Perceptual diversity for object recognition in robotic systems and applications

Abstract

The present disclosure relates to perceptual diversity for object recognition in robotic systems and applications, in particular to the use of a variety of analytical techniques to detect objects in a detection area. A variety of analysis techniques may be used to analyze sensor data corresponding to the detection region. SELECTING MULTIPLE BASED AT LEAST ON AT LEAST TWO OF A PLURALITY OF ANALYTIC TECHNOLOGIES HAVING COMPUTE DEVELOCITY BY PERFORMING COMPUTE ANALYTICS OF SENSOR DATA OF DIFFERENT TYPES BY PERFORMING COMPUTE ANALYTICS OF DIFFERENT TYPES and an analysis technology.

Description

Background Art

[0001] Object detection is an important task for many robotic and autonomous systems. For example, an object detection system may be used to detect objects (e.g., people, animals, materials, stationary objects, moving objects, and / or other objects), and more specifically, may be used to detect objects within certain areas, which may be designated as object detection areas ("detection areas"). For example, a detection system may be used to determine whether an object is within a certain range of a machine's operating area, where the operating area may include an area within a certain distance of the machine and may therefore be designated as a detection area.

[0002] Object detection may be used in certain safety measures to help prevent machines from operating unsafely and / or damaging property. For example, areas or spaces surrounding or containing the robot's operating area that are designated as potentially hazardous areas may be included in detection zones where it is critical to perform object detection. For example, in response to making an object detection corresponding to a detection zone, the detection system may cause the machine to operate in a particular manner, such as slowing down or stopping operation. For example, in response to the system making a detection corresponding to a detection zone, the machine may be stopped.

[0003] Currently, some methods for performing detection corresponding to a detection area include using a two-dimensional (2D) planar laser rangefinder or laser radar to scan within the detection area. For example, a 2D planar laser rangefinder or laser radar can be used to scan a circular plane of 270 to 360 degrees around the machine. In these cases, detection can be performed when an object intersects the circular plane. For example, a 2D planar laser rangefinder or laser radar can detect objects within the circular plane. However, this method is limited to detection in 2D space and may not be applicable in some cases. For example, this method may not be able to correctly detect in a three-dimensional (3D) plane. For example, an object may be located above the scanned circular plane, such as a flying object and / or a hanging object. A 2D planar laser rangefinder or laser radar may not be able to correctly detect these flying objects and / or hanging objects.

[0004] In addition, some current methods may be limited due to the use of specific detection technologies. For example, a specific detection system may use a specific detection technology to analyze input data obtained using a sensor to determine a detection result. For example, a specific detection technology may include an algorithm for analyzing input data. In some cases, the algorithm may not be suitable for a specific detection area or the specifications of an example system. For example, the algorithm may need to be pre-configured to calculate a specific set of disparities, which may prevent inference time adjustment. Such an algorithm may not be suitable for detection area requirements because the requirement may require flexible use of the trade-off between latency and depth quantization at inference time. Summary of the invention

[0005] According to one or more embodiments of the present disclosure, one or more systems and / or methods may be configured to detect objects in a detection area using multiple analysis techniques. For example, in some embodiments, multiple analysis techniques may be used to analyze sensor data corresponding to a detection area obtained using one or more sensors. In some embodiments, multiple analysis techniques may be selected based on at least two of the multiple analysis techniques with computational diversity, by performing different types of computational analysis on the sensor data relative to each other. In some embodiments, in order to further increase robustness and diversity, at least two of the multiple analysis techniques may be implemented on different types of computing or hardware platforms, devices, or components, respectively.

[0006] One or more embodiments of the present disclosure may help more accurately detect the presence of an object in a detection zone. Additionally or alternatively, detection of an object may be performed in a variety of ways, which may improve the robustness of the detection. For example, according to one or more embodiments, the diverse ways of detecting may improve the accuracy of detecting objects in various detection zones under different detection conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] The present system, method and application for detecting objects within a detection area are described in detail below in conjunction with the accompanying drawings, wherein:

[0008] Figure 1A-1B An example system configured to determine object detection in a detection area according to one or more embodiments of the present disclosure is shown;

[0009] Figure 1C An example schematic diagram 130 is shown that visualizes the tradeoff between safety and productivity when combining one or more detection algorithms in accordance with one or more embodiments of the present disclosure;

[0010] Figure 1D shows example detection areas according to some embodiments of the present disclosure;

[0011] Figure 2A-2B An example system configured to determine object detection in a detection area according to one or more embodiments of the present disclosure is shown;

[0012] Figure 3 is an example flow chart illustrating a method for detecting an object in a detection area of ​​a machine according to one or more embodiments of the present disclosure;

[0013] Figure 4A is an illustration of an example autonomous vehicle according to one or more embodiments of the present disclosure;

[0014] Figure 4B According to one or more embodiments of the present disclosure Figure 4A Examples of autonomous vehicle camera positions and fields of view;

[0015] Figure 4C According to one or more embodiments of the present disclosure Figure 4A a block diagram of an example system architecture for an example autonomous vehicle;

[0016] Figure 4D A cloud-based server and Figure 4A System diagram of communication between example autonomous vehicles;

[0017] Figure 5 is a block diagram of an example computing device suitable for implementing one or more embodiments of the present disclosure; and

[0018] Figure 6 is a block diagram of an example data center suitable for implementing one or more embodiments of the present disclosure. DETAILED DESCRIPTION

[0019] The present disclosure discloses systems and methods related to detecting objects in certain areas, which are referred to as "detection areas". In some embodiments, the detection area may include any area of ​​interest around or near a machine. For example, the detection area may include an operating area of ​​a vehicle or machine (commonly referred to as a "machine"). For example, the operating area of ​​a machine may include any area where a part of the machine can move or operate. For example, an operating area corresponding to a robotic arm may include any area where the robotic arm can move or move to. Additionally or alternatively, the operating area may include any area into which the operation of the machine may introduce something. For example, the operating area may include an area where electromagnetic waves, radiation, etc. (e.g., X-rays, laser beams, etc.) may exist due to emissions from the machine.

[0020] Additionally or alternatively, the detection area may include a hazardous area or a restricted area (often referred to as a "restricted area"). For example, a restricted area may include an area where access to persons, animals, substances, or objects may need to be restricted due to confidentiality considerations, contamination considerations, safety considerations, etc. In some cases, a restricted area may include an area that may be restricted with respect to access and / or objects or substances (e.g., dust, radiation, certain gases, etc.). In the present disclosure, a general reference to an object or multiple objects may include persons, substances, or any other applicable type of thing that may enter the detection area.

[0021] In some embodiments, the detection area may include an area related to the self-machine. For example, the detection area may include an area in front of, behind, and / or around the self-machine. Additionally or alternatively, the detection area may include an area above or below the self-machine. In these or other embodiments, the detection area may move with the self-machine as the self-machine moves. For example, graphically shown, a specific detection area may be represented as a boundary shape surrounding the self-machine. The boundary shape may be determined to be a specific size and shape based on the detection capabilities or requirements of the self-machine and the machine. For example, the boundary shape may be determined to cover an area that the self-machine should keep a distance from to avoid collision. In these and other embodiments, as the self-machine moves, the boundary shape may move with the self-machine. For example, the boundary shape may maintain a position and / or orientation relative to the current position and / or posture of the self-machine.

[0022] In some embodiments, object detection may be performed based at least on detection data generated using one or more sensors. For example, in some embodiments, the detection system may include one or more analysis algorithms that may be applied to the detection data to determine that an object is detected in a detection area.

[0023] One or more embodiments disclosed herein may involve detecting an object in a detection area associated with one or more self-machines, which may include any applicable machine or system capable of performing one or more autonomous or semi-autonomous operations. Example self-machines may include, but are not limited to, vehicles (land, sea, space, and / or air), robots, robotic platforms, etc. For example, a self-machine computing application may include one or more applications executable by an autonomous vehicle or semi-autonomous vehicle, such as Figures 4A-4D An example autonomous vehicle 400 (referred to herein interchangeably as "vehicle 400" or "self-machine 400") is described. In this disclosure, references to an "autonomous vehicle" or "semi-autonomous vehicle" may include any vehicle that can be configured to perform one or more autonomous or semi-autonomous navigation or driving operations. Thus, such vehicles may also include vehicles that require an operator or that can also perform such operations.

[0024] The systems and methods described herein may be used by, but are not limited to, non-autonomous vehicles or machines, semi-autonomous vehicles or machines (e.g., in one or more advanced driver assistance systems (ADAS)), autonomous vehicles or machines, driverless and driverless robots or robotic platforms, warehouse vehicles, off-road vehicles, vehicles coupled to one or more trailers, airships, boats, shuttles, emergency response vehicles, motorcycles, electric or motorized bicycles, aircraft, engineering vehicles, underwater vehicles, drones, and / or other vehicle types. Further, the systems and methods described herein may be used for a variety of purposes, including, by way of example and not limitation, for machine control, machine motion, machine driving, synthetic data generation, model training, perception, augmented reality, virtual reality, mixed reality, robotics, safety and surveillance, simulation and digital twins, autonomous or semi-autonomous machine applications, deep learning, environmental simulation, object or actor simulation and / or digital twins, data center processing, conversational AI (e.g., by employing one or more language models (e.g., one or more large language models (LLMs))), light transport simulation (e.g., ray tracing, path tracing, etc.), collaborative content creation for 3D assets, cloud computing, and / or any other suitable application.

[0025] The disclosed embodiments may be included in a variety of different systems, such as automotive systems (e.g., control systems for autonomous or semi-autonomous machines, perception systems for autonomous or semi-autonomous machines), systems implemented using robots, aerial systems, medical systems, boating systems, smart area monitoring systems, systems for performing deep learning operations, systems for performing simulation operations, systems for performing digital twin operations, systems implemented using edge devices, systems including one or more virtual machines (VMs), systems for performing synthetic data generation operations, systems implemented at least in part in a data center, systems for performing conversational AI operations (e.g., systems implementing one or more language models such as large language models (LLMs)), systems for hosting real-time streaming applications, systems for presenting one or more of virtual reality content, augmented reality content, or mixed reality content, systems for performing light transport simulations, systems for performing collaborative content creation of 3D assets, systems implemented at least in part using cloud computing resources, and / or other types of systems.

[0026] Embodiments of the present disclosure help improve the accuracy of object detection in a detection area. For example, in one or more embodiments of the present disclosure, a detection system can perform detection of an object in a variety of ways that can improve the accuracy of the detection system. For example, according to one or more embodiments, a detection system can use different techniques that can be used for object detection in a detection area. In some embodiments, different techniques can be selected based on computational diversity between different techniques. Additionally or alternatively, different techniques can be selected based on implementation diversity between different techniques.

[0027] One or more embodiments of the present disclosure contribute to improving accuracy relative to some traditional object detection methods. For example, some traditional object detection methods may include the use of a two-dimensional (2D) planar laser rangefinder or laser radar. For example, a 2D planar laser rangefinder or laser radar can be used to scan a circular plane of 270 to 360 degrees around the detection system. In these cases, detection can be performed when the object intersects the circular plane. However, this method may be limited to 2D space and / or may be limited in height (based on the type of laser radar sensor deployed), and may not be applicable to certain situations. For example, when the detection area includes a vertical plane, this method may not be valid. For example, the detection area may include areas where the ground is uneven and / or areas where the object may be located in different vertical planes. Using a 2D planar laser rangefinder or laser radar, uneven ground may cause false positive detections due to some parts of the ground intersecting the circular plane.

[0028] In addition, some conventional methods may include using a specific computing platform. For example, the specific computing platform may include a specific hardware platform and / or a specific software platform that can acquire sensor data and process the sensor data to determine the detection. However, using a specific computing platform may present a risk of common failures. For example, computational errors may occur on a specific computing platform, which may result in false positive detections and / or missed detections.

[0029] Embodiments of the present disclosure will be explained with reference to the accompanying drawings. It should be understood that the accompanying drawings are diagrams and schematics of such example embodiments and are not restrictive and are not necessarily drawn to scale. In the drawings, unless otherwise specified, features with the same number represent the same structure and function.

[0030] With reference to the accompanying drawings, Figure 1A-1B An example system 100 configured to determine object detection in a detection area according to one or more embodiments of the present disclosure is shown. In some embodiments, the system 100 can be implemented on a machine. For example, the system 100 can be implemented on a Figures 4A-4D For example, the system 100 may be configured to detect objects in a detection area of ​​the vehicle 400. In some cases, the system 100 may be used by the vehicle 400 to detect objects while the vehicle 400 is traveling. In another example, the system 100 may be implemented on a machine with moving parts (e.g., a robotic arm). In these cases, the system 100 may be configured to detect objects in an operating area of ​​the machine (e.g., an area that the robotic arm can reach and / or needs to operate safely).

[0031] As described herein, in general, system 100 may include a first computing platform (“first platform”) 104, a second computing platform (“second platform”) 106, and a third computing platform (“third platform”). In some embodiments, system 100 may include additional computing platforms. First platform 104, second platform 106, and third platform 108 may be configured to determine first detection 116a, second detection 116b, and third detection 116c, respectively, based on image data 102 and / or other sensor data (e.g., lidar, radar, ultrasonic, etc.). Therefore, when image data 102 is mentioned herein, it may also include sensor data of other types or modalities.

[0032] The image data 102 may include any type of data representing a detection area. For example, the image data 102 may include an image depicting the detection area. For example, in some embodiments, the image data 102 may be obtained using one or more sensors. The one or more sensors may include any suitable system, device, or equipment that can be used to obtain the image data 102, such as a laser scanner, a camera system, and / or any other suitable optoelectronic device. In the present disclosure, reference to an "image" may also refer to image data corresponding thereto. Similarly, an image may include representations of image data, as well as other sensor data representations (e.g., range images, point clouds, etc.) associated with other sensor data modalities that are also used herein without departing from the scope of the present disclosure.

[0033] In some embodiments, image data 102 obtained using one or more sensors may include a certain number of pixels. For example, the certain number of pixels may depend on the size of image data 102, specifications of one or more sensors, specifications of a computing platform, etc. For example, one or more sensors may be relatively high-resolution sensors and accordingly may be used to generate image data having a greater number of pixels than one or more sensors having a lower resolution.

[0034] In some embodiments, the number of pixels and / or other properties of the image data 102 may be transformed. For example, the image data 102 obtained using one or more sensors may be pre-processed using one or more pre-processing functions to make it more suitable for processing by the first platform 104, the second platform 106, and / or the third platform 108. For example, the image data 102 may be transformed to a specific size (e.g., number of pixels) and format. The one or more pre-processing functions may include any suitable functions to convert the raw image data into a format more suitable for processing. For example, the one or more pre-processing functions may include format conversion, noise reduction, tone mapping, multiple exposure merging, distortion correction, cropping, reduction, correction, etc.

[0035] In some embodiments, one or more objects may be represented in the image data 102. Furthermore, in the present disclosure, an object represented in the image data 102 and depicted by a corresponding image may also be referred to as being "present in" or "in" the image data 102 or the corresponding image. Furthermore, a pixel of the image data 102 that includes information corresponding to an object may be referred to as being "occupied by" the object. Similarly, a pixel that depicts an object or a portion of an object may be referred to as "comprising" the object, or the object may be referred to as being "present in" or "in" the pixel. Furthermore, regions and / or objects depicted or represented by the image data 102 and / or pixels may be considered to "correspond to" the image data 102 and / or pixels. Furthermore, in the present disclosure, a reference to a pixel may include a reference to one or more objects or regions corresponding to the pixel.

[0036] In some embodiments, the image data 102 may include information about which pixels are occupied by one or more objects. For example, if at least a portion of a particular pixel includes at least a portion of one or more objects, the particular pixel may be considered occupied. Conversely, if a portion of one or more objects is not present in the particular pixel, the particular pixel may be considered unoccupied. For example, the particular pixel corresponds to an area in the detection area that does not have any objects. For example, the particular pixel may correspond to an area that includes a background, sky, etc.

[0037] In some embodiments, one or more sensors may include a stereo camera module. In these and other embodiments, the stereo camera module may include two or more cameras that may be used to generate a three-dimensional (3D) depiction of an area. In some embodiments, the area depicted by the stereo camera module may correspond to the detection area. In some embodiments, the two or more cameras may be configured to acquire images of the area from different locations. For example, the stereo module may include a first camera and a second camera, wherein the first camera may be configured to acquire a first image of the area, and the second camera may be configured to acquire a second image of the area. In some cases, the first camera and the second camera may be offset from each other so that the first camera and the second camera have different viewing angles. In these and other embodiments, the first image and the second image may have a parallax due to different viewing angles. For example, one or more objects found in the first image and the second image may occupy different sets of pixels. In these cases, the parallax may indicate a displacement (e.g., a distance) between different sets of pixels of the first image and the second image. In some embodiments, the presence of a parallax may indicate that there is depth in the area, and the parallax may be used to estimate the depth. In some embodiments, the image data 102 may include at least the first image and the second image. Additionally or alternatively, the image data 102 may include a parallax between the first image and the second image.

[0038] In some embodiments, the first platform 104, the second platform 106, and the third platform 108 may include a first depth module 110a, a second depth module 110b, and a third depth module 110c, respectively (collectively referred to herein as "depth modules 110"). In some embodiments, one or more depth modules 110 may be implemented using hardware, including one or more processors, central processing units (CPUs), graphics processing units (GPUs), data processing units (DPUs), parallel processing units (PPUs), microprocessors (e.g., for performing one or more operations or controlling the execution of one or more operations), field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), accelerators (e.g., deep learning accelerators (DLAs), optical flow accelerators (OFAs)), programmable vision accelerators (including one or more direct memory address (DMA) systems and / or vector processing units (VPUs)), and / or other processor types. In some other cases, one or more of these modules may be implemented using a combination of hardware and software. In the present disclosure, an operation described as being performed by a corresponding module may include an operation that the corresponding module instructs the corresponding computing system to perform. In these or other embodiments, one or more of these modules may be provided by (e.g. Figures 4A-4D , Figure 5 and / or Figure 6 In some embodiments, the first depth module 110a, the second depth module 110b, and the third depth module 110c may be implemented using different hardware. For example, in some embodiments, the first depth module 110a may be implemented using OFA, the second depth module 110b may be implemented using DLA, and the third depth module 110c may be implemented using a GPU.

[0039] In some embodiments, one or more depth modules 110 may be configured to obtain image data 102. One or more depth modules 110 may further be configured to identify one or more objects depicted by image data 102. For example, one or more depth modules 110 may be configured to determine the number of objects present in image data 102 and / or the location of the objects. Additionally or alternatively, one or more depth modules 110 may be configured to determine depths corresponding to one or more objects identified in image data 102. For example, a first depth module 110a may be configured to determine a first depth 112a, a second depth module 110b may be configured to determine a second depth 112b, and a third depth module 110c may be configured to determine a third depth 112c. The first depth 112a, the second depth 112b, and the third depth 112c may be generally or collectively referred to as "multiple depths 112" or "depths 112".

[0040] In some embodiments, the one or more depths 112 may include a relationship between the one or more objects and a reference point. For example, in some embodiments, the one or more depths may be determined as a continuous depth, which may refer to the distance of the one or more objects from a reference point (e.g., an origin or rig coordinate corresponding to a machine or ego machine, such as the center of a vehicle's rear axle, or a point on a robotic arm).

[0041] Additionally or alternatively, one or more depths 112 may correspond to a disparity of one or more objects from a reference point. For example, objects farther from a reference point may have a greater depth than objects closer to the reference point. In these and other embodiments, any suitable measurement metric may be used to represent the distance corresponding to one or more depths 112. For example, the distance may be expressed in meters, centimeters, inches, etc.

[0042] In some embodiments, the reference point may correspond to the location of one or more sensors. For example, the reference point may be determined as the location of a stereo camera module. In these cases, one or more depths 112 of one or more objects may be determined as the distance from the stereo camera module to the one or more objects. In some embodiments, the reference point may be determined as any other suitable location relative to the system 100 and the detection area. For example, in some embodiments, the reference point may be located at the center, edge, front, back, or any other portion of a machine implementing the system 100.

[0043] Additionally or alternatively, in some embodiments, one or more depths 112 may be represented in a binary classification format. For example, depth may refer to whether one or more objects are closer or farther than a particular depth plane. For example, one or more depth planes may be located, each of the one or more depth planes being at a different distance from a reference point. In these and other embodiments, the location of one or more objects may be compared to a particular depth plane to determine whether one or more objects are closer or farther from a reference point than the particular depth plane. In some embodiments, one or more objects may be compared to multiple depth planes, which may provide a more refined depth quantification.

[0044] For example, the location of a particular object may be compared to a first depth plane and a second depth plane. It may be determined that the particular object is farther from a reference point than the first depth plane, but closer to the reference point than the second depth plane. In these cases, the particular object may be located between the first depth plane and the second depth plane. In these cases, comparing one or more objects to more depth planes may result in more detailed depth quantification of the one or more objects. In these and other embodiments, one or more depths 112 may indicate the relative depth relationship between the particular object and the multiple depth planes, respectively.

[0045] In some embodiments, the distance between one or more depth planes can be configured according to a specified depth quantization. For example, where greater depth quantization is desired, the distance between one or more depth planes can be narrower, and where less depth quantization is desired, the distance between one or more depth planes can be larger.

[0046] In some embodiments, one or more depths 112 may be determined on a pixel-by-pixel basis for pixels corresponding to one or more objects. In some embodiments, different portions of one or more objects may have different depths. In these and other embodiments, the depths of one or more objects may be determined by determining corresponding depths of corresponding portions of one or more objects corresponding to the pixels. For example, individual pixel depths may be determined for pixels occupied by one or more objects. For example, a distance between a reference point and each pixel occupied by one or more objects may be determined. In these and other embodiments, individual pixel depths may be used to determine one or more depths 112 in a continuous depth format.

[0047] Additionally or alternatively, individual pixel depths of pixels occupied by one or more objects may be determined in a binary classification format. This approach may provide a more detailed depth quantification than determining the depth of one or more objects as a whole. For example, a portion of an object represented by a particular pixel may be located closer to a reference point than a certain depth plane, while the remainder of the object may be located farther than a certain depth plane. This approach may indicate how only a small portion of an object is at a certain depth.

[0048] In some embodiments, a depth map may be used to represent one or more depths 112. The depth map may include a 3D representation of the depths of some (e.g., all) pixels present in the image data 102. For example, one or more objects may be projected into the 3D space based on individual pixel depths. For example, each pixel corresponding to one or more objects may be placed at certain locations in the 3D space based on a reference point and individual pixel depths.

[0049] In some embodiments, one or more depth modules 110 may be configured to determine the depth of one or more objects in different methods. For example, the first depth module 110a may determine the depth in a continuous depth format, while the second depth module 110b may determine the depth in a binary classification format. Additionally or alternatively, the different methods may involve using machine learning (ML)-based techniques instead of using one or more computer vision-based techniques. The different methods may also include analyzing or aggregating different properties of the data. For example, some ML-based techniques may implement 2D or 3D convolution techniques as part of training and / or depth determination. Additionally or alternatively, other ML-based techniques may implement iterative refinement techniques in which each iteration improves the estimate of the previous step as part of training and / or depth determination.

[0050] In some embodiments, different methods may include differences between the ability to manipulate one or more parameters corresponding to the method. In these and other embodiments, some methods may allow one or more parameters to be manipulated at runtime. Additionally or alternatively, some different methods may include flexible ranges, which may include the ability to define a certain range at runtime or near runtime, such that the method can only estimate the depth of objects within the defined range.

[0051] In some embodiments, one or more depth modules 110 may be implemented on different computing platforms. For example, in some embodiments, first platform 104, second platform 106, and third platform 108 may include hardware and / or software platforms, wherein at least two of first platform 104, second platform 106, and third platform 108 are different relative to each other. In some embodiments, different computing platforms may result in different results for one or more depth modules 110. Figure 2A-2B Different computing platforms are discussed in more detail.

[0052] Additionally or alternatively, one or more depth modules 110 may be configured to determine a depth confidence. For example, the first depth 112a may include a first depth confidence. In some embodiments, the first depth confidence may indicate how confident the first depth module 110a is about the depth of one or more objects determined by the first depth module 110a. For example, the first depth confidence may indicate how confident the first depth module 110a is about the depth determined for each of the one or more objects. In some embodiments, the first depth confidence may include how confident the first depth module 110a is about the depth determined for pixels occupied by the one or more objects.

[0053] In some embodiments, the depth determined by one or more depth modules 110 may be represented in the same format as the image data 102. For example, the image data 102 may be converted to a 16-bit image format for the first depth module 110a. In these and other cases, the first depth 112 may also be represented in a 16-bit image format.

[0054] In some embodiments, the first platform 104, the second platform 106, and the third platform 108 may include a first detection module 114a, a second detection module 114b, and a third detection module 114c, respectively (collectively referred to herein as "detection module 114"). In some embodiments, the detection module 114 may be implemented using the following hardware, including one or more processors, central processing units (CPUs), graphics processing units (GPUs), data processing units (DPUs), parallel processing units (PPUs), microprocessors (e.g., for executing or controlling the execution of one or more operations), field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), accelerators (e.g., deep learning accelerators (DLAs)), programmable vision accelerators (including one or more direct memory address (DMA) systems and / or vector processing units (VPUs)), and / or other processor types. In some other cases, one or more of these modules may be implemented using a combination of hardware and software. In the present disclosure, operations described as being performed by a corresponding module may include operations that the corresponding module may instruct the corresponding computing system to perform. In these or other embodiments, one or more of these modules may be implemented by (e.g. Figures 4A-4D , Figure 5 and / or Figure 6 The invention is implemented by one or more computing devices described in further detail.

[0055] In some embodiments, one or more detection modules 114 may be configured to obtain one or more depths 112. One or more detection modules 114 may be further configured to determine whether the depth of any object present in the one or more depths 112 is within a certain area of ​​the detection area. For example, the first detection module 114a may be configured to determine a first detection 116a, the second detection module 114b may be configured to determine a second detection 116b, and the third detection module 114c may be configured to determine a third detection 116c. The first detection 116a, the second detection 116b, and the third detection 116c may be generally or collectively referred to as "multiple detections 116" or "detections 116".

[0056] In some embodiments, one or more detections 116 may indicate whether detection of an object is performed within the detection area. In some embodiments, the detection may be represented using one or more voxels. For example, the detection area may be divided into one or more voxels or 3D cubes. One or more voxels may create a 3D cube grid covering the detection area. In some embodiments, one or more voxels may have the same size. In some embodiments, the size of one or more voxels may be determined based on the format of one or more depths 112. For example, one or more depths 112 may be represented in a 16-bit image format. In these cases, the size of one or more voxels may be determined to divide the detection area according to the 16-bit image format. In some embodiments, the first detection 116a, the second detection 116b, and the third detection 116c may be configured to have the same format. For example, one or more voxels of the first detection 116a, the second detection 116b, and the third detection 116c may be located at the same position and / or have the same size. However, in other embodiments, one or more voxels of the first detection 116a, the second detection 116b, and the third detection 116c may be located at different positions and / or have different sizes.

[0057] In some embodiments, the one or more detection modules 114 may be configured to project a depth map representing the one or more depths 112 onto one or more voxels. For example, pixels corresponding to different portions of the one or more objects may be placed on a 3D cube grid according to individual pixel depths. In these and other embodiments, the one or more detection modules 114 may be configured to determine which of the one or more voxels include a depth of any one of the one or more objects.

[0058] Additionally or alternatively, one or more detection modules 114 may be configured to determine whether one or more objects detected within the detection area are located within certain regions of interest within the detection area. For example, one or more detections 116 may indicate whether the depth of any one of the one or more objects falls within certain regions of interest. For example, one or more voxels that exist with at least a portion of one or more objects may be checked to determine where the one or more objects are located relative to the detection area. In some embodiments, certain regions of interest may include portions of the detection area. For example, certain regions of interest may include portions of the detection area within a certain distance from a machine implementing system 100. For example, an area near a machine may be of particular concern for safety reasons. In these cases, one or more voxels within a certain distance of the machine may be designated as regions of interest. In these cases, detection 116 may indicate whether any one of the one or more voxels within the region of interest includes any portion of one or more objects.

[0059] In some embodiments, the detection area may include one or more regions of interest. For example, the detection area may be divided into one or more sub-detection areas, each sub-detection area covering a specific region of interest. In some embodiments, the sub-detection areas may be determined based on the operating characteristics of the machine. For example, the machine may include a moving part (e.g., a robotic arm) that can operate within a specific operating area. In these and other embodiments, a first sub-detection area may include any area that a moving part can reach during operation, a second sub-detection area may include any area that should be kept unobstructed to prevent interference with operation, and a third sub-detection area may include an area that a moving part cannot reach. About the present disclosure Figure 1D Examples of different sub-detection areas may be discussed in more detail.

[0060] Additionally or alternatively, one or more detection modules 114 may be configured to determine a detection confidence. For example, the first detection 116a may include a first detection confidence, the second detection 116b may include a second detection confidence, and the third detection 116c may include a third detection confidence. In some embodiments, the first detection confidence may indicate how confident the first detection module 114a is in the detection of one or more objects determined by the first detection module 114a. For example, the first detection confidence may indicate how confident the first detection module 114a is in the detection determined for each of the one or more objects. In some embodiments, the first detection confidence may include how confident the first detection module 114a is in the detection of one or more objects in each of the one or more voxels. In some embodiments, the first detection confidence may be expressed as a number within a confidence range (e.g., 0-100). In some embodiments, the first detection confidence, the second detection confidence, and the third detection confidence may be expressed on the same confidence range.

[0061] In some embodiments, the system 100 may include an aggregation module 118. In some embodiments, the aggregation module 118 may be configured to obtain the first detection 116a, the second detection 116b, and the third detection 116c. The aggregation module 118 may also be configured to determine a final detection 120 based on at least the first detection 116a, the second detection 116b, and / or the third detection 116c.

[0062] In some embodiments, the aggregation module 118 may be implemented using the following hardware, including one or more processors, central processing units (CPUs), graphics processing units (GPUs), data processing units (DPUs), parallel processing units (PPUs), microprocessors (e.g., for performing or controlling the execution of one or more operations), field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), accelerators (e.g., deep learning accelerators (DLAs), optical flow accelerators (OFAs)), programmable vision accelerators (including one or more direct memory address (DMA) systems and / or vector processing units (VPUs)), and / or other processor types. In some other cases, the aggregation module 118 may be implemented using a combination of hardware and software. In the present disclosure, operations described as being performed by a corresponding module may include operations that the corresponding module may instruct the corresponding computing system to perform. In these or other embodiments, the aggregation module 118 may be implemented by (e.g., regarding Figure 4A – 4D, Figure 5 and / or Figure 6 The invention may be implemented by one or more computing devices (described in more detail).

[0063] In some embodiments, the final detection 120 may indicate whether detection of one or more objects is performed in the detection area. For example, the final detection 120 may indicate whether the one or more objects are in the image data 102. Additionally or alternatively, the final detection 120 may indicate the location at which the one or more objects are detected in the detection area. In some embodiments, the final detection 120 may be the result of combining the first detection 116a, the second detection 116b, and the third detection 116c. By combining two or more detections 116, the final detection 120 may more accurately indicate that one or more objects are detected in the detection area than using only one detection 116.

[0064] In some embodiments, the aggregation module 118 may be configured to aggregate the detections 116. For example, the first detection 116a, the second detection 116b, and the third detection 116c may be aggregated to determine the final detection 120. In some embodiments, the final detection 120 may be represented in the same format as the detections 116. For example, the final detection 120 may be represented using a final set of voxels, which may be located and sized the same as one or more voxels of the detections 116. In these and other embodiments, each voxel in the final set of voxels may have a corresponding voxel in each of the detections 116. In these and other embodiments, the final detections 120 may be determined voxel by voxel. For example, each of the final sets of voxels of the final detections 120 may be determined by aggregating corresponding voxels from the detections 116. In these and other cases, each voxel in the final set of voxels of the final detections 120 may indicate whether an object is present in the corresponding voxel.

[0065] In some embodiments, the aggregation of corresponding voxels may be done based on one or more combining algorithms. In this disclosure, a combining algorithm may refer to any suitable method of combining corresponding voxels from detection 116 to determine a final set of voxels for final detection 120.

[0066] In some embodiments, the particular combination algorithm may include a voting algorithm that may involve combining detections 116 based at least on the detection confidences included in detections 116. For example, for a final set of voxels (e.g., each voxel) of final detections 120, aggregation module 118 may analyze the respective voxels associated with each detection 116 and the detection confidences associated with the respective voxels.

[0067] For example, the first detection 116a may include a first detection confidence, the second detection 116b may include a second detection confidence, and the third detection 116c may include a third detection confidence. In these and other embodiments, the aggregation module 118 may place a greater weight on detections with higher detection confidences. For example, in the event that there is a conflict between multiple detections 116, a detection with a higher confidence may be preferred over another detection with a lower confidence.

[0068] For example, for a particular voxel in the final voxel set, the corresponding first voxel from the first detection 116a may indicate a detection (e.g., an object is present in the corresponding first voxel), while the corresponding second voxel from the second detection 116b and the corresponding third voxel from the third detection 116c may indicate a non-detection (e.g., an object is not present). In this case, the aggregation module 118 may observe the detection confidences of the first detection 116a, the second detection 116b, and the third detection 116c, and the detections with higher confidences may be given greater weight in determining the final detection 120 than the detections with lower confidences. For example, the first detection 116a may indicate that the detection for the particular voxel is complete with confidence (e.g., 100), while the second detection 116b and the third detection 116c may each indicate that the non-detection for the particular voxel is less confident (e.g., 30). In some of these cases, the detection of first detection 116a may be superior to the non-detection of second detection 116b and third detection 116c, and a particular voxel of final detection 120 may be determined to have a detection even though only one of the three detections indicated a detection.

[0069] In the event that there is a conflict between detections 116, aggregation module 118 may be configured to compare the overall confidence of the detections with the overall confidence of the non-detections. For example, in the above example, the overall confidence of the detections may be 100. In contrast, the overall confidence of the non-detections may be 60 (e.g., 30+30). Since the overall confidence of the detections is greater than the overall confidence of the non-detections, aggregation module 118 may determine that there is a detection in the particular voxel.

[0070] In some embodiments, the combined algorithm may include determining the trade-off between safety and productivity. Figure 1C An example diagram 130 of visualizing tradeoffs is shown. In some embodiments, the diagram 130 may include a sliding scale 132 indicating a preference between safety and productivity, in which an indicator 134 may be disposed. In these and other embodiments, when the indicator 134 moves toward a first side 136 of the sliding scale 132, it may indicate that safety takes precedence over productivity. Conversely, when the indicator 134 slides toward a second end 138 of the sliding scale 132, the productivity preference may increase, while the safety preference may decrease. In some embodiments, the aggregation module 118 may define a threshold requirement based at least on the position of the indicator 134 on the sliding scale 132. In these and other embodiments, the threshold requirement may define a number of detections required for the detection 116 to determine a detection for the final detection 120. For example, a particular voxel of the final detection 120 may be determined to represent the detection of one or more objects where the threshold requirement number of corresponding voxels from the detection 116 indicates a detection.

[0071] In some embodiments, safety may be more important than productivity. For example, as shown by the first indicator 134a, the indicator 134 may be placed closer to the first side 136 than the second end 138 of the sliding scale 132. In these cases, the threshold requirement may be set to a relatively low number, making it easier to meet, thereby making it easier to determine the detection for the final detection 120. For example, the threshold requirement may be set to indicate a slightly small or small number of detections 116. For example, a slightly small number may indicate less than half of the number of detections 116. For example, in the first detection 116a, the second detection 116b, and the third detection 116c, only one corresponding voxel indicating the detection may be found in a particular voxel of the final detection 120.

[0072] In some embodiments, productivity may be considered more important than safety. For example, as shown by the second indicator 134b, the indicator 134 can be placed slightly closer to the second end 138 of the sliding scale 132 than the first side 136. In these and other embodiments, the threshold requirement can be set to represent a number of at least most of the detections 116. For example, in some embodiments, the threshold requirement can be set to represent a number of slightly majority of the detections 116. For example, a slightly majority may indicate a number of detections 116 that is greater than half. In an instance with three detections, the threshold requirement can be set to two, where at least two of the three detections may be required to indicate detection of a particular voxel so that the final detection 120 finds detection of the particular voxel. In another example with five detections, at least three of the five detections may be required to indicate detection.

[0073] In some embodiments, as productivity is prioritized over safety, the threshold requirements may also increase. For example, instead of requiring three out of five detections to indicate a detection, four detections may be required. In some embodiments, the highest level of productivity may be sought. For example, as shown by the third indicator 134c, the indicator 134 may be placed at the end of the second end 138 of the sliding scale 132. In these and other embodiments, detections in all of the detections 116 may be required to find a detection for a particular voxel for the final detection 120. In some embodiments, the indicator 134 may be moved on the sliding scale 132 to adjust the threshold requirements based on the specific needs of the machine. In some embodiments, based on specific safety criteria, the threshold may be lowered for safety purposes—for example, if any detection is identified, the system may determine that a detection exists.

[0074] Back to Figure 1A-1B In some embodiments, after determining the detection results for each of the final voxel sets, the final detection 120 can be simplified to one or more general representations, indicating whether an object is detected in one or more sub-detection regions. For example, the final detection 120 may include one or more binary values, each binary value indicating that an object is detected in one or more sub-detection regions. One or more binary values ​​(e.g., 0 for not detected and 1 for detected) may simply indicate whether any object is detected in each sub-detection region. In these and other embodiments, one or more general representations may be determined based at least on the detection results corresponding to the final voxel set. For example, the one or more general representations may be a combined result of the detection results of the final voxel set located within the corresponding sub-detection region.

[0075] In some embodiments, the detection results of the final voxel set can be combined to include determining the number of final voxel groups with detection. For example, the number of final voxel groups indicating detection within a specific sub-detection area can be obtained. In these and other embodiments, the number can be compared with a reduced threshold to determine a general representation for a specific sub-detection area. The reduced threshold can define the number of voxels required to indicate detection to find a general representation for a specific sub-detection area. In some embodiments, the reduced threshold can be an arbitrary number of voxels suitable for the machine implementing system 100. For example, a specific machine may require an operating area without any objects or particles. In these cases, the reduced threshold can be determined to be as small as a single voxel. For example, in response to finding a detection in any one of the final voxel sets, a general representation can be determined to indicate detection. In another example, a specific machine may only be interested in detecting larger objects. In these cases, the reduced threshold may be larger.

[0076] In some embodiments, the system 100 may include a control module 122. In some embodiments, the control module 122 may be configured to obtain the final detection 120. In some embodiments, the control module 122 may be implemented using the following hardware, including one or more processors, central processing units (CPUs), graphics processing units (GPUs), data processing units (DPUs), parallel processing units (PPUs), microprocessors (e.g., for performing or controlling the performance of one or more operations), field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), accelerators (e.g., deep learning accelerators (DLAs), optical flow accelerators (OFAs)), programmable vision accelerators (including one or more direct memory address (DMA) systems and / or vector processing units (VPUs)), and / or other processor types. In some other cases, the control module 122 may be implemented using a combination of hardware and software. In the present disclosure, operations described as being performed by a corresponding module may include operations that the corresponding module may instruct the corresponding computing system to perform. In these or other embodiments, the control module 122 may be implemented by (e.g., regarding Figures 4A-4D , Figure 5 and / or Figure 6 The invention is implemented by one or more computing devices described in further detail.

[0077] In some embodiments, the control module 122 can be further configured to instruct a machine implementing the system 100 to perform or cease performing one or more operations based at least on the final detection 120. For example, in some embodiments, the control module 122 can instruct the machine to perform one or more mitigation operations in response to indicating the final detection 120 detected. In some embodiments, the one or more mitigation operations can include any operations that can be performed to maintain and / or enhance safety. For example, in some embodiments, the mitigation operation can include a notification about the detection. For example, the notification can include alerting security personnel, a warning alarm (e.g., visual and / or audio), or any other suitable type of notification.

[0078] In these and other embodiments, the mitigation operation may include causing a change in one or more operations of the machine corresponding to the sub-detection area where the detection was performed. For example, the mitigation operation may include causing a protective stop of the machine, limiting the speed of the machine, and / or operating a warning signal. The protective stop may include stopping one or more operations of the machine, such as any traveling or moving parts. In some embodiments, the control module 122 may select different mitigation operations based on the context and / or circumstances of the detection. For example, the control module 122 may select a mitigation operation based on the sub-detection area where the detection was performed.

[0079] For example, Figure 1D An example detection area 150 according to some embodiments of the present disclosure is shown. In some embodiments, the detection area 150 may be associated with a machine 152. For example, in some embodiments, the machine 152 may be a vehicle, such as the vehicle 400 in FIG. 4. In some embodiments, the detection area 150 may include one or more sub-detection areas. For example, the detection area 150 may include a first sub-detection area 154, a second sub-detection area 156, and a third sub-detection area 158. Although three sub-detection areas are shown in the figure, the detection area may include any number of suitable sub-detection areas. Additionally or alternatively, although a rectangular frame is shown in the figure, the sub-detection area may include any shape and / or format suitable for the machine 152. For example, the machine 152 may be interested in a circular-shaped area around the machine. Additionally or alternatively, one or more sub-detection areas may be similarly applied to all directions. For example, one or more sub-detection areas may include areas behind, around, above, and / or below the machine 152.

[0080] In some embodiments, the detection area 150 can be divided into one or more sub-detection areas based at least on the distance from the reference point. For example, the first sub-detection area 154 can include an area at a first distance from the reference point, the second sub-detection area 156 can include an area at a second distance from the reference point, excluding the area included in the first sub-detection area, and the third sub-detection area 158 can include an area at a third distance from the reference point, excluding the area covered by the first sub-detection area and the second sub-detection area. In some embodiments, the reference point can be located anywhere in or near the machine 152. For example, the reference point can be located at the front end of the machine 152. In another example, the reference point can be located at the center of the machine 152.

[0081] In some embodiments, one or more sub-detection areas may be determined based on the level of restriction on the entry and / or penetration of an object. For example, the first sub-detection area 154 may cover an area immediately adjacent to the machine 152 and may include the highest level of restriction. For example, the first sub-detection area 154 may include the physical operating area of ​​the machine 152. In these and other embodiments, the first sub-detection area 152 may be designated as a restricted area. The restricted area may correspond to an area where no objects should be operating at all times.

[0082] Additionally or alternatively, the second sub-detection area 156 may be designated or referred to as a protected area. In some embodiments, the protected area may include an area that contains and / or surrounds an operating area where potential dangers may exist. For example, the second sub-detection area 156 may include an area where the machine 152 may encounter an object while in motion, or may include an area immediately adjacent to a restricted area—so that any object in these areas may quickly enter the restricted area, so that protective measures can be taken in advance.

[0083] Additionally or alternatively, the third sub-detection area 158 may be designated or referred to as a warning zone. In some embodiments, a warning zone may include an area that may not affect the operation of the machine 152 but may not have appropriate preventive measures.

[0084] In some embodiments, the machine 152 may operate different mitigation operations based at least on the one or more sub-detection areas in which the one or more objects are detected. For example, detection in a first sub-detection area, a second sub-detection area, and a third sub-detection area may result in one or more different mitigation operations.

[0085] In some embodiments, one or more mitigation operations corresponding to the first sub-detection area 154 may include a protective stop and / or a limited speed of the machine 152. For example, one or more operations of the machine 152 (e.g., any traveling or moving parts) may be stopped and / or the speed of one or more operations may be reduced. For example, one or more operations may be stopped and / or reduced based at least on the position of one or more objects detected in the first sub-detection area 154. For example, in response to detecting one or more objects closer to the machine 152, one or more operations of the machine 152 may be stopped.

[0086] In some embodiments, one or more mitigation actions corresponding to second sub-detection area 156 may include a protective stop. In this and other cases, one or more objects detected in second sub-detection area 156 may not be close enough to enter first sub-detection area 154. In these cases, travel of machine 152 may be stopped to avoid collision with one or more objects.

[0087] Additionally or alternatively, in response to detecting one or more objects in the third sub-detection area 158, the mitigation operation may include operating the machine 152 at a limited speed and / or operating a warning signal. In some embodiments, the warning signal may include any visual and / or audible signal that may notify the presence of the machine 152. For example, the warning signal may include an alarm noise that may alert a worker in the third sub-detection area of ​​the presence of the machine 152.

[0088] Figure 2A-2B An example system 200 configured to determine that an object is detected in a detection area according to one or more embodiments of the present disclosure is shown. The system 200 may be Figure 1A-1B For example, system 200 may illustrate an example implementation of system 100 having a particular type of detector. Figure 1A-1B In the system 100, these detectors may be used as a depth module 110 implemented on a particular type of computing platform.

[0089] For example, the system 200 may include a semi-global matching (SGM) detector 206a implemented on an optical flow accelerator (OFA) 204a, a Bi3D detector 206b implemented on a deep learning accelerator (DLA) 204b, and an efficient semi-supervised (ESS) detector 206c implemented on a graphics processing unit (GPU) 204c. In some embodiments, the SGM detector 206a, the Bi3D detector 206b, and the ESS detector 206c may be collectively referred to as "deep detectors 206". In these and other embodiments, references to detectors may include the type of detector used (e.g., SGM, Bi3D, ESS) and / or the hardware platform on which the detector is implemented (e.g., OFA, DLA, GPU). For example, a reference to the SGM detector 206a may include a reference to the OFA 204a that implements the SGM detector 206a.

[0090] Although the depth detector 206 is shown with a specific example, the depth detector 206 may include any other suitable type of detector. For example, in some embodiments, the depth detector 206 may include one or more of SGM, Bi3D, ESS, block matching (BM), block matching with dynamic programming (BMDP), belief matching (BM), gradient feature matching (GF), directional gradient histogram (HOG), etc. Additionally or alternatively, although illustrated as being implemented on a specific hardware platform (e.g., OFA, DLA, and GPU), the depth detector 206 may be implemented on any other suitable hardware platform. For example, the depth detector 206 may be implemented using one or more of a central processing unit (CPU), a GPU, an OFA, a DLA, a programmable vision accelerator (PVA), a video input (VI) controller, an image signal processor (ISP), a video image synthesizer (VIC) engine, etc. In an embodiment, when different hardware is used for each detection, the hardware can avoid common failures. In this way, if one of the hardware elements has a problem or defect and produces incorrect results, these problems or defects will not propagate to the other hardware elements - so accurate detection can continue even if one or more detectors or detection modules are offline or not working properly.

[0091] In these and other embodiments, depth detector 206 may be configured to determine depth based at least on image data 202. In some embodiments, image data 202 may include one or more images depicting a detection area. For example, in some embodiments, image data 202 may be obtained using one or more sensors. In some embodiments, image data 202 may correspond to Figure 1A The image data 102 is shown.

[0092] In some embodiments, the depth detector 206 may obtain the image data 202 and determine the depth of one or more objects present in the image data 202. In these and other embodiments, the determined depths may be different due to differences between the one or more depth detectors 206. For example, the SGM detector 206a may be configured to determine the SGM depth 208a, the Bi3D detector 206b may determine the Bi3D depth 208b, and the ESS detector 206c may determine the ESS depth 208c. In the present disclosure, the SGM depth 208a, the Bi3D depth 208b, and the ESS depth 208c may be collectively referred to as "depth 208". In these and other embodiments, one or more depths 208 may be different from each other. For example, in some embodiments, one or more depth detectors 206 may correspond to different parameters that characterize the depth detector 206.

[0093] In some embodiments, the SGM detector 206a may be a computer vision-based detector configured to determine an SGM depth 208a based at least on the image data 202. Additionally or alternatively, the SGM depth 208a may correspond to a disparity of one or more objects from a reference point, wherein objects farther from the reference point may be at a greater depth than objects closer to the reference point.

[0094] In some embodiments, the SGM detector 206a may be configured to combine a local pixel matching cost function with a global energy minimization constraint. For example, the SGM detector 206a may receive image data 202 including a first image and a second image from a set of stereo cameras. For one or more pixels corresponding to one or more objects in the first image, the SGM detector 206a may determine the corresponding pixels in the second image. In some embodiments, the SGM detector 206a may apply a census transform to determine the corresponding pixels in the second image. For example, the SGM detector 206a may be configured to receive the image data 202 as a grayscale image represented in a brightness format. For example, the image data 202 may be represented in a 16-bit brightness format and / or an 8-bit brightness format. In these and other embodiments, the census transform may compare one or more pixels to a plurality of adjacent pixels. For example, the intensity (e.g., brightness value) of each pixel may be compared to the intensity of the adjacent pixels. The comparison result may be represented as a number that may be used to determine the corresponding pixels between the first image and the second image.

[0095] In some embodiments, the SGM detector 206a may perform regression-based classification. For example, a disparity value may be assigned to one or more pixels of the image data 202. In these and other embodiments, a reference to a pixel may include a reference to one or more objects or regions corresponding to the pixel. The disparity value may represent a distance from one or more pixels in the first image to a corresponding pixel in the second image. In these and other embodiments, the disparity value may be used to determine a distance of one or more pixels from a particular reference point, which distance represents a depth of the one or more pixels.

[0096] In some embodiments, the SGM detector 206a can cause one or more specifications to be adjusted at runtime. For example, the SGM detector 206a can adjust the number of iterations and / or the accuracy of the depth estimate. For example, the SGM detector 206a can iterate one or more times, wherein the iterations can refine the cost output of the SGM detector 206a. Additionally or alternatively, the output grid size granularity is configurable. For example, the output grid size can be configured according to the desired depth estimate quality. For example, a smaller grid size can provide a more detailed depth estimate than a larger grid size.

[0097] In some embodiments, the Bi3D detector 206b may be an ML-based detector configured to use a series of binary classifications to determine the depth of one or more objects in the image data 202. For example, rather than testing whether one or more pixels are at a particular depth (e.g., assigning a disparity value to each pixel), the Bi3D detector 206b may classify one or more pixels corresponding to one or more objects as being closer or farther than a given depth plane. As such, the Bi3D detector 206b may implement a 2D convolution technique that aggregates features within the same depth plane. In some embodiments, such as Figure 1A-1B As described in further detail in , the Bi3D detector 206b may determine depth in a binary classification format.

[0098] In some embodiments, the Bi3D detector 206b may include a flexible range. For example, the Bi3D detector 206b may be able to estimate the depth of objects within a specific range (e.g., an area immediately adjacent to a machine). In some cases, the specific range may be determined by specifying a depth plane to be used for binary classification. Additionally or alternatively, the Bi3D detector 206b may allow parameters to be manipulated during runtime. For example, one or more pixels may be classified in a manner such that they are located at or farther from a given depth plane, and iteration may be performed for multiple depth planes. In some embodiments, by iterative classification, a more refined depth quantization may be obtained.

[0099] In some embodiments, the Bi3D detector 206b may be configured to receive image data 202 represented in RGB format. For example, the image data 202 may be represented in RGB8 format, where one or more pixels present in the image data are represented by three colors (e.g., red, green, and blue), and each color is represented by an 8-bit integer. In some embodiments, the Bi3D detector 206b may be configured to receive image data in any other suitable format.

[0100] In some embodiments, the ESS detector 206 c can be configured to use an ML-based algorithm to determine the ESS depth 208 c. In some embodiments, similar to the Bi3D detector 206 b, the ESS detector 206 c can obtain the image data 202 in RGB format. In these and other embodiments, the ESS detector 206 c can be configured to apply regression-based classification to the image data 202 to determine the ESS depth 208 c based on at least the image data 202.

[0101] In these and other embodiments, while the ESS detector 206c may be similar to the Bi3D detector 206b in that both are ML-based detectors, the ESS detector 206c and the Bi3D detector 206b may include different methods of representing depth. For example, while the Bi3D detector 206b uses different depth planes to represent depth in binary classification, the ESS detector 206c may represent depth in a continuous depth format. For example, the ESS detector 206c represents depth as a distance from a certain reference point, rather than representing whether a particular pixel corresponding to an object is closer or farther than a particular depth plane. In some embodiments, the continuous depth format employed by the ESS detector 206c may be combined with the present disclosure. Figure 1A-1B Describe in more detail.

[0102] In some embodiments, the ESS detector 206c may be configured to process all depth levels simultaneously to determine the distribution probability of one or more objects in the image data 202 at all possible depth levels. For example, the ESS detector 206c may implement a 3D convolution to aggregate data across multiple depth levels. For example, the ESS detector 206c may extract features (e.g., one or more objects) from the image data 202 and arrange the extracted features into a tensor that extends across multiple depth levels. In these and other embodiments, the tensor may be constructed and processed in such a way that, for one or more pixels corresponding to the extracted features of the image data 202, a probability distribution across multiple depth levels is generated. For example, the probability distribution may indicate the probability that the one or more pixels corresponding to the extracted features are at different depth levels. In some embodiments, the depth of the extracted features may be determined by finding the maximum value in the probability distribution. For example, the one or more pixels corresponding to the extracted features may be determined to be located at the depth level with the highest probability.

[0103] In some embodiments, the depth detector 206 may be implemented on different hardware platforms. For example, one or more depth detectors 206 may be implemented on different hardware platforms. In these and other embodiments, different hardware platforms may be selected based on parameters characterizing one or more depth detectors 206. For example, the OFA 204a may be suitable for use with the SGM detector 206a because the OFA 204a may be dedicated to optical flow and / or stereo vision calculations. For example, the OFA 204a may be configured to generate optical flow and / or disparity maps from image data. The DLA 204b may be suitable for use with the Bi3D detector 206b because the DLA 204b is used to run a neural network. The GPU 204c may be suitable for use with the ESS detector 206c because the GPU 204c may be used to provide high-performance parallel computing that the ESS detector 206c may use to simultaneously determine the depth of one or more objects across multiple depth planes.

[0104] In these and other embodiments, the depth detector 206 can have different properties and / or parameters suitable for the hardware platform. In some embodiments, the different properties and / or parameters can improve the robustness of the system 200. For example, different properties and / or parameters of the depth detector 206 and / or the hardware platform can reduce the possibility of errors when detecting one or more objects. For example, the differences between multiple depth detectors 206 can allow the system 200 to include different methods of detecting one or more objects. As discussed herein according to some embodiments, the different properties and / or parameters between the depth detectors are shown and summarized in Table 1 below.

[0105]

[0106] Table 1

[0107] In some embodiments, the system 200 may include a first detection module 212a, a second detection module 212b, and a third detection module 212c (collectively referred to as "detection modules 212"), which may correspond to Figure 1A 1. In some embodiments, the detection module 212 may be configured to obtain the depth 208 to determine one or more detections. For example, the first detection module 212a may be configured to determine an SGM detection 214a based on at least the SGM depth 208a, the second detection module 212b may be configured to determine a Bi3D detection 214b based on at least the Bi3D depth 208b, and the third detection module 212c may be configured to determine an ESS detection 214c based on at least the ESS depth 208c. In some embodiments, the SGM detection 214a, the Bi3D detection 214b, and the ESS detection 214c may be collectively referred to as "detections 214".

[0108] In these and other embodiments, one or more detections 214 may indicate whether detection of an object is performed within the detection area. Additionally or alternatively, one or more detection modules 212 may be configured to determine whether one or more objects detected within the detection area are located within certain regions of interest within the detection area. In some embodiments, the determination of one or more detections 214 may include a determination of whether the detection 116 is within a region of interest. Figure 1B One or more operations described.

[0109] In some embodiments, the detection module 212 may be implemented on one or more CPUs. For example, the first detection module 212a may be implemented on the first CPU 210a, the second detection module 212b may be implemented on the second CPU, and the third detection module 212b may be implemented on the third CPU 210c. In some embodiments, the first CPU 210a, the second CPU 210b, and the third CPU 210c may refer to the same CPU, for example, a single CPU may implement all the detection modules 212. In some embodiments, one or more of the first CPU 210a, the second CPU 210b, and the third CPU 210c may be different from each other.

[0110] In some embodiments, the system 200 may include an aggregation module 218 implemented on a fourth CPU 216. In some embodiments, the fourth CPU 216 may refer to the same CPU as one or more of the first CPU 210a, the second CPU 210b, and the third CPU 210c. In some embodiments, the fourth CPU 216 may be different from one or more other CPUs. In some embodiments, the aggregation module 218 may be configured to obtain the SGM detection 214a, the Bi3D detection 214b, and the ESS detection 214c. The aggregation module 218 may also be configured to determine the final detection 220 based on at least the SGM detection 214a, the Bi3D detection 214b, and / or the ESS detection 214c.

[0111] In some embodiments, the final detection 220 may indicate whether detection of one or more objects is performed in the detection area. For example, the final detection 220 may indicate whether the one or more objects are in the image data 202. Additionally or alternatively, the final detection 220 may indicate the location where the one or more objects are detected in the detection area. In some embodiments, the final detection 220 may be the result of combining the SGM detection 214a, the Bi3D detection 214b, and the ESS detection 214c. By combining two or more detections in the detection 214, the final detection 220 may more accurately indicate that one or more objects are detected in the detection area than using only one detection in the detection 214. In some embodiments, the aggregation module 218 may determine the final detection 220 by aggregating the detections 214. In some embodiments, one or more detections in the aggregated detection 214 may include information about the aggregated detection 116. Figure 1B In some embodiments, for example, regarding Figure 1B As described, one or more control operations may be performed based on the final detection 220 .

[0112] Figure 3 is a flow chart illustrating a method 300 for detecting an object in a detection area of ​​a machine according to one or more embodiments of the present disclosure. In some embodiments, one or more operations of the method 300 may be directed to Figure 1A-1B One or more operations of method 300 may be performed by any suitable system, device, or apparatus, for example, system 100, Figures 4A-4D Autonomous vehicle systems described, Figure 5 Computing devices described and / or related Figure 6 Describe the data system.

[0113] Method 300 may include one or more blocks. Although shown as discrete blocks, operations associated with one or more blocks of method 300 may be divided into additional blocks, combined into fewer blocks, or eliminated, depending on the specific implementation.

[0114] In some embodiments, method 300 may include box 302. At box 302, multiple analysis techniques may be used to analyze sensor data corresponding to the operating area of ​​the machine. In some embodiments, multiple analysis techniques may be used to detect objects in the operating area. In these and other embodiments, the multiple analysis techniques may include one or more depth techniques that may differ in how the depth from the reference point is determined. For example, the one or more depth techniques may obtain sensor data and determine the depth of one or more objects present in the sensor data. In some embodiments, the sensor data may include stereo image data. For example, a stereo camera may be used to obtain sensor data corresponding to the operating area of ​​the machine.

[0115] In some embodiments, at least two of the plurality of analysis techniques may have computational diversity. For example, at least two of the analysis techniques may be configured to perform different types of computational analysis on the sensor data. In some embodiments, computational diversity between the plurality of analysis techniques may provide robust detection of the object. For example, a single analysis technique may include certain computational errors and / or limitations. By implementing different analysis techniques, certain computational errors and / or limitations may be reduced. In some embodiments, for example, with respect to Figure 1A-1B , examples of various analysis techniques may be further discussed and illustrated in this disclosure.

[0116] In some embodiments, at least two of the plurality of techniques may have implementation diversity. For example, at least two of the analysis techniques may be implemented on different computing platforms. For example, a particular analysis technique may be implemented on a particular computing hardware that is best suited for that particular analysis technique. Figures 1A-1C , examples of different computing platforms and corresponding analysis techniques may be further discussed and illustrated in this disclosure.

[0117] At box 304, detection of objects within the operating area can be determined based at least on the analysis. For example, based at least on the results of multiple analysis techniques, it can be determined whether one or more objects are present in the sensor data. For example, individual detection results corresponding to each of the multiple analysis techniques can be determined. In these and other cases, an overall detection result can be determined based at least on the individual detection results. In some embodiments, the determination of the overall detection result can include combining the individual detection results. In some cases, the individual detection results can be combined based on one or more combining algorithms to determine an overall detection result that represents the presence and / or location of one or more objects within the machine's operating area. In some embodiments, the determination of the overall detection result can include one or more operations corresponding to determining a final detection 120, see Figures 1A-1C Described.

[0118] At block 306, one or more operations of the machine may be controlled based at least on the detection of the object. For example, in response to the detection of the object, one or more mitigation operations may be performed by the machine. In some embodiments, the one or more mitigation operations may vary based on the location of the object detection relative to the machine. For example, the one or more mitigation operations may vary based at least on the distance of the object from the machine. In some embodiments, for example, reference Figure 1A-1B and Figure 1D , examples of varying mitigation operations may be further discussed and illustrated in this disclosure.

[0119] Without departing from the scope of the present disclosure, method 300 may be modified, added to, or omitted. For example, the operations of method 300 may be implemented in different orders. Additionally or alternatively, two or more operations may be performed simultaneously. In addition, the operations and actions outlined are provided only as examples, and some of these operations and actions may be optional, combined into fewer operations and actions, or expanded into additional operations and actions, without departing from the essence of the described embodiments.

[0120] Example Autonomous Vehicle

[0121] Figure 4A4 is a diagram of an example autonomous vehicle 400 according to some embodiments of the present disclosure. Autonomous vehicle 400 (alternatively, referred to herein as "vehicle 400") may include, but is not limited to, a passenger vehicle, such as a car, a truck, a bus, an emergency vehicle, a shuttle, an electric or motorized bicycle, a motorcycle, a fire truck, a police car, an ambulance, a boat, an engineering vehicle, an underwater vessel, a drone, and / or other types of vehicles (e.g., unmanned and / or capable of accommodating one or more passengers). Autonomous vehicles are generally described according to the automation levels defined by the National Highway Traffic Safety Administration (NHTSA), a division of the U.S. Department of Transportation, and the Society of Automotive Engineers (SAE) "Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles" (Standard No. J3016-201506, issued on June 15, 2018, Standard No. J3016-201609, issued on September 30, 2016, and previous and future versions of the standard). The vehicle 400 is capable of implementing functions that comply with one or more of levels 3-5 of autonomous driving levels. The vehicle 400 is capable of implementing functions that comply with one or more of levels 1-5 of autonomous driving levels. For example, depending on the embodiment, the vehicle 400 is capable of driver assistance (level 1), partial automation (level 2), conditional automation (level 3), high automation (level 4) and / or full automation (level 5). The term "autonomy" as used herein may include any and / or all types of autonomy of 400 or other machines, such as full autonomy, high autonomy, conditional autonomy, partial autonomy, providing assisted autonomy, semi-autonomy, main autonomy or other names.

[0122] Vehicle 400 may include components such as a chassis, a body, wheels (e.g., 2, 4, 6, 8, 18, etc.), tires, axles, and other components of the vehicle. Vehicle 400 may include a propulsion system 450, such as an internal combustion engine, a hybrid power plant, an all-electric engine, and / or another propulsion system type. Propulsion system 450 may be connected to a drive train of vehicle 400, which may include a transmission, to achieve propulsion of vehicle 400. Propulsion system 450 may be controlled in response to receiving a signal from throttle / accelerator 452.

[0123] A steering system 454, which may include a steering wheel, may be used to steer vehicle 400 (e.g., along a desired path or route) when propulsion system 450 is operating (e.g., when the vehicle is in motion). Steering system 454 may receive signals from steering actuator 456. For fully automated (Level 5) functionality, a steering wheel may be optional.

[0124] Brake sensor system 446 may be used to operate vehicle brakes in response to receiving signals from brake actuator 448 and / or brake sensors.

[0125] May include one or more CPUs, system on chip (SoC) 404 ( Figure 4C ) and / or one or more controllers 436 of one or more GPUs may provide signals (e.g., representing commands) to one or more components and / or systems of the vehicle 400. For example, the one or more controllers may send signals to operate vehicle brakes via one or more brake actuators 448, to operate a steering system 454 via one or more steering actuators 456, and / or to operate a propulsion system 450 via one or more throttles / accelerators 452. The one or more controllers 436 may include one or more onboard (e.g., integrated) computing devices (e.g., supercomputers) that process sensor signals and output operating commands (e.g., signals representing commands) to enable autonomous driving and / or assist a human driver in driving the vehicle 400. The one or more controllers 436 may include a first controller 436 for autonomous driving functions, a second controller 436 for functional safety functions, a third controller 436 for artificial intelligence functions (e.g., computer vision), a fourth controller 436 for infotainment functions, a fifth controller 436 for redundancy in emergency situations, and / or other controllers. In some examples, a single controller 436 may handle two or more of the above functions, two or more controllers 436 may handle a single function, and / or any combination thereof.

[0126] The one or more controllers 436 may provide signals for controlling one or more components and / or systems of the vehicle 400 in response to sensor data (e.g., sensor input) received from one or more sensors. The sensor data may be received from, for example and without limitation, a global navigation satellite system sensor 458 (e.g., a global positioning system sensor), a RADAR sensor 460, an ultrasonic sensor 462, a LIDAR sensor 464, an inertial measurement unit (IMU) sensor 466 (e.g., an accelerometer, a gyroscope, a magnetic compass, a magnetometer, etc.), a microphone 796, a stereo camera 468, a wide angle camera 470 (e.g., a fisheye camera), an infrared camera 472, a surround camera 474 (e.g., a 360 degree camera), a long-range and / or mid-range camera 498, a speed sensor 444 (e.g., for measuring the velocity of the vehicle 400), a vibration sensor 442, a steering sensor 440, a brake sensor (e.g., as part of a brake sensor system 446), and / or other sensor types.

[0127] One or more of the controllers 436 may receive inputs (e.g., represented by input data) from the instrument cluster 432 of the vehicle 400 and provide outputs (e.g., represented by output data, display data, etc.) via a human machine interface (HMI) display 434, an audible annunciator, a speaker, and / or via other components of the vehicle 400. These outputs may include information such as vehicle speed, velocity, time, map data (e.g., Figure 4C The HMI display 434 may include information such as the HD map 422 of the vehicle 400, location data (e.g., the location of the vehicle 400 on the map), directions, locations of other vehicles (e.g., occupancy grid), information about objects and states of objects as sensed by the controller 436, etc. For example, the HMI display 434 may display information about the presence of one or more objects (e.g., street signs, warning signs, traffic light changes, etc.) and / or information about driving maneuvers that the vehicle has made, is making, or will make (e.g., changing lanes now, leaving 34B in two miles, etc.).

[0128] The vehicle 400 further includes a network interface 424 that can communicate via one or more networks using one or more wireless antennas 426 and / or a modem. For example, the network interface 424 may be capable of communicating via LTE, WCDMA, UMTS, GSM, CDMA2000, etc. The one or more wireless antennas 426 may also enable communication between objects in the environment (e.g., vehicles, mobile devices, etc.) using one or more local area networks such as Bluetooth, Bluetooth LE, Z-wave, ZigBee, etc. and / or one or more low power wide area networks (LPWANs) such as LoRaWAN, SigFox, etc.

[0129] Figure 4B According to some embodiments of the present disclosure, Figure 4A An example of camera positions and fields of view for an example autonomous vehicle 400 is shown. The cameras and respective fields of view are an example embodiment and are not intended to be limiting. For example, additional and / or replaceable cameras may be included, and / or the cameras may be located at different locations on the vehicle 400.

[0130] The camera type for the camera may include, but is not limited to, a digital camera that may be suitable for use with components and / or systems of the vehicle 400. The camera may operate at an automotive safety integrity level (ASIL) B and / or at another ASIL. The camera type may have any image capture rate, such as 60 frames per second (fps), 120fps, 240fps, etc., depending on the embodiment. The camera may be capable of using a rolling shutter, a global shutter, another type of shutter, or a combination thereof. In some examples, the color filter array may include a red-white-white-white (RCCC) color filter array, a red-white-white-blue (RCCB) color filter array, a red-blue-green-white (RBGC) color filter array, a Foveon X3 color filter array, a Bayer sensor (RGGB) color filter array, a monochrome sensor color filter array, and / or another type of color filter array. In some embodiments, a clear pixel camera such as a camera with an RCCC, RCCB, and / or RBGC color filter array may be used in an effort to improve light sensitivity.

[0131] In some examples, one or more of the cameras can be used to perform advanced driver assistance system (ADAS) functions (e.g., as part of a redundant or fail-safe design). For example, a multi-function monocular camera can be installed to provide functions including lane departure warning, traffic sign assistance, and intelligent headlight control. One or more of the cameras (e.g., all of the cameras) can simultaneously record and provide image data (e.g., video).

[0132] One or more of the cameras may be mounted in a mounting assembly such as a custom designed (3-D printed) assembly to cut off stray light and reflections from within the car (e.g., reflections from the dashboard reflected in the windshield mirror) that may interfere with the camera's image data capture capabilities. With respect to the wing mirror mounting assembly, the wing mirror assembly may be custom 3-D printed so that the camera mounting plate matches the shape of the wing mirror. In some examples, one or more cameras may be integrated into the wing mirror. For side view cameras, one or more cameras may also be integrated into the four pillars at each corner of the cab.

[0133] A camera (e.g., a front-facing camera) having a field of view that includes a portion of the environment in front of the vehicle 400 can be used for surround view to help identify the forward path and obstacles, as well as assist in providing information critical to generating an occupancy grid and / or determining a preferred vehicle path with the help of one or more controllers 436 and / or control SoCs. The front-facing camera can be used to perform many of the same ADAS functions as LIDAR, including emergency braking, pedestrian detection, and collision avoidance. The front-facing camera can also be used for ADAS functions and systems, including lane departure warning (LDW), autonomous cruise control (ACC), and / or other functions such as traffic sign recognition.

[0134] A variety of cameras may be used in the front-facing configuration, including, for example, a monocular camera platform including a CMOS (complementary metal oxide semiconductor) color imager. Another example may be a wide-angle camera 470, which may be used to sense objects (e.g., pedestrians, intersection traffic, or bicycles) entering the field of view from the periphery. Figure 4B Only one wide-angle camera is shown in the figure, but there may be any number of wide-angle cameras 470 on the vehicle 400. In addition, long-range cameras 498 (e.g., a long-view stereo camera pair) may be used for depth-based object detection, especially for objects for which a neural network has not been trained. Long-range cameras 498 may also be used for object detection and classification and basic object tracking.

[0135] One or more stereo cameras 468 may also be included in the front configuration. The stereo camera 468 may include an integrated control unit including an expandable processing unit that may provide a multi-core microprocessor and programmable logic (FPGA) with an integrated CAN or Ethernet interface on a single chip. Such a unit may be used to generate a 3-D map of the vehicle environment, including distance estimates for all points in the image. An alternative stereo camera 468 may include a compact stereo vision sensor that may include two camera lenses (one on each side) and an image processing chip that may measure the distance from the vehicle to the target object and use the generated information (e.g., metadata) to activate autonomous emergency braking and lane departure warning functions. In addition to those described herein or alternatively, other types of stereo cameras 468 may be used.

[0136] A camera (e.g., a side view camera) having a field of view of a portion of the environment including the sides of the vehicle 400 can be used for surround viewing, providing information used to create and update the occupancy grid and generate side impact collision warnings. Figure 4B The four surround cameras 474 shown in FIG. 4 can be placed on the vehicle 400. The surround cameras 474 can include wide-angle cameras 470, fisheye cameras, 360-degree cameras, and / or the like. For example, four fisheye cameras can be placed in front, back, and sides of the vehicle. In an alternative arrangement, the vehicle can use three surround cameras 474 (e.g., left, right, and back), and can utilize one or more other cameras (e.g., a forward-facing camera) as a fourth surround camera.

[0137] A camera having a field of view that includes a portion of the environment behind the vehicle 400 (e.g., a rear view camera) can be used to assist with parking, surround view, rear collision warning, and creating and updating an occupancy grid. A variety of cameras can be used, including but not limited to cameras that are also suitable as front-facing cameras as described herein (e.g., long-range and / or mid-range cameras 498, stereo cameras 468, infrared cameras 472, etc.).

[0138] Figure 4C According to some embodiments of the present disclosure, Figure 4A Block diagram of an example system architecture of an example autonomous vehicle 400. It should be understood that this arrangement and other arrangements described herein are set forth merely as examples. Other arrangements and elements (e.g., machines, interfaces, functions, sequences, functional groupings, etc.) may be used in addition to or in place of those shown, and some elements may be omitted entirely. Further, many of the elements described herein are functional entities that may be implemented as discrete or distributed components or in combination with other components, and in any appropriate combination and location. The various functions described herein as being performed by an entity may be implemented by hardware, firmware, and / or software. For example, the various functions may be implemented by a processor executing instructions stored in a memory.

[0139] Figure 4C Each of the components, features, and systems of vehicle 400 is illustrated as being connected via bus 402. Bus 402 may include a controller area network (CAN) data interface (alternatively, referred to herein as a "CAN bus"). CAN may be a network inside vehicle 400 that assists in controlling various features and functions of vehicle 400, such as actuation of brakes, acceleration, braking, steering, windshield wipers, and the like. The CAN bus may be configured to have tens or even hundreds of nodes, each with its own unique identifier (e.g., CAN ID). The CAN bus may be read to find steering wheel angle, ground speed, engine speed per minute (RPM), button position, and / or other vehicle status indicators. The CAN bus may be ASIL B compliant.

[0140] Although bus 402 is described as a CAN bus here, this is not intended to be limiting. For example, in addition to or alternatively to the CAN bus, FlexRay and / or Ethernet can be used. In addition, although bus 402 is represented by a single line, this is not intended to be limiting. For example, there can be any number of buses 402, which may include one or more CAN buses, one or more FlexRay buses, one or more Ethernet buses, and / or one or more other types of buses using different protocols. In some examples, two or more buses 402 can be used to perform different functions, and / or can be used for redundancy. For example, a first bus 402 can be used for a collision avoidance function, and a second bus 402 can be used for drive control. In any example, each bus 402 can communicate with any component of the vehicle 400, and two or more buses 402 can communicate with the same component. In some examples, each SoC 404, each controller 436, and / or each computer in the vehicle can have access to the same input data (e.g., input from a sensor of the vehicle 400), and can be connected to a common bus such as a CAN bus.

[0141] The vehicle 400 may include one or more controllers 436, such as those described herein. Figure 4A Those controllers described. Controller 436 can be used for a variety of functions. Controller 436 can be coupled to any other different components and systems of vehicle 400, and can be used for control of vehicle 400, artificial intelligence of vehicle 400, infotainment for vehicle 400, and / or the like.

[0142] The vehicle 400 may include one or more system on chip (SoC) 404. The SoC 404 may include a CPU 406, a GPU 408, a processor 410, a cache 412, an accelerator 414, a data store 416, and / or other components and features not shown. The SoC 404 may be used to control the vehicle 400 in a variety of platforms and systems. For example, one or more SoCs 404 may be combined with an HD map 422 in a system (e.g., a system of the vehicle 400), and the HD map may be downloaded from one or more servers (e.g., a server) via a network interface 424. Figure 4D one or more servers 478) to obtain map refreshes and / or updates.

[0143] CPU 406 may include a CPU cluster or CPU complex (alternatively, referred to herein as "CCPLEX"). CPU 406 may include multiple cores and / or L2 caches. For example, in some embodiments, CPU 406 may include eight cores in a coherent multiprocessor configuration. In some embodiments, CPU 406 may include four dual-core clusters, each of which has a dedicated L2 cache (e.g., 2MB L2 cache). CPU 406 (e.g., CCPLEX) may be configured to support simultaneous cluster operations, so that any combination of clusters of CPU 406 can be active at any given time.

[0144] CPU 406 may implement power management capabilities including one or more of the following features: each hardware block may be automatically clock gated when idle to save dynamic power; each core clock may be gated when the core is not actively executing instructions due to the execution of WFI / WFE instructions; each core may be independently power gated; each core cluster may be independently clock gated when all cores are clock gated or power gated; and / or each core cluster may be independently power gated when all cores are power gated. CPU 406 may further implement an enhanced algorithm for managing power states, in which allowed power states and expected wake-up times are specified, and hardware / microcode determines the optimal power state to enter for the core, cluster, and CCPLEX. The processing core may support a simplified power state entry sequence in software, with the work being offloaded to the microcode.

[0145] GPU 408 may include an integrated GPU (alternatively, referred to herein as an "iGPU"). GPU 408 may be programmable and efficient for parallel workloads. In some examples, GPU 408 may use an enhanced tensor instruction set. GPU 408 may include one or more streaming microprocessors, each of which may include an L1 cache (e.g., an L1 cache with at least 96KB storage capacity), and two or more of these streaming microprocessors may share an L2 cache (e.g., an L2 cache with 512KB storage capacity). In some embodiments, GPU 408 may include at least eight streaming microprocessors. GPU 408 may use a computing application programming interface (API). In addition, GPU 408 may use one or more parallel computing platforms and / or programming models (e.g., NVIDIA's CUDA).

[0146] In the case of automotive and embedded use, GPU 408 can be power optimized to achieve optimal performance. For example, GPU 408 can be manufactured on fin field effect transistors (FinFETs). However, this is not intended to be limiting, and GPU 408 can be manufactured using other semiconductor manufacturing processes. Each streaming microprocessor can merge several mixed precision processing cores divided into multiple blocks. For example and without limitation, 64 PF32 cores and 32 PF64 cores can be divided into four processing blocks. In such an example, each processing block can be allocated 16 FP32 cores, 8 FP64 cores, 16 INT32 cores, two mixed precision NVIDIA tensor cores for deep learning matrix arithmetic, L0 instruction cache, thread beam (warp) scheduler, dispatch unit and / or 64KB register file. In addition, the streaming microprocessor can include independent parallel integer and floating point data paths to provide efficient execution of workloads using a mix of calculations and addressing calculations. The streaming microprocessor may include independent thread scheduling capabilities to allow for finer-grained synchronization and collaboration between parallel threads. A streaming microprocessor may include a combined L1 data cache and shared memory unit to increase performance while simplifying programming.

[0147] GPU 408 can include high bandwidth memory (HBM) and / or a 16GB HBM2 memory subsystem that provides a peak memory bandwidth of approximately 900 GB / s in some examples. In some examples, synchronous graphics random access memory (SGRAM), such as fifth generation graphics double data rate synchronous random access memory (GDDR5), can be used in addition to or in lieu of HBM memory.

[0148] The GPU 408 may include unified memory technology that includes access counters to allow memory pages to be more accurately migrated to the processor that accesses them most frequently, thereby improving the efficiency of memory ranges shared between processors. In some examples, address translation service (ATS) support can be used to allow the GPU 408 to directly access the CPU 406 page table. In such an example, when the GPU 408 memory management unit (MMU) experiences a miss, an address translation request can be transmitted to the CPU 406. In response, the CPU 406 can look up the virtual-physical mapping for the address in its page table and transmit the translation back to the GPU 408. In this way, the unified memory technology can allow a single unified virtual address space to be used for the memory of both the CPU 406 and the GPU 408, thereby simplifying GPU 408 programming and porting applications to the GPU 408.

[0149] In addition, GPU 408 may include access counters that can track how often GPU 408 accesses the memory of other processors. Access counters can help ensure that memory pages are moved to the physical memory of the processor that accesses them most frequently.

[0150] SoC 404 may include any number of caches 412, including those described herein. For example, cache 412 may include an L3 cache available to both CPU 406 and GPU 408 (e.g., connected to both CPU 406 and GPU 408). Cache 412 may include a write-back cache that may track the state of a line, for example, by using a cache coherence protocol (e.g., MEI, MESI, MSI, etc.). Depending on the embodiment, the L3 cache may include 4MB or more, although smaller cache sizes may also be used.

[0151] SoC 404 may include one or more arithmetic logic units (ALUs) that may be used to perform processing for any of a variety of tasks or operations with respect to vehicle 400, such as processing a DNN. Additionally, SoC 404 may include a floating point unit (FPU) or other math coprocessor or digital coprocessor type for performing mathematical operations within the system. For example, SoC 404 may include one or more FPUs integrated as execution units within CPU 406 and / or GPU 408.

[0152] SoC 404 may include one or more accelerators 414 (e.g., hardware accelerators, software accelerators, or combinations thereof). For example, SoC 404 may include a hardware acceleration cluster, which may include optimized hardware accelerators and / or large on-chip memory. The large on-chip memory (e.g., 4MB SRAM) may enable the hardware acceleration cluster to accelerate neural networks and other calculations. The hardware acceleration cluster may be used to supplement GPU 408 and offload some tasks of GPU 408 (e.g., free up more cycles of GPU 408 for performing other tasks). As an example, accelerator 414 may be used for targeted workloads (e.g., perception, convolutional neural networks (CNNs), etc.) that are sufficiently stable to be easily controlled for acceleration. When used herein, the term "CNN" may include all types of CNNs, including region-based or regional convolutional neural networks (RCNNs) and fast RCNNs (e.g., for object detection).

[0153] Accelerator 414 (e.g., hardware acceleration cluster) may include a deep learning accelerator (DLA). DLA may include one or more tensor processing units (TPUs) that may be configured to provide an additional 10 trillion operations per second for deep learning applications and reasoning. TPU may be an accelerator configured to perform image processing functions (e.g., for CNN, RCNN, etc.) and optimized for performing image processing functions. DLA may be further optimized for a specific set of neural network types and floating point operations and reasoning. The design of DLA may provide higher performance per millimeter than a general-purpose GPU, and far exceeds the performance of a CPU. TPU may perform several functions, including a single instance convolution function, support, for example, INT8, INT16, and FP16 data types for both features and weights, and post-processor functions.

[0154] The DLA can quickly and efficiently execute neural networks, particularly CNNs, on processed or unprocessed data for any of a wide variety of functions, such as, but not limited to: a CNN for object recognition and detection using data from a camera sensor; a CNN for distance estimation using data from a camera sensor; a CNN for emergency vehicle detection and identification and detection using data from a microphone; a CNN for facial recognition and vehicle owner identification using data from a camera sensor; and / or a CNN for safety and / or security-related events.

[0155] The DLA can perform any function of the GPU 408, and by using an inference accelerator, for example, the designer can target any function to either the DLA or the GPU 408. For example, the designer can focus the processing of CNNs and floating point operations on the DLA, and leave other functions to the GPU 408 and / or other accelerators 414.

[0156] The accelerator 414 (e.g., a hardware acceleration cluster) may include a programmable vision accelerator (PVA), which may be alternatively referred to herein as a computer vision accelerator. The PVA may be designed and configured to accelerate computer vision algorithms for advanced driver assistance systems (ADAS), autonomous driving, and / or augmented reality (AR) and / or virtual reality (VR) applications. The PVA may provide a balance between performance and flexibility. For example, each PVA may include, for example and without limitation, any number of reduced instruction set computer (RISC) cores, direct memory access (DMA), and / or any number of vector processors.

[0157] The RISC core can interact with an image sensor (e.g., an image sensor of any camera described herein), an image signal processor, and / or the like. Each of these RISC cores can include any amount of memory. Depending on the embodiment, the RISC core can use any of a number of protocols. In some examples, the RISC core can execute a real-time operating system (RTOS). The RISC core can be implemented using one or more integrated circuit devices, application specific integrated circuits (ASICs), and / or storage devices. For example, the RISC core can include an instruction cache and / or a tightly coupled RAM.

[0158] The DMA may enable components of the PVA to access system memory independently of the CPU 406. The DMA may support any number of features used to provide optimizations for the PVA, including, but not limited to, support for multi-dimensional addressing and / or circular addressing. In some examples, the DMA may support addressing in up to six or more dimensions, which may include block width, block height, block depth, horizontal block stepping, vertical block stepping, and / or depth stepping.

[0159] The vector processor can be a programmable processor that can be designed to efficiently and flexibly perform programming for computer vision algorithms and provide signal processing capabilities. In some examples, the PVA can include a PVA core and two vector processing subsystem partitions. The PVA core can include a processor subsystem, one or more DMA engines (e.g., two DMA engines), and / or other peripherals. The vector processing subsystem can operate as the main processing engine of the PVA, and can include a vector processing unit (VPU), an instruction cache, and / or a vector memory (e.g., VMEM). The VPU core can include a digital signal processor, such as, for example, a single instruction multiple data (SIMD), a very long instruction word (VLIW) digital signal processor. The combination of SIMD and VLIW can enhance throughput and rate.

[0160] Each of the vector processors may include an instruction cache and may be coupled to a dedicated memory. As a result, in some examples, each of the vector processors may be configured to execute independently of the other vector processors. In other examples, the vector processors included in a particular PVA may be configured to employ data parallelism. For example, in some embodiments, multiple vector processors included in a single PVA may execute the same computer vision algorithm, but on different regions of an image. In other examples, the vector processors included in a particular PVA may execute different computer vision algorithms simultaneously on the same image, or even execute different algorithms on sequence images or portions of an image. Among other things, any number of PVAs may be included in a hardware acceleration cluster, and any number of vector processors may be included in each of these PVAs. In addition, the PVA may include additional error correction code (ECC) memory to enhance overall system security.

[0161] The accelerator 414 (e.g., a hardware acceleration cluster) may include an on-chip computer vision network and SRAM to provide high bandwidth, low latency SRAM for the accelerator 414. In some examples, the on-chip memory may include at least 4MB of SRAM consisting of, for example and not limited to, eight field-configurable memory blocks, which may be accessed by both the PVA and the DLA. Each pair of memory blocks may include an advanced peripheral bus (APB) interface, a configuration circuit system, a controller, and a multiplexer. Any type of memory may be used. The PVA and DLA may access the memory via a backbone that provides high-speed memory access to the PVA and DLA. The backbone may include an on-chip computer vision network that interconnects the PVA and DLA to the memory (e.g., using APB).

[0162] The on-chip computer vision network may include an interface that determines that both the PVA and DLA provide ready and valid signals before transmitting any control signals / addresses / data. Such an interface may provide separate phases and separate channels for transmitting control signals / addresses / data, as well as burst communication for continuous data transmission. This type of interface may comply with ISO 26262 or IEC 61508 standards, but other standards and protocols may also be used.

[0163] In some examples, SoC 404 may include a real-time ray tracing hardware accelerator such as described in U.S. Patent Application No. 16 / 101,232 filed on August 10, 2018. The real-time ray tracing hardware accelerator may be used to quickly and efficiently determine the position and range of objects (e.g., within a world model) in order to generate real-time visualization simulations for RADAR signal interpretation, for sound propagation synthesis and / or analysis, for SONAR system simulations, for general wave propagation simulations, for comparison with LIDAR data for the purpose of positioning and / or other functions, and / or for other purposes. In some embodiments, one or more tree traversal units (TTUs) may be used to perform one or more ray tracing related operations.

[0164] The accelerator 414 (e.g., a hardware accelerator cluster) has a wide range of uses in autonomous driving. The PVA can be a programmable visual accelerator that can be used in key processing stages in ADAS and autonomous vehicles. The capabilities of the PVA are a good match for algorithmic domains that require predictable processing, low power, and low latency. In other words, the PVA performs well on semi-dense or dense rule computations, and even on small data sets that require predictable runtimes with low latency and low power. Therefore, in the context of a platform for autonomous vehicles, the PVA is designed to run classic computer vision algorithms because they are efficient at object detection and integer math.

[0165] For example, according to one embodiment of the technology, PVA is used to perform computer stereo vision. In some examples, semi-global matching based algorithms can be used, but this is not intended to be limiting. Many applications for Level 3-5 autonomous driving require instant motion estimation / stereo matching (e.g., structure from motion, pedestrian recognition, lane detection, etc.). PVA can perform computer stereo vision functions on inputs from two monocular cameras.

[0166] In some examples, PVA can be used to perform dense optical flow. According to processing raw RADAR data (e.g., using 4D fast Fourier transform) to provide processed RADAR. In other examples, PVA is used for time-of-flight depth processing, which is for example by processing raw time-of-flight data to provide processed time-of-flight data.

[0167] DLA can be used to run any type of network to enhance control and driving safety, including, for example, a neural network that outputs a confidence measure for each object detection. Such a confidence value can be interpreted as a probability, or as providing a relative "weight" of each detection compared to other detections. This confidence value enables the system to make further decisions about which detections should be considered true positive detections rather than false positive detections. For example, the system can set a threshold for confidence and only consider detections that exceed the threshold as true positive detections. In an automatic emergency braking (AEB) system, a false positive detection will cause the vehicle to automatically perform emergency braking, which is obviously undesirable. Therefore, only the most confident detection should be considered a trigger for AEB. DLA can run a neural network for regressing confidence values. The neural network can take at least some subset of parameters as its input, such as bounding box dimensions, ground plane estimates obtained (e.g., from another subsystem), inertial measurement unit (IMU) sensor 466 output related to vehicle 400 orientation and distance, 3D position estimates of objects obtained from neural networks and / or other sensors (e.g., LIDAR sensor 464 or RADAR sensor 460), etc.

[0168] SoC 404 may include one or more data stores 416 (e.g., memory). Data store 416 may be on-chip memory of SoC 404 that may store neural networks to be executed on the GPU and / or DLA. In some examples, data store 416 may be large enough to store multiple instances of a neural network for redundancy and safety. Data store 412 may include an L2 or L3 cache 412. References to data store 416 may include references to memory associated with a PVA, DLA, and / or other accelerator 414 as described herein.

[0169] SoC 404 may include one or more processors 410 (e.g., embedded processors). Processor 410 may include a boot and power management processor, which may be a dedicated processor and subsystem for handling boot power and management functions and related safety implementations. The boot and power management processor may be part of the SoC 404 boot sequence and may provide runtime power management services. The boot power and management processor may provide clock and voltage programming, auxiliary system low power state transitions, SoC 404 thermal and temperature sensor management, and / or SoC 404 power state management. Each temperature sensor may be implemented as a ring oscillator whose output frequency is proportional to the temperature, and SoC404 may use the ring oscillator to detect the temperature of CPU 406, GPU 408, and / or accelerator 414. If it is determined that the temperature exceeds a threshold, the boot and power management processor may enter a temperature fault routine and place SoC 404 in a lower power state and / or place vehicle 400 in a driver safety parking mode (e.g., parking vehicle 400 safely).

[0170] Processor 410 may further include a set of embedded processors that may be used as an audio processing engine. The audio processing engine may be an audio subsystem that allows for full hardware support for multi-channel audio through multiple interfaces and a wide range of flexible audio I / O interfaces. In some examples, the audio processing engine is a dedicated processor core having a digital signal processor with dedicated RAM.

[0171] The processor 410 may further include an always-on processor engine that may provide the necessary hardware features to support low-power sensor management and wake-up use cases. The always-on processor engine may include a processor core, tightly coupled RAM, supporting peripherals (e.g., timers and interrupt controllers), various I / O controller peripherals, and routing logic.

[0172] The processor 410 may further include a safety cluster engine, which includes a dedicated processor subsystem that handles safety management of automotive applications. The safety cluster engine may include two or more processor cores, tightly coupled RAM, supporting peripherals (e.g., timers, interrupt controllers, etc.), and / or routing logic. In safety mode, the two or more cores may operate in lockstep mode and act as a single core with comparison logic to detect any differences between their operations.

[0173] Processor 410 may further include a real-time camera engine, which may include a dedicated processor subsystem for handling real-time camera management.

[0174] Processor 410 may further include a high dynamic range signal processor, which may include an image signal processor, which is a hardware engine that is part of the camera processing pipeline.

[0175] The processor 410 may include a video image compositer, which may be a processing block (e.g., implemented on a microprocessor) that implements video post-processing functions required for the video playback application to produce a final image for the player window. The video image compositer may perform lens distortion correction for the wide-angle camera 470, the surround camera 474, and / or for an in-cab surveillance camera sensor. The in-cab surveillance camera sensor is preferably monitored by a neural network running on another instance of the advanced SoC, configured to recognize in-cab events and respond accordingly. The in-cab system may perform lip reading to activate mobile phone service and place a call, dictate an email, change a vehicle destination, activate or change the vehicle's infotainment system and settings, or provide voice-activated web surfing. Certain functions are only available to the driver when the vehicle is operating in autonomous mode and are disabled in other circumstances.

[0176] The video image compositer may include enhanced temporal noise reduction for both spatial and temporal noise reduction. For example, in the case of motion in the video, the noise reduction appropriately weights the spatial information and reduces the weight of information provided by adjacent frames. In the case where an image or portion of an image does not include motion, the temporal noise reduction performed by the video image compositer may use information from previous images to reduce noise in the current image.

[0177] The video image compositor may also be configured to perform stereoscopic rectification on the input stereoscopic footage frames. The video image compositor may further be used for user interface composition when the operating system desktop is in use and the GPU 408 does not need to continuously render new surfaces. Even when the GPU 408 is powered on and active for 3D rendering, the video image compositor may be used to offload the GPU 408 to improve performance and responsiveness.

[0178] SoC 404 may further include a mobile industry processor interface (MIPI) camera serial interface for receiving video and input from a camera, a high-speed interface, and / or a video input block that may be used for camera and related pixel input functions. SoC 404 may further include an input / output controller that may be controlled by software and may be used to receive I / O signals that are not committed to a specific role.

[0179] SoC 404 may further include a wide range of peripheral interfaces to enable communication with peripherals, audio codecs, power management and / or other devices. SoC 404 may be used to process data from cameras (connected via Gigabit multimedia serial link and Ethernet), sensors (e.g., LIDAR sensor 464, RADAR sensor 460, etc., which may be connected via Ethernet), data from bus 402 (e.g., speed of vehicle 400, steering wheel position, etc.), data from GNSS sensor 458 (connected via Ethernet or CAN bus). SoC 404 may further include dedicated high-performance mass storage controllers, which may include their own DMA engines, and which may be used to free up CPU 406 from routine data management tasks.

[0180] SoC 404 can be an end-to-end platform with a flexible architecture that spans automation levels 3-5, providing a comprehensive functional safety architecture that leverages and efficiently uses computer vision and ADAS technologies to achieve diversity and redundancy, together with deep learning tools to provide a platform for a flexible and reliable driving software stack. SoC 404 can be faster, more reliable, and even more energy efficient and space efficient than conventional systems. For example, when combined with CPU 406, GPU 408, and data storage 416, accelerator 414 can provide a fast and efficient platform for level 3-5 autonomous vehicles.

[0181] The technology thus provides capabilities and functionality that cannot be achieved by conventional systems. For example, computer vision algorithms can be executed on CPUs, which can be configured using high-level programming languages ​​such as the C programming language to execute a wide variety of processing algorithms across a wide variety of visual data. However, CPUs often fail to meet the performance requirements of many computer vision applications, such as those related to, for example, execution time and power consumption. In particular, many CPUs are unable to execute complex object detection algorithms in real time, which is a requirement for in-vehicle ADAS applications and a requirement for practical Level 3-5 autonomous vehicles.

[0182] In contrast to conventional systems, by providing a CPU complex, a GPU complex, and a hardware acceleration cluster, the techniques described herein allow multiple neural networks to be executed simultaneously and / or sequentially, and the results to be combined together to achieve Level 3-5 autonomous driving functions. For example, a CNN executed on a DLA or dGPU (e.g., GPU 420) may include text and word recognition, allowing the supercomputer to read and understand traffic signs, including signs for which a neural network has not been specifically trained. The DLA may further include a neural network capable of recognizing, interpreting, and providing semantic understanding of the signs, and passing that semantic understanding to a path planning module running on the CPU complex.

[0183] As another example, as required for Level 3, 4, or 5 driving, multiple neural networks may be running simultaneously. For example, a warning sign consisting of "Caution: Flashing Lights Indicate Icing Conditions" along with electric lights may be interpreted by several neural networks independently or collectively. The sign itself may be recognized as a traffic sign by a deployed first neural network (e.g., a trained neural network), and the text "Flashing Lights Indicate Icing Conditions" may be interpreted by a deployed second neural network that informs the vehicle's path planning software (preferably executing on a CPU complex) that icing conditions exist when the flashing lights are detected. The flashing lights may be identified by operating a deployed third neural network over multiple frames that informs the vehicle's path planning software of the presence (or absence) of the flashing lights. All three neural networks may be running simultaneously, for example, within the DLA and / or on the GPU 408.

[0184] In some examples, a CNN for facial recognition and owner recognition can use data from the camera sensor to identify the presence of an authorized driver and / or owner of the vehicle 400. The always-on sensor processing engine can be used to unlock the vehicle and turn on the lights when the owner approaches the driver's door, and in security mode, disable the vehicle when the owner leaves the vehicle. In this way, the SoC 404 provides security against theft and / or carjacking.

[0185] In another example, a CNN for emergency vehicle detection and identification can use data from microphone 796 to detect and identify emergency vehicle sirens. In contrast to conventional systems that use general classifiers to detect sirens and manually extract features, SoC 404 uses CNN to classify environmental and urban sounds and to classify visual data. In a preferred embodiment, the CNN running on the DLA is trained to identify the relative closing rate of emergency vehicles (e.g., by using the Doppler effect). The CNN can also be trained to identify emergency vehicles specific to the local area in which the vehicle is operating, as identified by the GNSS sensor 458. Thus, for example, when operating in Europe, the CNN will seek to detect European sirens, and when in the United States, the CNN will seek to identify only North American sirens. Once an emergency vehicle is detected, with the assistance of the ultrasonic sensor 462, the control program can be used to execute emergency vehicle safety routines to slow the vehicle, drive to the side of the road, stop the vehicle, and / or idle the vehicle until the emergency vehicle passes.

[0186] The vehicle may include a CPU 418 (e.g., a discrete CPU or dCPU) that may be coupled to the SoC 404 via a high-speed interconnect (e.g., PCIe). The CPU 418 may include, for example, an X86 processor. The CPU 418 may be used to perform any of a variety of functions, including, for example, arbitrating potentially inconsistent results between ADAS sensors and the SoC 404, and / or monitoring the status and health of the controller 436 and / or the infotainment SoC 430.

[0187] The vehicle 400 may include a GPU 420 (e.g., a discrete GPU or dGPU) that may be coupled to the SoC 404 via a high-speed interconnect (e.g., NVIDIA's NVLINK). The GPU 420 may provide additional artificial intelligence functionality, for example, by executing redundant and / or different neural networks, and may be used to train and / or update the neural network based on inputs (e.g., sensor data) from sensors of the vehicle 400.

[0188] The vehicle 400 may further include a network interface 424, which may include one or more wireless antennas 426 (e.g., one or more wireless antennas for different communication protocols, such as cellular antennas, Bluetooth antennas, etc.). The network interface 424 can be used to enable wireless connections with the cloud (e.g., with a server 478 and / or other network devices), with other vehicles, and / or with computing devices (e.g., a passenger's client device) via the Internet. In order to communicate with other vehicles, a direct link can be established between the two vehicles, and / or an indirect link can be established (e.g., across a network and through the Internet). The direct link can be provided using a vehicle-to-vehicle communication link. The vehicle-to-vehicle communication link can provide the vehicle 400 with information about vehicles approaching the vehicle 400 (e.g., vehicles in front of, to the side of, and / or behind the vehicle 400). This functionality can be part of the cooperative adaptive cruise control functionality of the vehicle 400.

[0189] The network interface 424 may include a SoC that provides modulation and demodulation functions and enables the controller 436 to communicate over a wireless network. The network interface 424 may include a radio frequency front end for up-conversion from baseband to radio frequency and down-conversion from radio frequency to baseband. The frequency conversion may be performed by a known process and / or may be performed using a super-heterodyne process. In some examples, the radio frequency front end function may be provided by a separate chip. The network interface may include a wireless function for communicating via LTE, WCDMA, UMTS, GSM, CDMA2000, Bluetooth, Bluetooth LE, Wi-Fi, Z-wave, ZigBee, LoRaWAN, and / or other wireless protocols.

[0190] The vehicle 400 may further include data storage 428, which may include off-chip storage (e.g., outside the SoC 404). The data storage 428 may include one or more storage elements, including RAM, SRAM, DRAM, VRAM, flash memory, hard disk, and / or other components and / or devices that can store at least one bit of data.

[0191] The vehicle 400 may further include a GNSS sensor 458. The GNSS sensor 458 (e.g., GPS, assisted GPS sensor, differential GPS (DGPS) sensor, etc.) is used to assist in mapping, perception, occupancy grid generation, and / or path planning functions. Any number of GNSS sensors 458 may be used, including, for example and without limitation, GPS using a USB connector with an Ethernet to serial (RS-232) bridge.

[0192] The vehicle 400 may further include a RADAR sensor 460. The RADAR sensor 460 may be used by the vehicle 400 for remote vehicle detection even in darkness and / or inclement weather conditions. The RADAR functional safety level may be ASIL B. The RADAR sensor 460 may use CAN and / or bus 402 (e.g., to transmit data generated by the RADAR sensor 460) for control and access to object tracking data, accessing Ethernet in some examples to access raw data. A variety of RADAR sensor types may be used. For example and without limitation, the RADAR sensor 460 may be suitable for front, rear, and side RADAR use. In some examples, a pulse Doppler RADAR sensor is used.

[0193] The RADAR sensor 460 may include different configurations, such as long-range with a narrow field of view, short-range with a wide field of view, short-range side coverage, and the like. In some examples, the long-range RADAR may be used for adaptive cruise control functions. The long-range RADAR system may provide a wide field of view (e.g., within 250m) achieved through two or more independent browsing. The RADAR sensor 460 may help distinguish between static objects and moving objects, and may be used by the ADAS system for emergency braking assistance and forward collision warnings. The long-range RADAR sensor may include a single-station multimode RADAR with multiple (e.g., six or more) fixed RADAR antennas and high-speed CAN and FlexRay interfaces. In an example with six antennas, the central four antennas may create a focused beam pattern designed to record the surroundings of the vehicle 400 at a higher rate with minimal traffic interference from adjacent lanes. The other two antennas may expand the field of view, making it possible to quickly detect vehicles entering or leaving the lane of the vehicle 400.

[0194] As an example, a medium-range RADAR system may include a range of up to 160m (front) or 80m (rear) and a field of view of up to 42 degrees (front) or 150 degrees (rear). A short-range RADAR system may include, but is not limited to, a RADAR sensor designed to be mounted on both ends of a rear bumper. When mounted on both ends of a rear bumper, such a RADAR sensor system may create two beams that continuously monitor the blind spots behind and beside the vehicle.

[0195] Short-range RADAR systems can be used in ADAS systems for blind spot detection and / or lane change assistance.

[0196] The vehicle 400 may further include ultrasonic sensors 462. The ultrasonic sensors 462, which may be placed on the front, rear, and / or sides of the vehicle 400, may be used for parking assistance and / or creating and updating occupancy grids. A variety of ultrasonic sensors 462 may be used, and different ultrasonic sensors 462 may be used for different detection ranges (e.g., 2.5 m, 4 m). The ultrasonic sensors 462 may operate at a functional safety level of ASIL B.

[0197] The vehicle 400 may include a LIDAR sensor 464. The LIDAR sensor 464 may be used for object and pedestrian detection, emergency braking, collision avoidance, and / or other functions. The LIDAR sensor 464 may be ASIL B for functional safety level. In some examples, the vehicle 400 may include multiple LIDAR sensors 464 (e.g., two, four, six, etc.) that may use Ethernet (e.g., to provide data to a Gigabit Ethernet switch).

[0198] In some examples, LIDAR sensor 464 may be able to provide a list of objects and their distances for a 360-degree field of view. Commercially available LIDAR sensor 464 may have, for example, an advertised range of approximately 100m, an accuracy of 2cm-3cm, and support for 100Mbps Ethernet connections. In some examples, one or more non-protruding LIDAR sensors 464 may be used. In such examples, LIDAR sensor 464 may be implemented as a small device that can be embedded in the front, back, side, and / or corner of vehicle 400. In such examples, LIDAR sensor 464 may provide a field of view of up to 120 degrees horizontally and 35 degrees vertically, with a range of 200m, even for low reflectivity objects. Front-mounted LIDAR sensor 464 may be configured for a horizontal field of view between 45 and 135 degrees.

[0199] In some examples, LIDAR technologies such as 3D flash LIDAR may also be used. 3D flash LIDAR uses flashes of laser as an emission source to illuminate the vehicle's surroundings up to about 200m. The flash LIDAR unit includes a receiver that records the laser pulse transmission time and reflected light on each pixel, which in turn corresponds to the range from the vehicle to the object. Flash LIDAR can allow highly accurate and distortion-free images of the surrounding environment to be generated with each laser flash. In some examples, four flash LIDAR sensors can be deployed, one on each side of the vehicle 400. Available 3D flash LIDAR systems include solid-state 3D staring array LIDAR cameras (e.g., non-browsing LIDAR devices) with no moving parts other than fans. The flash LIDAR device can use 5 nanosecond Class I (eye-safe) laser pulses per frame, and can capture the reflected laser light in the form of a 3D range point cloud and co-registered intensity data. By using flash LIDAR, and because flash LIDAR is a solid-state device with no moving parts, LIDAR sensor 464 may be less susceptible to motion blur, vibration, and / or shock.

[0200] The vehicle may further include an IMU sensor 466. In some examples, the IMU sensor 466 may be located at the center of the rear axle of the vehicle 400. The IMU sensor 466 may include, for example and without limitation, an accelerometer, a magnetometer, a gyroscope, a magnetic compass, and / or other sensor types. In some examples, for example, in a six-axis application, the IMU sensor 466 may include an accelerometer and a gyroscope, and in a nine-axis application, the IMU sensor 466 may include an accelerometer, a gyroscope, and a magnetometer.

[0201] In some embodiments, IMU sensor 466 can be implemented as a miniature high-performance GPS-assisted inertial navigation system (GPS / INS) that combines micro-electromechanical system (MEMS) inertial sensors, high-sensitivity GPS receivers, and advanced Kalman filter algorithms to provide estimates of position, velocity, and attitude. Thus, in some examples, IMU sensor 466 can enable vehicle 400 to estimate heading by directly observing and correlating velocity changes from GPS to IMU sensor 466 without input from a magnetic sensor. In some examples, IMU sensor 466 and GNSS sensor 458 can be combined into a single integrated unit.

[0202] The vehicle may include microphones 496 positioned in and / or around the vehicle 400. The microphones 496 may be used for, among other things, emergency vehicle detection and identification.

[0203] The vehicle may further include any number of camera types, including a stereo camera 468, a wide angle camera 470, an infrared camera 472, a surround camera 474, a long-range and / or mid-range camera 498, and / or other camera types. These cameras can be used to capture image data around the entire periphery of the vehicle 400. The type of camera used depends on the embodiment and the requirements of the vehicle 400, and any combination of camera types can be used to provide the necessary coverage around the vehicle 400. In addition, the number of cameras can vary depending on the embodiment. For example, the vehicle may include six cameras, seven cameras, ten cameras, twelve cameras, and / or another number of cameras. As an example and not limitation, the cameras can support Gigabit Multimedia Serial Link (GMSL) and / or Gigabit Ethernet. Each of the cameras described herein may be capable of supporting Gigabit Multimedia Serial Link (GMSL) and / or Gigabit Ethernet. Figure 4A and Figure 4B Described in more detail.

[0204] The vehicle 400 may further include a vibration sensor 442. The vibration sensor 442 may measure vibrations of components of the vehicle, such as an axle. For example, changes in vibration may indicate changes in the road surface. In another example, when two or more vibration sensors 442 are used, the difference between the vibrations may be used to determine friction or slip of the road surface (e.g., when there is a vibration difference between a powered drive shaft and a free-spinning shaft).

[0205] The vehicle 400 may include an ADAS system 438. In some examples, the ADAS system 438 may include a SoC. The ADAS system 438 may include autonomous / adaptive / automatic cruise control (ACC), cooperative adaptive cruise control (CACC), forward collision warning (FCW), automatic emergency braking (AEB), lane departure warning (LDW), lane keeping assist (LKA), blind spot warning (BSW), rear cross traffic warning (RCTW), collision warning system (CWS), lane centering (LC), and / or other features and functions.

[0206] The ACC system may use a RADAR sensor 460, a LIDAR sensor 464, and / or a camera. The ACC system may include a longitudinal ACC and / or a lateral ACC. The longitudinal ACC monitors and controls the distance to the vehicle immediately in front of the vehicle 400, and automatically adjusts the vehicle speed to maintain a safe distance from the vehicle in front. The lateral ACC performs distance keeping and suggests that the vehicle 400 change lanes when necessary. The lateral ACC is related to other ADAS applications such as LCA and CWS.

[0207] CACC uses information from other vehicles, which can be received indirectly from other vehicles via a wireless link or through a network connection (e.g., through the Internet) via a network interface 424 and / or a wireless antenna 426. A direct link can be provided by a vehicle-to-vehicle (V2V) communication link, while an indirect link can be an infrastructure-to-vehicle (I2V) communication link. Typically, the V2V communication concept provides information about the vehicle immediately ahead (e.g., the vehicle immediately ahead of the vehicle 400 and in the same lane as it), while the I2V communication concept provides information about traffic farther ahead. The CACC system may include either or both of the I2V and V2V information sources. Given information about the vehicle ahead of the vehicle 400, CACC can be more reliable, and it is possible to improve the smoothness of traffic flow and reduce road congestion.

[0208] The FCW system is designed to alert the driver to hazards so that the driver can take corrective action. The FCW system uses a front camera and / or RADAR sensor 460 coupled to a dedicated processor, DSP, FPGA and / or ASIC, which is electrically coupled to driver feedback such as a display, speaker and / or vibrating components. The FCW system can provide warnings in the form of, for example, sound, visual warnings, vibrations and / or rapid brake pulses.

[0209] The AEB system detects an impending front collision with another vehicle or other object and can automatically apply the brakes if the driver does not take corrective action within specified time or distance parameters. The AEB system can use a front camera and / or RADAR sensor 460 coupled to a dedicated processor, DSP, FPGA and / or ASIC. When the AEB system detects a hazard, it typically first alerts the driver to take corrective action to avoid the collision, and if the driver does not take corrective action, the AEB system can automatically apply the brakes in an effort to prevent or at least mitigate the effects of the predicted collision. The AEB system may include technologies such as dynamic brake support and / or collision approach braking.

[0210] The LDW system provides visual, audible, and / or tactile warnings, such as steering wheel or seat vibrations, to alert the driver when the vehicle 400 crosses a lane marking. When the driver indicates an intention to leave the lane, by activating a turn signal, the LDW system is not activated. The LDW system may use a front-facing camera coupled to a dedicated processor, DSP, FPGA, and / or ASIC that is electrically coupled to driver feedback such as a display, speaker, and / or vibrating component.

[0211] The LKA system is a variation of the LDW system. If the vehicle 400 begins to leave the lane, the LKA system provides steering input or braking to correct the vehicle 400. The BSW system detects and warns the driver of vehicles in the car's blind spot. The BSW system may provide visual, audible, and / or tactile alerts to indicate that it is unsafe to merge or change lanes. The system may provide additional warnings when the driver uses a turn signal. The BSW system may use one or more rear-facing cameras and / or one or more RADAR sensors.

[0212] The RCTW system may provide visual, audible, and / or tactile notifications when an object is detected outside the range of the rear camera while the vehicle 400 is in reverse. Some RCTW systems include AEB to ensure that the vehicle brakes are applied to avoid a crash. The RCTW system may use one or more rear RADAR sensors 460 coupled to a dedicated processor, DSP, FPGA, and / or ASIC that is electrically coupled to driver feedback such as a display, speaker, and / or vibrating component.

[0213] Conventional ADAS systems may be prone to false positive results, which may annoy and distract the driver, but are typically not catastrophic because the ADAS system alerts the driver and allows the driver to decide whether the safety condition really exists and take action accordingly. However, in the autonomous vehicle 400, in the case of conflicting results, the vehicle 400 itself must decide whether to pay attention to the results from the main computer or the auxiliary computer (e.g., the first controller 436 or the second controller 436). For example, in some embodiments, the ADAS system 438 can be a backup and / or auxiliary computer for providing perception information to the backup computer rationality module. The backup computer rationality monitor can run redundant and diverse software on hardware components to detect faults in perception and dynamic driving tasks. The output from the ADAS system 438 can be provided to the supervisory MCU. If the outputs from the main computer and the auxiliary computer conflict, the supervisory MCU must determine how to coordinate the conflict to ensure safe operation.

[0214] In some examples, the master computer can be configured to provide a confidence score to the supervisory MCU, indicating the master computer's confidence in the selected result. If the confidence score exceeds a threshold, the supervisory MCU can follow the direction of the master computer, regardless of whether the auxiliary computer provides conflicting or inconsistent results. In the case where the confidence score does not meet the threshold and the master computer and the auxiliary computer indicate different results (e.g., conflicts), the supervisory MCU can arbitrate between these computers to determine the appropriate result.

[0215] The supervisory MCU may be configured to run a neural network that is trained and configured to determine conditions under which the auxiliary computer provides a false alarm based on outputs from the primary computer and the auxiliary computer. Thus, the neural network in the supervisory MCU may learn when the output of the auxiliary computer may be trusted and when it may not. For example, when the auxiliary computer is a RADAR-based FCW system, the neural network in the supervisory MCU may learn when the FCW system is identifying a metal object that is not actually dangerous, such as a drain grate or manhole cover that triggers an alarm. Similarly, when the auxiliary computer is a camera-based LDW system, the neural network in the supervisory MCU may learn to ignore the LDW when a cyclist or pedestrian is present and lane departure is actually the safest strategy. In an embodiment that includes a neural network running on the supervisory MCU, the supervisory MCU may include at least one of a DLA or GPU suitable for running the neural network with associated memory. In a preferred embodiment, the supervisory MCU may include a component of SoC 404 and / or be included as a component of SoC 404.

[0216] In other examples, the ADAS system 438 may include an auxiliary computer that uses traditional computer vision rules to perform ADAS functions. In this way, the auxiliary computer can use classic computer vision rules (if-then), and the presence of a neural network in the supervisory MCU can improve reliability, safety, and performance. For example, diverse implementations and intentional non-identity make the entire system more fault-tolerant, especially for failures caused by software (or software-hardware interface) functions. For example, if there is a software vulnerability or error in the software running on the main computer and the non-identical software code running on the auxiliary computer provides the same overall result, the supervisory MCU can be more confident that the overall result is correct and that the vulnerability in the software or hardware on the main computer does not cause a substantial error.

[0217] In some examples, the output of the ADAS system 438 can be fed to the perception block of the main computer and / or the dynamic driving task block of the main computer. For example, if the ADAS system 438 indicates a forward collision warning due to an object immediately ahead, the perception block can use this information in identifying the object. In other examples, the auxiliary computer can have its own neural network that is trained and thus reduces the risk of false positives as described herein.

[0218] The vehicle 400 may further include an infotainment SoC 430 (e.g., an in-vehicle infotainment system (IVI)). Although illustrated and described as an SoC, the infotainment system may not be an SoC and may include two or more discrete components. The infotainment SoC 430 may include a combination of hardware and software that may be used to provide audio (e.g., music, personal digital assistant, navigation instructions, news, radio, etc.), video (e.g., TV, movies, streaming, etc.), phone (e.g., hands-free calling), network connectivity (e.g., LTE, Wi-Fi, etc.), and / or information services (e.g., navigation system, rear parking assistance, radio data system, vehicle-related information such as fuel level, total distance covered, brake fuel level, oil level, door open / closed, air filter information, etc.) to the vehicle 400. For example, the infotainment SoC 430 may include a radio, a disc player, a navigation system, a video player, USB and Bluetooth connectivity, an onboard computer, onboard entertainment, Wi-Fi, steering wheel audio controls, hands-free voice controls, a head-up display (HUD), an HMI display 434, a telematics device, a control panel (e.g., for controlling and / or interacting with various components, features, and / or systems), and / or other components. The infotainment SoC 430 may further be used to provide information (e.g., visual and / or auditory) to a user of the vehicle, such as information from an ADAS system 438, autonomous driving information such as planned vehicle maneuvers, trajectories, surrounding environment information (e.g., intersection information, vehicle information, road information, etc.), and / or other information.

[0219] The infotainment SoC 430 may include GPU functionality. The infotainment SoC 430 may communicate with other devices, systems, and / or components of the vehicle 400 via a bus 402 (e.g., a CAN bus, Ethernet, etc.). In some examples, the infotainment SoC 430 may be coupled to a supervisory MCU so that in the event of a failure of a master controller 436 (e.g., a main and / or backup computer of the vehicle 400), the GPU of the infotainment system may perform some self-driving functions. In such an example, the infotainment SoC 430 may place the vehicle 400 in a driver-safe parking mode as described herein.

[0220] The vehicle 400 may further include an instrument cluster 432 (e.g., a digital instrument panel, an electronic instrument cluster, a digital instrument panel, etc.). The instrument cluster 432 may include a controller and / or a supercomputer (e.g., a separate controller or a supercomputer). The instrument cluster 432 may include a set of instruments, such as a speedometer, a fuel level, an oil pressure, a tachometer, an odometer, a turn indicator, a shift position indicator, a seat belt warning light, a parking brake warning light, an engine fault light, an airbag (SRS) system information, lighting controls, safety system controls, navigation information, etc. In some examples, information may be displayed and / or shared between the infotainment SoC 430 and the instrument cluster 432. In other words, the instrument cluster 432 may be included as part of the infotainment SoC 430, or vice versa.

[0221] Figure 4D For a cloud-based server and Figure 4A 4. System diagram of communication between example autonomous vehicles 400. System 476 may include server 478, network 490, and vehicles including vehicle 400. Server 478 may include multiple GPUs 484 (A)-484 (H) (collectively referred to as GPU 484 here), PCIe switches 482 (A)-482 (H) (collectively referred to as PCIe switches 482 here), and / or CPUs 480 (A)-480 (B) (collectively referred to as CPU 480 here). GPU 484, CPU 480, and PCIe switches may be interconnected with high-speed interconnects and / or PCIe connections 486 such as, for example and without limitation, NVLink interface 488 developed by NVIDIA. In some examples, GPU 484 is connected via NVLink and / or NVSwitch SoC, and GPU 484 and PCIe switch 482 are connected via PCIe interconnects. Although eight GPUs 484, two CPUs 480, and two PCIe switches are shown, this is not intended to be limiting. Depending on the embodiment, each of the servers 478 may include any number of GPUs 484, CPUs 480, and / or PCIe switches. For example, each of the servers 478 may include eight, sixteen, thirty-two, and / or more GPUs 484.

[0222] Server 478 may receive image data over network 490 and from a vehicle, the image data representing images showing unexpected or changed road conditions, such as recently begun road work. Server 478 may transmit neural network 492, updated neural network 492, and / or map information 494, including information about traffic and road conditions, over network 490 and to the vehicle. Updates to map information 494 may include updates to HD map 422, such as information about construction sites, potholes, curves, flooding, or other obstacles. In some examples, neural network 492, updated neural network 492, and / or map information 494 may have been generated from new training and / or data received from any number of vehicles in the environment and / or based on experience from training performed at a data center (e.g., using server 478 and / or other servers).

[0223] Server 478 can be used to train a machine learning model (e.g., a neural network) based on training data. The training data can be generated by the vehicle, and / or can be generated in a simulation (e.g., using a game engine). In some examples, the training data is labeled (e.g., when the neural network benefits from supervised learning) and / or undergoes other preprocessing, while in other examples, the training data is not labeled and / or preprocessed (e.g., when the neural network does not require supervised learning). Training can be performed according to any one or more classes of machine learning techniques, including but not limited to the following classes: supervised training, semi-supervised training, unsupervised training, self-learning, reinforcement learning, joint learning, transfer learning, feature learning (including principal component and cluster analysis), multilinear subspace learning, manifold learning, representation learning (including alternate dictionary learning), rule-based machine learning, anomaly detection, and any variants or combinations thereof. Once the machine learning model is trained, the machine learning model can be used by the vehicle (e.g., transmitted to the vehicle via network 490), and / or the machine learning model can be used by server 478 to remotely monitor the vehicle.

[0224] In some examples, server 478 can receive data from the vehicle and apply the data to the latest real-time neural network for real-time intelligent reasoning. Server 478 may include a deep learning supercomputer and / or a dedicated AI computer powered by GPU 484, such as DGX and DGX Station machines developed by NVIDIA. However, in some examples, server 478 may include a deep learning infrastructure of a data center powered only by CPUs.

[0225] The deep learning infrastructure of server 478 may be capable of rapid real-time inference, and may use this capability to assess and verify the health of the processors, software, and / or associated hardware in vehicle 400. For example, the deep learning infrastructure may receive periodic updates from vehicle 400, such as a sequence of images and / or objects located in the sequence of images that vehicle 400 has located (e.g., via computer vision and / or other machine learning object classification techniques). The deep learning infrastructure may run its own neural network to identify objects and compare them to the objects identified by vehicle 400, and if the results do not match and the infrastructure concludes that the AI ​​in vehicle 400 has failed, server 478 may transmit a signal to vehicle 400 instructing the fail-safe computer of vehicle 400 to take control, notify passengers, and complete a safe parking maneuver.

[0226] For reasoning, the server 478 may include a GPU 484 and one or more programmable reasoning accelerators (e.g., NVIDIA's TensorRT 3). The combination of GPU-powered servers and reasoning acceleration can make real-time responses possible. In other examples, such as when performance is not so important, CPU, FPGA, and other processor-powered servers can be used for reasoning.

[0227] Example computing device

[0228] Figure 5 5 is a block diagram of an example computing device 500 suitable for implementing some embodiments of the present disclosure. The computing device 500 may include an interconnect system 502 that directly or indirectly couples the following devices: memory 504, one or more central processing units (CPUs) 506, one or more graphics processing units (GPUs) 508, a communication interface 510, an input / output (I / O) port 512, an input / output component 514, a power supply 516, one or more presentation components 518 (e.g., a display), and one or more logic units 520. In at least one embodiment, the computing device 500 may include one or more virtual machines (VMs), and / or any of its components may include virtual components (e.g., virtual hardware components). For non-limiting examples, the one or more GPUs 508 may include one or more vGPUs, the one or more CPUs 506 may include one or more vCPUs, and / or the one or more logic units 520 may include one or more virtual logic units. Thus, computing device 500 may include discrete components (eg, a complete GPU dedicated to computing device 500 ), virtual components (eg, a portion of a GPU dedicated to computing device 500 ), or a combination thereof.

[0229] although Figure 5The various blocks of are shown as being connected via an interconnect system 502 having wires, but this is not intended to be limiting and is merely for clarity. For example, in some embodiments, a presentation component 518 such as a display device may be considered an I / O component 514 (e.g., if the display is a touch screen). As another example, CPU 506 and / or GPU 508 may include memory (e.g., memory 504 may represent a storage device in addition to the memory of GPU 508, CPU 506, and / or other components). In other words, Figure 5 The computing devices described herein are illustrative only. No distinction is made between categories such as "workstations," "servers," "laptops," "desktops," "tablets," "client devices," "mobile devices," "handheld devices," "game consoles," "electronic control units (ECUs)," "virtual reality systems," and / or other device or system types, as all are considered Figure 5 within the range of computing devices.

[0230] The interconnection system 502 may represent one or more links or buses, such as an address bus, a data bus, a control bus, or a combination thereof. The interconnection system 502 may include one or more links or bus types, such as an industry standard architecture (ISA) bus, an extended industry standard architecture (EISA) bus, a video electronics standard association (VESA) bus, a peripheral component interconnect (PCI) bus, a peripheral component interconnect express (PCIe) bus, and / or another type of bus or link. In some embodiments, there is a direct connection between components. As an example, the CPU 506 may be directly connected to the memory 504. In addition, the CPU 506 may be directly connected to the GPU 508. In the case where there is a direct or point-to-point connection between components, the interconnection system 502 may include a PCIe link to perform the connection. In these examples, the computing device 500 does not need to include a PCI bus.

[0231] Memory 504 may include any of a variety of computer-readable media. Computer-readable media may be any available media that can be accessed by computing device 500. Computer-readable media may include volatile and nonvolatile media and removable and non-removable media. By way of example and not limitation, computer-readable media may include computer storage media and communication media.

[0232] Computer storage media may include volatile and nonvolatile media and / or removable and non-removable media implemented in any method or technology for storing information such as computer-readable instructions, data structures, program modules, and / or other data types. For example, memory 504 may store computer-readable instructions (e.g., representing programs and / or program elements, such as an operating system). Computer storage media may include, but are not limited to, RAM, ROM, EEPROM, flash memory or other storage technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by computing device 500. As used herein, computer storage media does not include the signals themselves.

[0233] Computer storage media may contain computer readable instructions, data structures, program modules, and / or other data types in a modulated data signal such as a carrier wave or other transmission mechanism, and include any information delivery media. The term "modulated data signal" may refer to a signal that has one or more of its characteristics set or changed in such a way that information is encoded into the signal. By way of example and not limitation, computer storage media may include wired media such as a wired network or a direct wired connection, and wireless media such as sound, RF, infrared, and other wireless media. Any combination of the above should also be included within the scope of computer readable media.

[0234] The CPU 506 may be configured to execute at least some of the computer-readable instructions so as to control one or more components of the computing device 500 to perform one or more of the methods and / or processes described herein. Each of the CPUs 506 may include one or more cores (e.g., one, two, four, eight, twenty-eight, seventy-two, etc.) capable of processing a large number of software threads simultaneously. The CPU 506 may include any type of processor, and may include different types of processors, depending on the type of computing device 500 implemented (e.g., a processor with fewer cores for mobile devices and a processor with more cores for servers). For example, depending on the type of computing device 500, the processor may be an Advanced RISC Mechanism (ARM) processor implemented using Reduced Instruction Set Computing (RISC) or an x86 processor implemented using Complex Instruction Set Computing (CISC). The computing device 500 may also include one or more CPUs 506 in addition to one or more microprocessors or supplementary coprocessors such as math coprocessors.

[0235] In addition to or in place of the CPU 506, the GPU 508 may also be configured to execute at least some computer-readable instructions to control one or more components of the computing device 500 to perform one or more methods and / or processes described herein. One or more GPUs 508 may be integrated GPUs (e.g., with one or more CPUs 506) and / or one or more GPUs 508 may be discrete GPUs. In an embodiment, one or more GPUs 508 may be coprocessors of one or more CPUs 506. The computing device 500 may use the GPU 508 to render graphics (e.g., 3D graphics) or perform general-purpose computing. For example, the GPU 508 may be used for general-purpose computing on a GPU (GPGPU). The GPU 508 may include hundreds or thousands of cores capable of processing hundreds or thousands of software threads simultaneously. The GPU 508 may generate pixel data for outputting an image in response to a rendering command (e.g., a rendering command from the CPU 506 received via a host interface). The GPU 508 may include a graphics memory such as a display memory for storing pixel data or any other suitable data (e.g., GPGPU data). Display memory may be included as part of memory 504. GPU 508 may include two or more GPUs operating in parallel (e.g., via a link). The link may connect the GPUs directly (e.g., using NVLINK) or through a switch (e.g., using NVSwitch). When combined together, each GPU 508 may generate pixel data or GPGPU data for different portions of output or for different outputs (e.g., a first GPU for a first image and a second GPU for a second image). Each GPU may include its own memory or may share memory with other GPUs.

[0236] In addition to or in lieu of the CPU 506 and / or the GPU 508, the logic unit 520 may be configured to execute at least some computer-readable instructions to control one or more components of the computing device 500 to perform one or more methods and / or processes described herein. In an embodiment, the CPU 506, the GPU 508, and / or the logic unit 520 may discretely or jointly perform any combination of methods, processes, and / or portions thereof. The one or more logic units 520 may be a part of and / or integrated into the one or more CPUs 506 and / or the one or more GPUs 508 and / or the one or more logic units 520 may be a discrete component of the CPU 506 and / or the GPU 508 or otherwise external thereto. In an embodiment, the one or more logic units 520 may be a processor of the one or more CPUs 506 and / or the one or more GPUs 508.

[0237] Examples of logic unit 520 include one or more processing cores and / or components thereof, such as a data processing unit (DPU), a tensor core (TC), a tensor processing unit (TPU), a pixel vision core (PVC), a vision processing unit (VPU), a graphics processing cluster (GPC), a texture processing cluster (TPC), a streaming multiprocessor (SM), a tree traversal unit (TTU), an artificial intelligence accelerator (AIA), a deep learning accelerator (DLA), an arithmetic logic unit (ALU), an application specific integrated circuit (ASIC), a floating point unit (FPU), an input / output (I / O) element, a peripheral component interconnect (PCI) or a peripheral component interconnect express (PCIe) element, etc.

[0238] The communication interface 510 may include one or more receivers, transmitters, and / or transceivers that enable the computing device 500 to communicate with other computing devices via an electronic communication network, including wired and / or wireless communications. The communication interface 510 may include components and functions that enable communication over any of a number of different networks, such as wireless networks (e.g., Wi-Fi, Z-wave, Bluetooth, Bluetooth LE, ZigBee, etc.), wired networks (e.g., communicating over Ethernet or InfiniBand), low power wide area networks (e.g., LoRaWAN, SigFox, etc.), and / or the Internet. In one or more embodiments, the logic unit 520 and / or the communication interface 510 may include one or more data processing units (DPUs) to transmit data received over a network and / or over the interconnect system 502 directly to one or more GPUs 508 (e.g., memory in a GPU 508).

[0239] The I / O port 512 can enable the computing device 500 to be logically coupled to other devices including I / O components 514, presentation components 518, and / or other components, some of which can be built into (e.g., integrated into) the computing device 500. Illustrative I / O components 514 include microphones, mice, keyboards, joysticks, game pads, game controllers, satellite dishes, browsers, printers, wireless devices, and the like. The I / O components 514 can provide a natural user interface (NUI) that processes air gestures, voice, or other physiological inputs generated by a user. In some instances, the input can be transmitted to an appropriate network element for further processing. The NUI can implement any combination of voice recognition, stylus recognition, facial recognition, biometric recognition, gesture recognition on and adjacent to the screen, air gestures, head and eye tracking, and touch recognition associated with the display of the computing device 500 (as described in more detail in this disclosure). The computing device 500 may include a depth camera such as a stereo camera system, an infrared camera system, an RGB camera system, touch screen technology, and combinations of these for gesture detection and recognition. Additionally, computing device 500 may include an accelerometer or gyroscope to enable motion detection (e.g., as part of an inertial measurement unit (IMU)). In some examples, the output of the accelerometer or gyroscope may be used by computing device 500 to render immersive augmented reality or virtual reality.

[0240] The power supply 516 may include a hardwired power supply, a battery power supply, or a combination thereof. The power supply 516 may provide power to the computing device 500 to enable the components of the computing device 500 to operate.

[0241] The presentation component 518 may include a display (e.g., a monitor, a touch screen, a television screen, a head-up display (HUD), other display types, or a combination thereof), a speaker, and / or other presentation components. The presentation component 518 may receive data from other components (e.g., GPU 508, CPU 506, DPU, etc.) and output the data (e.g., as an image, video, sound, etc.).

[0242] Sample Data Center

[0243] Figure 6 An example data center 600 is shown, which may be used in at least one embodiment of the present disclosure. The data center 600 may include a data center infrastructure layer 610, a framework layer 620, a software layer 630, and an application layer 640.

[0244] like Figure 6As shown, the data center infrastructure layer 610 may include a resource coordinator 612, grouped computing resources 614, and node computing resources ("node CRs") 616(1)-616(N), where "N" represents any complete positive integer. In at least one embodiment, the node CRs 616(1)-616(N) may include, but are not limited to, any number of central processing units (CPUs) or other processors (including DPUs, accelerators, field programmable gate arrays (FPGAs), graphics processors or graphics processing units (GPUs), etc.), memory devices (e.g., dynamic read-only memory), storage devices (e.g., solid-state drives or disk drives), network input / output (NW I / O) devices, network switches, virtual machines (VMs), power modules and cooling modules, etc. In some embodiments, one or more of the node CRs 616(1)-616(N) may correspond to a server having one or more of the above computing resources. Furthermore, in some embodiments, node CRs 616 ( 1 )- 616 (N) may include one or more virtual components, such as vGPUs, vCPUs, etc., and / or one or more of node CRs 616 ( 1 )- 616 (N) may correspond to virtual machines (VMs).

[0245] In at least one embodiment, the computing resources 614 of the grouping may include a separate grouping (not shown) of the node CR616 contained in one or more racks, or many racks (also not shown) contained in the data center of each geographical location. The separate grouping of the node CR616 in the computing resources 614 of the grouping may include computing, network, memory or storage resources that can be configured or assigned to support the grouping of one or more workloads. In at least one embodiment, several node CR616 including CPU, GPU, DPU and / or other processors may be grouped in one or more racks to provide computing resources to support one or more workloads. One or more racks may also include any number of power modules, cooling modules and / or network switches in any combination.

[0246] Resource coordinator 612 may configure or otherwise control one or more nodes CR 616(1)-616(N) and / or grouped computing resources 614. In at least one embodiment, resource coordinator 612 may include a software design infrastructure (SDI) management entity for data center 600. Resource coordinator 612 may include hardware, software, or some combination thereof.

[0247] In at least one embodiment, Figure 6As shown, the framework layer 620 may include a job scheduler 633, a configuration manager 634, a resource manager 636, and a distributed file system 638. The framework layer 620 may include a framework that supports software 632 of the software layer 630 and / or one or more applications 642 of the application layer 640. The software 632 or the application 642 may include web-based service software or applications, such as those provided by Amazon Web Services, Google Cloud, and Microsoft Azure. The framework layer 620 may be, but is not limited to, a free and open source software network application framework, such as Apache Spark, which may utilize the distributed file system 638 for large-scale data processing (e.g., "big data"). TM (hereinafter referred to as "Spark"). In at least one embodiment, the job scheduler 633 may include a Spark driver to facilitate scheduling of workloads supported by various layers of the data center 600. In at least one embodiment, the configuration manager 634 may be able to configure different layers, such as the software layer 630 and the framework layer 620 including Spark and a distributed file system 638 for supporting large-scale data processing. The resource manager 636 can manage cluster or group computing resources mapped to or allocated to support the distributed file system 638 and the job scheduler 633. In at least one embodiment, the cluster or group computing resources may include grouped computing resources 614 at the data center infrastructure layer 610. The resource manager 636 can coordinate with the resource coordinator 612 to manage these mapped or allocated computing resources.

[0248] In at least one embodiment, software 632 included in software layer 630 may include software used by at least a portion of node CRs 616(1)-616(N), grouped computing resources 614, and / or distributed file system 638 of framework layer 620. One or more types of software may include, but are not limited to, Internet web page search software, email virus scanning software, database software, and streaming video content software.

[0249] In at least one embodiment, one or more applications 642 included in the application layer 640 may include one or more types of applications used by at least a portion of the node CRs 616(1)-616(N), the grouped computing resources 614, and / or the distributed file system 638 of the framework layer 620. The one or more types of applications may include, but are not limited to, any number of genomics applications, cognitive computing, and machine learning applications, including training or inference software, machine learning framework software (e.g., PyTorch, TensorFlow, Caffe, etc.), and / or other machine learning applications used in conjunction with one or more embodiments.

[0250] In at least one embodiment, any of configuration manager 634, resource manager 636, and resource coordinator 612 can implement any number and type of self-modification actions based on any number and type of data acquired in any technically feasible manner. The self-modification actions can relieve the data center operator of data center 600 from making potentially bad configuration decisions and can avoid underutilized and / or underperforming portions of the data center.

[0251] The data center 600 may include tools, services, software, or other resources for training one or more machine learning models or using one or more machine learning models to predict or infer information according to one or more embodiments described herein. For example, a machine learning model may be trained by calculating weight parameters according to a neural network architecture using the software and computing resources described in this disclosure with respect to the data center 600. In at least one embodiment, by using weight parameters calculated by one or more training techniques, the resources described in this disclosure with respect to the data center 600 may be used to infer or predict information using trained machine learning models corresponding to one or more neural networks, such as but not limited to those described herein.

[0252] In at least one embodiment, the data center 600 can use a CPU, an application-specific integrated circuit (ASIC), a GPU, an FPGA, and / or other hardware (or virtual computing resources corresponding thereto) to use the above resources to perform training and / or reasoning. In addition, one or more software and / or hardware resources described in the present disclosure can be configured as a service to allow users to train or perform information reasoning, such as image recognition, speech recognition, or other artificial intelligence services.

[0253] Example network environment

[0254] A network environment suitable for implementing embodiments of the present disclosure may include one or more client devices, servers, network attached storage (NAS), other backend devices, and / or other device types. The client devices, servers, and / or other device types (e.g., each device) may be configured to communicate with the client device, server, or other device types. Figure 5 The backend device 600 may be implemented on one or more instances of the computing device 500 of the present invention—for example, each device may include similar components, features, and / or functions of the computing device 500. In addition, in the case of implementing a backend device (e.g., a server, NAS, etc.), the backend device may be included as part of the data center 600, examples of which are described herein with respect to Figure 6 Describe in more detail.

[0255] The components of the network environment can communicate with each other through the network, which can be wired, wireless, or both. The network can include multiple networks, or a network of multiple networks. For example, the network can include one or more wide area networks (WANs), one or more local area networks (LANs), one or more public networks (such as the Internet and / or the Public Switched Telephone Network (PSTN)), and / or one or more private networks. In the case where the network includes a wireless telecommunications network, components such as base stations, communication towers, or even access points (and other components) can provide wireless connections.

[0256] Compatible network environments may include one or more peer-to-peer network environments (in which case the server may not be included in the network environment), and one or more client-server network environments (in which case one or more servers may be included in the network environment). In a peer-to-peer network environment, the functionality described herein with respect to the server may be implemented on any number of client devices.

[0257] In at least one embodiment, the network environment may include one or more cloud-based network environments, distributed computing environments, combinations thereof, and the like. A cloud-based network environment may include a framework layer, a job scheduler, a resource manager, and a distributed file system implemented on one or more servers, which may include one or more core network servers and / or edge servers. The framework layer may include a framework for supporting software at the software layer and / or one or more applications at the application layer. The software or application may include network-based service software or application programs, respectively. In an embodiment, one or more client devices may use network-based service software or application programs (e.g., by accessing the service software and / or application programs via one or more application programming interfaces (APIs)). The framework layer may be, but is not limited to, a type of free and open source software network application framework, such as a distributed file system that may be used for large-scale data processing (e.g., "big data").

[0258] A cloud-based network environment can provide cloud computing and / or cloud storage that performs any combination of the computing and / or data storage functions (or one or more portions thereof) described herein. Any of these various functions can be distributed across multiple locations from a central or core server (e.g., one or more data centers that can be distributed across states, regions, countries, the world, etc.). If the connection to the user (e.g., client device) is relatively close to an edge server, the core server can assign at least a portion of the functionality to the edge server. A cloud-based network environment can be private (e.g., limited to a single organization), public (e.g., available to many organizations), and / or a combination thereof (e.g., a hybrid cloud environment).

[0259] The client device may include Figure 5 At least some of the components, features, and functionality of the described example computing device 500. By way of example and not limitation, the client device may be embodied as a personal computer (PC), a laptop computer, a mobile device, a smartphone, a tablet computer, a smart watch, a wearable computer, a personal digital assistant (PDA), an MP3 player, a virtual reality head mounted display, a global positioning system (GPS) or device, a video player, a camera, a surveillance device or system, a vehicle, a vessel, an aircraft, a virtual machine, a drone, a robot, a handheld communication device, a hospital device, a gaming device or system, an entertainment system, an in-vehicle computer system, an embedded system controller, a remote control, an appliance, a consumer electronic device, a workstation, an edge device, any combination of these described devices, or any other suitable device.

[0260] The present disclosure may be described in the general context of machine-usable instructions or computer codes executed by a computer or other machine such as a personal digital assistant or other handheld device, including computer-executable instructions such as program modules. Typically, program modules including routines, programs, objects, components, data structures, etc. refer to codes that perform specific tasks or implement specific abstract data types. The present disclosure may be practiced in a variety of system configurations, including handheld devices, consumer electronics, general-purpose computers, more specialized computing devices, etc. The present disclosure may also be practiced in a distributed computing environment in which tasks are performed by remote processing devices linked through a communication network.

[0261] As used herein, the description of "and / or" about two or more elements should be interpreted as referring to only one element or a combination of elements. For example, "element A, element B and / or element C" may include only element A, only element B, only element C, element A and element B, element A and element C, element B and element C, or elements A, B and C. In addition, "at least one of element A or element B" may include at least one of element A, at least one of element B, or at least one of element A and at least one of element B. Further, "at least one of element A and element B" may include at least one of element A, at least one of element B, or at least one of element A and at least one of element B. In addition, the use of the term "based on" should not be interpreted as "based only on" or "based only on". On the contrary, the first element "based on" the second element includes an instance in which the first element is based on the second element but may also be based on one or more additional elements.

[0262] The subject matter of the present disclosure is described in detail herein to meet statutory requirements. However, the description itself is not intended to limit the scope of the present disclosure. On the contrary, the inventors have contemplated that the claimed subject matter may also be embodied in other ways to include steps different from the steps described herein in conjunction with other current or future technologies or combinations of similar steps. Moreover, although the terms "step" and / or "block" may be used herein to imply different elements of the method employed, these terms should not be interpreted as implying any particular order among or between the various steps disclosed herein, unless the order of the steps is explicitly described.

Claims

1. A method comprising: The sensor data corresponding to the operating area of ​​the machine is analyzed using a variety of analysis techniques including: at least two analysis techniques having computational diversity by performing different types of computational analysis on the sensor data; and By at least two analysis techniques with implementation diversity implemented on different types of computing platforms; determining detection of an object within the operating area based at least on the analyzing; and One or more operations of the machine are controlled based at least on the detection.

2. The method according to claim 1, wherein: The sensor data includes stereoscopic image data.

3. The method according to claim 1, wherein: The plurality of analysis techniques include a plurality of depth techniques that differ in how depth from one or more reference points is determined.

4. The method according to claim 3, wherein: The multiple deep technologies include at least semi-global matching SGM, Bi3D and efficient semi-supervised deep ESS.

5. The method according to claim 1, wherein: Different types of computing platforms include one or more of the following: central processing unit CPU, graphics processing unit GPU, optical flow accelerator OFA, deep learning accelerator DLA, programmable vision accelerator PVA, video input VI controller, image signal processor ISP or video image synthesizer VIC engine.

6. The method according to claim 1, wherein: Determining the detection of the object comprises: determining a plurality of individual test results corresponding to respective ones of the plurality of analysis techniques; and An overall test result is determined based at least on the plurality of individual test results.

7. The method according to claim 6, wherein: Determining the overall test result includes: The overall test result is determined by combining the plurality of individual test results, wherein the one or more individual test results are weighted for the combination based at least on respective confidence levels associated with one or more of the plurality of individual test results.

8. The method according to claim 6, wherein: Determining the overall test result includes: The overall detection result is determined based on whether a threshold number of the plurality of individual detection results is indicated to be detected.

9. The method according to claim 8, wherein: The threshold amount is based at least on a target safety level associated with the operating area.

10. A system comprising: One or more processors, the one or more processors causing execution of operations including: analyzing sensor data corresponding to the detection area using a plurality of analysis techniques, the plurality of analysis techniques comprising computational diversity and implementational diversity; determining detection of an object within the detection area based at least on the analyzing; as well as One or more operations are performed based at least on the detecting.

11. The system according to claim 10, wherein: The sensor data includes stereoscopic image data.

12. The system according to claim 10, wherein: The plurality of analysis techniques include a plurality of depth techniques that differ in how depth from one or more reference points is determined.

13. The system according to claim 12, wherein: The multiple deep technologies include at least semi-global matching SGM, Bi3D and efficient semi-supervised deep ESS.

14. The system according to claim 10, wherein: The multiple analysis technologies are implemented on multiple computing platforms, which include one or more of the following: a central processing unit CPU, a graphics processing unit GPU, an optical flow accelerator OFA, a deep learning accelerator DLA, a programmable vision accelerator PVA, a video input VI controller, an image signal processor ISP or a video image synthesizer VIC engine.

15. The system according to claim 10, wherein: Determining the detection of the object comprises: determining a plurality of individual test results corresponding to respective ones of the plurality of analysis techniques; and An overall test result is determined based at least on the plurality of individual test results.

16. The system of claim 15, wherein: Determining the overall test result includes: The overall test result is determined by combining the plurality of individual test results, wherein the one or more individual test results are weighted for the combination based at least on respective confidence levels associated with one or more of the plurality of individual test results.

17. The system of claim 15, wherein: Determining the overall test result includes: The overall detection result is determined based on whether a threshold number of the plurality of individual detection results is indicated to be detected.

18. The system of claim 17, wherein: The threshold amount is based at least on a target safety level associated with the operating area.

19. The system of claim 10, wherein: The system is included in at least one of the following: control systems for autonomous or semi-autonomous machines; Perception systems for autonomous or semi-autonomous machines; A system for performing simulation operations; Systems for performing digital twin operations; A system for performing light transport simulations; A system for performing collaborative content creation of 3D assets; Systems for performing deep learning operations; A system for presenting at least one of augmented reality content, virtual reality content, or mixed reality content; A system for hosting one or more real-time streaming applications; Systems implemented using edge devices; Systems implemented using robots; Systems for performing conversational AI operations; A system implementing one or more large language models (LLMs); Systems for generating synthetic data; A system comprising one or more virtual machines VM; A system implemented at least in part in a data center; or A system implemented at least in part using cloud computing resources.

20. A system comprising: Processing circuitry for performing one or more operations associated with a machine based at least on a final detection result determined based at least on two or more individual detection results computed using two or more different algorithms executed on two or more different hardware components of the machine.

Images