction of the storage unit are both convolution boundaries, the detection data stored in the storage unit is transmitted to the cache module.

13. The laser radar according to claim 2, wherein the detection areas of different properties include any one or more of the following: rectangular detection areas, discrete detection areas and strip detection areas.

14. A detection method of a laser radar, the laser radar comprising a transmitting device, a detecting device, a storage device and a data processing device, wherein: The detection device comprises a pixel array, the emission device can emit a detection beam, and an echo beam after the detection beam is reflected on an obstacle can form a detection area on the pixel array, and the pixel array comprises a plurality of pixels, each pixel can respond to the echo beam and generate detection data; The data processing device is coupled to the detection device and the storage device; The detection method comprises: S11: emits a detection beam to detect obstacles; S12: responding to the echo light beam and generating detection data; S13: storing the detection data of the pixels covered by the detection area in blocks into a plurality of storage units of the storage device, wherein the storage device comprises a plurality of storage units distributed in an array, each storage unit can store the detection data of a plurality of pixels, and the number of pixels corresponding to each storage unit is less than the number of pixels covered by the detection area, so as to store the detection data of the pixels covered by the detection area in blocks into the plurality of storage units of the storage device; and S14: performing block processing on the detection data stored in the storage device.

15. The detection method according to claim 14, wherein the step S11 comprises: Different types of detection light beams are emitted to detect obstacles, wherein the echo light beams of the different types of detection light beams after being reflected on the obstacles can form detection areas with different properties on the pixel array.

16. The detection method according to claim 14, wherein the number of pixels corresponding to the storage device is not less than the number of pixels covered by any of the detection areas.

17. The detection method according to claim 14, wherein the step S13 comprises: According to the spatial correspondence between the pixel array and the storage unit array, the detection data of the pixels covered by the detection area are respectively stored in the corresponding multiple storage units.

18. The detection method according to claim 14, wherein the data processing device comprises one or more convolution modules, and the step S14 comprises: The detection data stored in the storage device is subjected to block-by-block convolution processing by the convolution module, specifically comprising performing convolution processing block by block in units of storage units to obtain a reconstructed pixel array; and A laser radar point cloud is generated based on the recombined pixel array.

19. The detection method according to claim 18, wherein the step S14 comprises: Determine the boundary attributes of the first direction and the second direction of each storage unit, wherein the boundary attributes include convolution boundaries or non-convolution boundaries; and determine the detection data transmitted to the convolution module according to the boundary attributes of the first direction and the second direction of each storage unit.

20. The detection method according to claim 19, wherein the step of determining the detection data to be transmitted to the convolution module comprises: When the boundary attribute of the storage unit in the first direction and / or the second direction is a non-convolution boundary, the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction are transmitted to the convolution module.

21. The detection method according to claim 19, wherein the step of determining the detection data transmitted to the convolution module further comprises: When the boundary attributes of the first direction and the second direction of the storage unit are both convolution boundaries, the detection data stored in the storage unit is transmitted to the convolution module.

22. The detection method according to any one of claims 14 to 21, wherein the data processing device further comprises an accumulator, and the step S13 comprises: The detection data generated by the plurality of echo light beams on each pixel are accumulated pixel by pixel and stored in a storage unit corresponding to the pixel.

23. The detection method according to any one of claims 14 to 21, wherein the storage unit is provided with an additional storage location for storing part of the detection data in other storage units adjacent to the storage unit.

24. The detection method according to any one of claims 14 to 21, wherein the data processing device further comprises: a control module and a cache module, wherein the control module is coupled to the storage device; The step S14 further comprises: at least transmitting the detection data stored in each storage unit to the cache module block by block; The detection data stored in the cache module is transmitted to the convolution module.

25. The detection method according to claim 24, wherein the step of at least transmitting the detection data stored in each storage unit to the cache module block by block comprises: According to the boundary attributes of the first direction and the second direction of each storage unit, at least the detection data stored in each storage unit is transmitted to the cache module block by block; When the boundary attribute of the storage unit in the first direction and / or the second direction is a non-convolution boundary, the detection data stored in the storage unit and part of the detection data stored in other storage units adjacent to the storage unit along the first direction and / or the second direction are transmitted to the cache module through the control module; or When the boundary attributes of the first direction and the second direction of the storage unit are both convolution boundaries, the detection data stored in the storage unit is transmitted to the cache module through the control module.

26. The detection method according to claim 15, wherein the detection areas of different properties include any one or more of the following: rectangular detection areas, discrete detection areas and strip detection areas.

Images