Deep learning-based environment modeling for vehicle environment visualization

Abstract

The invention relates to deep learning based environment modeling for vehicle environment visualization. In various examples, an environment visualization pipeline may determine whether to generate or otherwise enable visualization using an environment modeling pipeline that models the environment as a 3D bowl or using an environment modeling pipeline that models the environment with some other 3D representation (e.g., detected 3D surface topology). Determination may be made based on various factors, such as self-machine status (e.g., indicative of one or more detected features of a specified operating scene, proximity to a detected object, speed of the self-machine, etc.), estimated image quality of a corresponding environmental visualization, and / or other factors. Thus, the environment around self-machines (e.g., vehicles, robots, and / or other types of objects) may be visualized in systems such as parking visualization systems, look-around systems, and / or other systems.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of U.S. Provisional Application No. 63 / 566,129, filed on March 15, 2024, and U.S. Provisional Application No. 63 / 565,885, filed on March 15, 2024. The contents of each of the foregoing applications are incorporated herein by reference in their entirety. Background Art

[0003] Surround view systems provide vehicle occupants with a visual representation of the area surrounding the vehicle. For the driver, surround view systems provide the driver with the ability to see nearby areas of the environment, including blind spots where the driver's view is obscured by components of the vehicle or other objects in the environment, without having to reposition their view (e.g., turning their head, leaving the driver's seat, leaning in a certain direction, etc.). This visualization can assist and facilitate various driving maneuvers, such as smoothly entering or exiting a parking space without colliding with vulnerable road users (such as pedestrians) or objects (such as curbs or other vehicles). An increasing number of vehicles, particularly luxury brands or newer models, are being produced equipped with surround view systems. Surround view systems in existing vehicles typically use fisheye cameras (typically mounted on the front, left, rear, and right sides of the vehicle body) to perceive the surrounding area from multiple directions. In some technologies, frames from various cameras are stitched together using camera parameter alignment and overlapping areas are combined using blending techniques to provide a top-down 360° surround view visualization.

[0004] There are a variety of techniques for generating visualizations of the surrounding environment, some of which rely on different assumptions about the geometry of the surrounding environment. In reality, no single model is perfect, so choosing a particular technique to generate surround view visualizations can be considered a design choice with various advantages and disadvantages. For example, in some existing surround view systems, two-dimensional (2D) images are used to approximate a three-dimensional (3D) visual representation of the vehicle's surroundings by modeling the geometry of the vehicle's surroundings as a virtual 3D bowl. The 3D bowl typically includes a flat, circular ground plane for the bowl's inner portion, which is connected to an outer bowl, which is represented as a curved surface that rises from the ground plane to a certain height or has a slope that increases proportionally with the distance from the bowl's center. Therefore, some conventional systems project (e.g., stitch) images onto the 3D bowl, render a view of the projected image data onto the 3D bowl from the perspective of a virtual camera, and present the rendered view on a monitor visible to a vehicle occupant or operator (e.g., driver). However, the projection and / or stitching process introduces various artifacts, including geometric distortions (e.g., size or shape misalignment), texture distortions (e.g., blurring, ghosting, object disappearance, object distortion), and color distortions. Because these artifacts may block or omit useful visual information and are generally distracting to the driver, they may interfere with safe vehicle operation in certain scenarios.

[0005] Therefore, there is a need for improved visualization techniques that reduce visual artifacts, better represent useful visual information, and / or otherwise improve the visual quality of the resulting images. Summary of the Invention

[0006] Embodiments of the present disclosure relate to environment modeling for vehicular environment visualization. Systems and methods are disclosed for determining whether to implement different environment modeling pipelines to generate visualizations. Compared to conventional systems, such as those described above, systems and methods are disclosed for determining whether to generate or otherwise implement a visualization using an environment modeling pipeline that models the surrounding environment as a 3D bowl or using an environment modeling pipeline that models the surrounding environment using some other 3D representation (e.g., a detected 3D surface topology). The determination can be made based on various factors, such as the state of the ego machine (e.g., one or more detected features indicating proximity to a detected object for a specified operating scenario (e.g., a parking scenario, an off-road scenario, etc.), the speed of the ego machine, etc.), an estimated or predicted image quality of the corresponding environment visualization, and / or other factors. The present technology can be used to visualize the environment surrounding an ego machine (e.g., a vehicle, a robot, and / or other types of objects) in systems such as parking visualization systems, surround view systems, and / or other systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] The system and method for environment modeling for vehicle environment visualization are described in detail below with reference to the accompanying drawings, wherein:

[0008] Figure 1 is an example environment visualization pipeline according to some embodiments of the present disclosure;

[0009] Figure 2 is an example environment model switching state machine according to some embodiments of the present disclosure;

[0010] Figure 3A is an example frame representing a rear camera view according to some embodiments of the present disclosure;

[0011] Figure 3B is an example frame representing a front camera view according to some embodiments of the present disclosure;

[0012] Figure 3C is an example frame representing a left camera view according to some embodiments of the present disclosure;

[0013] Figure 3D is an example frame representing a right camera view according to some embodiments of the present disclosure;

[0014] Figure 3E is an example bowl visualization from a third-person perspective according to some embodiments of the present disclosure, and Figure 3F It is from Figure 3E Example surface topology visualization of the third-person view direction;

[0015] Figure 3G is an example bowl visualization from a bay view according to some embodiments of the present disclosure, and Figure 3H It is from Figure 3G Example surface topology visualization of the viewing direction of the cabin view;

[0016] Figure 3I is an example bowl visualization of a viewing direction from a rim-protected view according to some embodiments of the present disclosure, and Figure 3J It is from Figure 3I Example surface topology visualization of the viewing direction of the edge-preserving view;

[0017] Figure 3K is an example bowl visualization of a viewing direction from a cabin view in a parking space according to some embodiments of the present disclosure, and Figure 3L It is from Figure 3K Example surface topology visualization of the viewing direction of a cabin view in a parking space;

[0018] Figure 3Mis another example surface topology visualization from a third-person viewpoint according to some embodiments of the present disclosure;

[0019] Figure 3N is an example surface topology visualization from a top-down view according to some embodiments of the present disclosure;

[0020] Figure 4 is a flow chart illustrating a method for switching between visualizations according to some embodiments of the present disclosure;

[0021] Figure 5A is an illustration of an example autonomous vehicle according to some embodiments of the present disclosure;

[0022] Figure 5B According to some embodiments of the present disclosure Figure 5A Examples of camera positions and fields of view for autonomous vehicles;

[0023] Figure 5C According to some embodiments of the present disclosure Figure 5A a block diagram of an example system architecture for an example autonomous vehicle;

[0024] Figure 5D is a cloud-based server and Figure 5A System diagram of an example of communication between autonomous vehicles;

[0025] Figure 6 is a block diagram of an example computing device suitable for implementing some embodiments of the present disclosure; and

[0026] Figure 7 is a block diagram of an example data center suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION

[0027] Systems and methods are disclosed relating to environment modeling for vehicle environment visualization. Although the present disclosure may relate to an example autonomous or semi-autonomous vehicle or machine 500 (referred to herein alternatively as "vehicle 500" or "ego-machine 500"), the examples relate to Figures 5A-5DThe present disclosure may be described with respect to surround view visualization for a vehicle, but this is not intended to be limiting. For example, 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, manned and unmanned robots or robotic platforms, warehouse vehicles, off-road vehicles, vehicles coupled to one or more trailers, flying boats, ships, space shuttles, emergency response vehicles, motorcycles, electric or motorized bicycles, airplanes, construction vehicles, trains, underwater vehicles, remotely controlled vehicles (e.g., drones), and / or other vehicle types. Furthermore, while the present disclosure may be described with respect to surround view visualization for a vehicle, this is not intended to be limiting, and the systems and methods described herein may be used in augmented reality, virtual reality, mixed reality, robotics, safety and surveillance, autonomous or semi-autonomous machine applications, and / or any other technology field in which vehicle environment visualization may be used.

[0028] Example systems and methods are disclosed that determine whether to generate a visualization using an environment modeling pipeline that models the surrounding environment as a 3D bowl or using an environment modeling pipeline that models the surrounding environment using some other 3D representation (e.g., detected 3D surface topology or 3D point cloud). This determination can be made based on various factors, such as the state of the ego machine (e.g., proximity to detected pedestrians, one or more detected features indicative of a specified operating scenario, such as a detected parking scenario in which the ego machine is operating or predicted to begin operating), estimated image quality of one or more source images and / or corresponding environment visualizations, and / or other factors. The present technology can be used to visualize the environment surrounding an ego machine (e.g., a vehicle, robot, and / or other type of object) in systems such as parking visualization systems, surround view systems, and / or other systems.

[0029] At a high level, an environment visualization pipeline (e.g., of a self-machine) can support using multiple environment modeling techniques to generate different types of visualizations of a surrounding environment in different scenarios, e.g., based on the strengths of using each of the multiple environment modeling techniques in different scenarios. For example, the environment visualization pipeline can support an environment modeling pipeline that models the surrounding environment as a detected 3D surface topology when generating a visualization (e.g., in certain scenarios, such as when the speed of self-motion is low, objects are detected within a threshold distance, and / or other scenarios), and / or an environment modeling pipeline that models the surrounding environment as a 3D bowl when generating a visualization (e.g., in other scenarios, such as when the speed of self-motion is high, objects are not detected within a threshold distance, and / or other scenarios).

[0030] Taking a 3D bowl model as an example, depending on the implementation and / or scenario, any known technique can be used to model the surrounding environment as a 3D bowl and / or generate a visualization of a 3D bowl modeled around the surrounding environment (e.g., a bowl visualization). Compared to environment modeling techniques that model the surrounding environment using other means (e.g., detected 3D surface topology), many implementations of bowl visualizations have relatively low processing requirements, allowing them to be generated more quickly and computationally cheaply compared to other techniques. Furthermore, by projecting or texturing image data onto the 3D bowl model, the image data can be effectively stretched onto the bowl, reducing or eliminating gaps caused by occlusions typically produced by other techniques. While closing these gaps may be considered beneficial, the stretching effect may result in exaggerated scaling. In some cases, the benefit of avoiding gaps in the resulting visualization may be considered worth the cost of some visual artifacts. However, in certain implementations and / or scenarios (e.g., scenarios where objects or obstacles (such as vehicles or pedestrians) are detected within a threshold distance), it may be desirable to use a different visualization technique.

[0031] Thus, in some implementations and / or scenarios, the environment can be modeled using the detected 3D surface topology to generate a visualization of the environment. For example, as described in more detail in U.S. Provisional Application No. 63 / 565,885, the contents of which are incorporated herein by reference, the environment surrounding the self-machine can be visualized by extracting one or more depth maps from sensor data (e.g., image data), converting the depth maps into a detected 3D surface topology of the surrounding environment, and / or texturing the detected 3D surface topology with the sensor data. The detected 3D surface topology can model the geometry of the surrounding environment as a 3D surface using a (e.g., truncated) signed distance function (SDF) that encodes the distance between each voxel in the 3D mesh and the detected 3D surface topology represented by the one or more depth maps. A 3D surface representation (e.g., a 3D mesh) of the detected 3D surface can be extracted from the detected 3D surface topology and can be smoothed using any known technique. In this way, sensor data (e.g., image data) can be back-projected onto a 3D surface representing a detected 3D surface topology (e.g., a 3D surface mesh) using corresponding depth values ​​in each back-projection (e.g., represented by an unsmoothed 3D surface representation, rendered using unstructured light map rendering, etc.) to generate a textured 3D surface. In this way, surface topology visualization can be generated by rendering a view of the textured 3D surface, or by generating an untextured view of the detected 3D surface topology (e.g., a 2D image with placeholder pixels), projecting the depth values ​​into the 2D view (e.g., generating a corresponding depth map), and texturing the 2D view using the corresponding (e.g., per-pixel) depth values.

[0032] In some implementations and / or scenarios, the detected 3D surface topology can be represented as a (e.g., characterized) point cloud, and the detected 3D surface topology can be used to model the environment to generate a visualization of the environment (e.g., a surface topology visualization). For example, sensor data (e.g., images from different cameras of the self-machine) can be applied to one or more neural networks to extract corresponding depth maps, and the depth maps can be characterized (e.g., by assigning a color of a corresponding pixel from the input image to each pixel in the depth map). The characterized depth map can be uplifted to generate and / or characterize a sparse point cloud representing the detected 3D surface topology of the surrounding environment (e.g., where each point of the characterized point cloud carries a color value of a corresponding pixel from the image data). Thus, a surface topology visualization can be generated by rendering a view of the characterized point cloud, and / or applying the resulting (e.g., sparse) image to a neural renderer to densify and generate a photo-realistic image.

[0033] Compared to environment visualization techniques that model the surrounding environment as a 3D bowl, modeling the surrounding environment as detected 3D surface topology can provide a more realistic view of the surrounding environment in some scenarios, and many visual artifacts that may appear in the bowl visualization (e.g., scale expansion, object disappearance, etc.) are less likely to appear in the corresponding surface topology visualization. For example, nearby objects (e.g., moving obstacles or their subclasses such as vehicles, pedestrians, and / or others) are less likely to be distorted, blurred, or omitted in the surface topology visualization. However, in some scenarios, the surface topology visualization may be rendered in a view that includes areas of the environment that are not observed in the source sensor (e.g., image) data. Therefore, rendering the surface topology visualization may cause artifacts in the surface topology visualization, such as gaps (e.g., occlusions) where there is no image data (e.g., or historical image data) representing a particular area. Furthermore, in some implementations (e.g., those using a neural renderer), running a visualization pipeline that models the surrounding environment as detected 3D surface topology may be computationally more expensive than alternative supporting methods (e.g., using a 3D bowl model) and may introduce visual artifacts into the surface topology visualization in some scenarios (e.g., when the image cannot be rendered quickly enough). In this regard, in some implementations and / or scenarios, it may be desirable to model the surrounding environment as detected 3D surface topology when a measure of detected image quality of the surface topology visualization (e.g., a measure of the number or size of occluded regions) is above a specified threshold.

[0034] Thus, to capture the benefits of certain environment modeling techniques in certain scenarios, in some embodiments, a state machine may be used to select, enable, or switch between supported environment modeling techniques, which may be based on any number of detectable factors, such as the state of the ego machine (e.g., proximity to detected pedestrians; proximity to detected objects; detected operating scenarios, such as designated parking scenarios or off-road scenarios; speed of ego motion; and / or other); estimated or predicted image quality of the corresponding visualization (e.g., which may be based on the confidence level of the corresponding depth estimate used to generate the visualization); selections and / or preferences of the operator of the ego machine; and / or other. Some implementations may provide different selectable visualization options supported by different environment modeling techniques. Additionally or alternatively, some visualizations or selectable visualization options may be enabled or disabled based on the state of the ego machine or the estimated and / or predicted image quality of the corresponding visualization.

[0035] Typically, visualizations generated by modeling the surrounding environment using a 3D bowl will be very clear, will not have holes, and in many implementations will be computationally cheaper to run than other supported modeling techniques. Therefore, in some embodiments, one or more visualizations that rely on the 3D bowl model can be enabled (e.g., displayed or made available for selection) as a baseline or default. In an example implementation, one or more detectable features can be used to determine whether to disable one or more visualizations that rely on the 3D bowl model and / or enable one or more visualizations that rely on a different model of the surrounding environment (e.g., a detected 3D surface topology) and / or one or more visualizations that rely on neural rendering. Example features that may be used to enable one or more visualizations that rely on detected 3D surface topology and / or neural rendering include: detected close objects of a specified class (e.g., vehicles) within a threshold proximity, detected close objects of an unsupported class (e.g., trees or other objects that supported object detectors may not be designed to detect) within a threshold proximity, such as by using detected depth estimates, detected ego-machine speed below a threshold (e.g., representative of a parking scenario), detected operator intent (e.g., representative of a parking scenario), confidence in the prediction of the extracted depth below a threshold (e.g., which may indicate that a detected 3D surface topology that relies on an extracted depth below a threshold will be less accurate), detected operation scenarios (e.g., in the 3D bowl model false assuming that the detected 3D surface topology is enabled in off-road driving in some implementations where the ground is flat, and in parking scenarios in some implementations where better visualization can be expected to be generated at lower speeds), the amount of detected or predicted occlusion (e.g., this may indicate that stretching image data to cover a bowl visualization of a hole during texturing will provide better visualization), the selected or applicable viewport (e.g., virtual camera placement orientation) or corresponding type of visualization, the availability of sensor data from a previous time slice (e.g., in some embodiments that generate visualizations such as top-down views or other views that rely on detected 3D surface topology generated using sensor data from a previous time slice), operator selection, and / or other factors (e.g., such as a combination thereof). Example features that can be used to enable visualization of one or more simulations (e.g., with simulated representations of portions of the environment (e.g., nearby lanes) and / or detected objects (e.g., nearby vehicles or pedestrians)) include: detecting objects (e.g., pedestrians) between the ego machine and the virtual camera (e.g., such that a mirror image of the object can be rendered due to texturing), detecting partially occluded objects (e.g., rendering a 3D asset that simulates an object of the corresponding class to represent the occluded object rather than rendering the partially observed object), ego machine speed above a certain threshold (e.g., representing highway speed), and / or other (e.g., such as combinations thereof).

[0036] Taking an example implementation of using detected speed and detected proximity to detected objects to enable or disable one or more visualizations relying on a 3D bowl model (bowl visualization) and / or relying on detected 3D surface topology (e.g., surface topology visualization), when the detected vehicle speed is below a specified speed threshold (e.g., less than 8-10 kilometers per hour (km / hr)) and the nearest detected object is closer to the vehicle than a certain threshold proximity (e.g., less than 3 meters), the surface topology visualization can be enabled (e.g., for display or user selection). Conversely, when the vehicle speed is above the speed threshold and no objects are detected within the threshold proximity, the bowl visualization can be enabled. In some embodiments, when the distance of the detected object from the ego machine is less than a threshold distance (e.g., less than 1 meter), raw image data (e.g., an image / video feed generated using the ego machine's camera) is enabled. In some embodiments, when the vehicle's speed is above a speed threshold (e.g., 35 km / hr), a simulated visualization (e.g., a visualization having simulated representations of nearby lanes and / or simulated assets corresponding to detected objects (e.g., cars, trucks, motorcycles, pedestrians, and / or others)) is enabled for display.

[0037] In an example data stream, input images from one or more cameras of an ego machine (e.g., a vehicle) can be used to visualize the environment surrounding the ego machine using multiple supported environment modeling techniques and / or view directions. For example, a display visible to an operator and / or occupant of the ego machine can present different visualizations of the environment generated from the perspective of virtual cameras positioned and / or oriented in different viewing directions (e.g., side view, top-down view, forward view, rear view, view from below the ego machine, and / or other), and / or can present some indication of which views are available for selection. In certain detected scenes, visualizations from one or more different viewing directions can be generated (or made available for selection) as surface topology visualizations, and in certain detected scenes, visualizations from one or more different viewing directions can be generated (or made available for selection) as bowl visualizations. In an example embodiment, a bowl visualization (e.g., a side view generated using an environment modeling pipeline that models the surrounding environment as a 3D bowl) can be automatically displayed or made available for selection as the default visualization for one or more viewing directions (e.g., or for all available viewing directions). In some implementations, a bowl visualization is used as a default visualization because it is computationally inexpensive to generate compared to other supported techniques and should generally produce high-quality visualizations in most scenarios. Upon detection of one or more specified scenarios (e.g., a parking scenario, an estimated or predicted image quality of a surface topology visualization above a threshold), a surface topology visualization (e.g., a view generated using an environment modeling pipeline that models the surrounding environment as a detected 3D surface topology) may be automatically displayed or made selectable. In this regard, in certain scenarios (e.g., where the surface topology visualization may provide a better visualization), the surface topology visualization may be enabled. For example, in scenarios where a pedestrian is detected approaching the ego-machine (e.g., within one to two meters (m)), the surface topology visualization may be enabled (e.g., because the bowl visualization may represent the pedestrian with deformations, or not at all).

[0038] Examples of scenarios in which surface topology visualizations may be generated, displayed, and / or selectable include: detection of an object of a specified class (e.g., a pedestrian); detection of an object (e.g., a class of objects such as cars, objects exceeding a threshold size (e.g., a curb exceeding a threshold height), and / or other) within a threshold distance of the ego machine (e.g., within two or three meters of the ego machine); detection of an off-road scenario (e.g., based on an inferred enabled driving mode, e.g., based on detection of an object located on the driving surface and having a detected height within a threshold height range, which may indicate a large obstacle that may be driven over); detection or prediction of a parking scenario, such as entering or exiting a parking space, parallel parking, etc. (e.g., based on operator input, such as indicating an automated parking maneuver or selecting a parking space on a display), detection of a speed of ego-motion below a specified threshold (e.g., 8-10 km / hr), predicted driver intent to park, detected trajectory, detection of a vacant parking space, detection of one or more driver actions (such as signaling or shifting gears), detection of the ego-machine backing up, and / or some combination thereof); some measure of estimated or predicted image quality of the resulting visualization (e.g., less than a certain number, amount, or coverage of occlusions, at least some threshold confidence level of extracted depth values ​​represented by a 3D model of the surrounding environment); selections and / or preferences of the operator of the ego-machine (e.g., for surface topology visualization); and / or other (e.g., such as a combination thereof).

[0039] Examples of scenarios in which a bowl visualization may be generated, displayed, and / or selectable include: by default, detected objects (e.g., or certain types of objects, such as cars or objects larger than a threshold size) exceeding a certain threshold distance from the ego-machine (e.g., two or three meters); detected driving on a road (e.g., based on an enabled driving mode, such as inferred based on no objects detected above a threshold height on the driving surface); detected speed of ego-motion above a specified threshold (e.g., 8-10 km / hr); some indication that the surface topology visualization may not provide improved visualization (e.g., exceeding a certain number, amount, or coverage of occlusions, below a certain threshold confidence level of extracted depth values); selections and / or preferences of an operator of the ego-machine (e.g., for bowl visualization); and / or other (e.g., such as a combination thereof). In some embodiments, the bowl visualization is always selectable for display.

[0040] Any suitable technique may be used to determine the state of the self-machine (e.g., detected proximity to a detected object; a detected operating scenario in which the self-machine is operating or predicted to begin operating; the speed of the self-motion; and / or other). For example, in some embodiments, one or more sensors of the self-machine (e.g., a camera, a radar (RADAR) sensor, a laser radar (LiRAR) sensor, an ultrasonic sensor) may be used to generate sensor data representing the environment surrounding the self-machine, and any known object detection and / or tracking technique may be used to detect approaching objects (e.g., moving obstacles or subclasses thereof, such as vehicles, pedestrians, and / or other objects) and estimate their positions in the environment (e.g., relative to the self-machine) based on the sensor data. In some embodiments, a camera of the self-machine is used to generate detected or predicted depth values ​​corresponding to an input image to determine the proximity of the self-machine to a detected object represented in the image. In some embodiments, one or more operating scenarios may be specified, and any known technique may be used to detect one or more features indicative of a specified operating scenario. For example, detectable features that may indicate a parking scenario include the presence of stop lines or parking spaces, nearby stationary vehicles, proximity to buildings or structures, reduced traffic flow, reduced speed or deceleration of the self-machine, detected operator intent (e.g., based on eye movements, gaze direction, facial expressions, steering patterns, and / or other), and / or other detectable features that may indicate an off-road driving scenario include the absence of road markings, rough terrain, uneven surfaces, vegetation, obstacles (e.g., rocks or trees), engagement of an off-road driving mode, suspension adjustments, and / or other detectable features.

[0041] Any known technique and / or any suitable metric may be used to determine the estimated or predicted image quality of the visualization (e.g., the number, amount, and / or coverage of occlusions; the confidence level of the corresponding depth estimate; and / or other). For example, if a measure of the number, size, or (e.g., percentage) coverage of occluded regions in the image data used for the textured visualization is above a specified threshold, the surface topology visualization may be disabled. In some embodiments, a simulated representation of a detected object (e.g., a 3D model representing a typical pedestrian) may be inserted into the visualization in place of the real image data in various circumstances (e.g., based on a determination that the detected object is occluded, located between the ego machine and the virtual camera, and / or other circumstances). For example, if an object (e.g., a pedestrian) is detected and the object can be rendered in a photorealistic manner (e.g., the quality of the object rendering is above a threshold), the object may be rendered in a photorealistic manner (e.g., based on real image data showing the pedestrian's face). If the object cannot be rendered in a photorealistic manner (e.g., the quality of the object rendering is below a threshold, e.g., because only half of the pedestrian is depicted in the input image), a simulated representation of the object may be inserted into the visualization instead of rendering the object based on the real image data. For example, a simulated asset (e.g., a 3D mannequin of a person, a 3D model of a car, or other simulated model) can be inserted into a 3D model of an environment or a corresponding 2D visualization to replace an object in the visualization. As another example, content generated by a generative artificial intelligence model can be inserted into a 3D model of an environment or a corresponding 2D visualization to replace an object in the visualization. The quality of the object rendering (e.g., the confidence that the object is rendered in a photorealistic manner) can be determined by any known technique and / or using any suitable metric.

[0042] In some embodiments, a particular viewing direction of the surface topology visualization can be enabled for display in certain scenarios before other viewing directions of the surface topology visualization are enabled for display. For example, after receiving a certain amount of historical image data, a top-down view of the surface topology visualization of the ego machine can be enabled for display to provide a visualization with above-threshold image quality (e.g., below a threshold amount of occlusion in the visualization). As another example, an edge-view visualization of the surface topology visualization of the ego machine can be enabled for display after detecting a specified operating scenario or after selection by an operator of the ego machine, because the edge-view visualization can generate a visualization with above-threshold image quality without requiring historical image data.

[0043] Thus, the techniques described herein can be used to determine whether to generate a bowl visualization and / or a surface topology visualization based on detectable factors (e.g., one or more features of the ego-machine state, some measure of estimated or predicted image quality of the visualization, and / or other factors). Determining whether to generate a bowl visualization and / or a surface topology visualization helps provide optimal visualization (e.g., reducing or avoiding visual artifacts such as zooming, object disappearance, and occlusion) in different detection scenarios, thereby improving the quality of the visualization and the operator's ability to safely navigate through the environment. Furthermore, determining that surface topology visualization is selectively enabled in certain scenarios (e.g., using a neural renderer) reduces the computational expense that would otherwise be required to generate that type of visualization more frequently. Thus, the techniques described herein can be used to reduce visual artifacts, enhance safe operation of a vehicle, and / or increase processing speed by selectively enabling different environment modeling techniques to support the generation of visualizations in different scenarios.

[0044] Example environment visualization pipeline

[0045] refer to Figure 1 , Figure 1 is an example environment visualization pipeline 100 according to some embodiments of the present disclosure. It should be understood that this arrangement and other arrangements described herein are presented by way of example only. Other arrangements and elements (e.g., machines, interfaces, functions, sequences, functional groupings, etc.) may be used in addition to or in place of the arrangements and elements shown, and some elements may be omitted entirely. In addition, many of the elements described herein are functional entities that may be implemented as discrete or distributed components or in conjunction with other components and in any suitable combination and location. The various functions performed by the entities described herein may be performed by hardware, firmware, and / or software. For example, the various functions may be performed by a processor executing instructions stored in a memory. In some embodiments, the systems, methods, and processes described herein may be implemented using hardware, firmware, and / or software. Figures 5A-5D Example of autonomous vehicle 500, Figure 6 The example computing device 600 and / or Figure 7 The example data center 700 may be performed using components, features, and / or functions similar to the components, features, and / or functions of the example data center 700.

[0046] As a high-level overview, the environment visualization pipeline 100 may be incorporated into or otherwise associated with an ego machine (e.g., as a surround view system for the ego machine), e.g. Figures 5A-5Dego vehicle 500 in FIG. The environment visualization pipeline 100 may include any number and type of sensors 101 (e.g., one or more cameras) that may be used to generate sensor data (e.g., image data 105) representing the surrounding environment. The environment visualization pipeline 100 may include an environment visualization generator 111 that uses the image data 105 to generate an environment visualization representing the surrounding environment and / or provide it to a display 180 visible to an occupant or operator (e.g., a driver or passenger) of the ego machine. Additionally or alternatively, the environment visualization pipeline 111 may stream the environment visualization, sensor data, and / or some other representation of the environment within and / or around the ego machine to a remote location.

[0047] In an example embodiment, a self-machine (e.g., Figures 5A-5D The autonomous vehicle 500 in FIG. 5 is equipped with any number and type of sensors 101 (e.g., one or more cameras, such as a fisheye camera), and the sensors 101 can be used to generate overlapping sensor data frames (e.g., overlapping image data) for each time slice. In general, any suitable sensor can be used, such as Figure 5A One or more of the stereo camera 568, wide-angle camera 570 (e.g., fisheye camera), infrared camera 572, surround camera 574 (e.g., 360° camera), and / or long-range and / or mid-range camera 598 of the vehicle 500 can be aligned. Any known technique can be used to align the sensor data. In the example configuration, four fisheye cameras are mounted on the front, left, rear, and right sides of the vehicle, where surrounding video is continuously captured (e.g., once per time slice). Figure 3A An example frame of image data representing a rear camera view 300A is shown. Figure 3B An example frame of image data representing a front camera view 300B is shown. Figure 3C An example frame of image data representing left camera view 300C is shown. Figure 3D An example image data frame representing right camera view 300D is shown. The vehicle's ego-motion can be generated using any known technique (e.g., via velocity detector 143) and synchronized with the timestamp of the video frame (e.g., image). For example, absolute or relative ego-motion data (e.g., position, orientation, position and rotational velocity, position and rotational acceleration) can be determined using a vehicle speed sensor, gyroscope, accelerometer, inertial measurement unit (IMU), and / or other devices via velocity detector 143.

[0048] In some embodiments, the environment modeling pipeline 113 of the environment visualization generator 111 (e.g., of the ego machine) can support using multiple environment modeling techniques to generate different types of visualizations of the surrounding environment in different scenarios (e.g., as determined by the environment modeling pipeline selection component 120), for example, based on the benefits of using each of the multiple environment modeling techniques in different scenarios. For example, the environment visualization generator 111 can support an environment modeling pipeline (e.g., provided by the detected 3D surface topology modeling pipeline 117) that models the surrounding environment as a detected 3D surface topology when generating visualizations (e.g., in some scenarios, such as low ego-motion speed, detected objects within a threshold distance, and / or other scenarios) and / or an environment modeling pipeline (e.g., provided by the 3D bowl modeling pipeline 115) that models the surrounding environment as a 3D bowl when generating visualizations in other scenarios (e.g., in other scenarios, such as high ego-motion speed, no detected objects within a threshold distance, and / or other scenarios). Visualizations can be generated (e.g., by view generator 170) from different viewing directions using a corresponding environment modeling pipeline selected for the detected scene (e.g., 2D images of the environment model can be generated from the perspective of virtual cameras positioned and oriented in different ways in the 3D scene, thereby providing various perspectives, such as a side view relative to the ego machine, a top-down view, a forward view relative to the ego machine, a rearward view relative to the ego machine, a view from below the ego machine, etc.). Figures 3E-3N Examples of different types of visualizations representing different viewing directions and an example environment modeling pipeline are shown in , as explained in more detail below.

[0049] In some implementations and / or scenarios, the environment modeling pipeline 113 can implement a 3D bowl modeling pipeline 115 to model the surrounding environment as a 3D bowl. Generally, the 3D bowl modeling pipeline 115 can use any known techniques to model the surrounding environment as a 3D bowl, the texture mapping component 160 can use corresponding sensor data (e.g., image data 105 from different cameras of the ego machine) to texture or otherwise apply graphical representations of visual elements (e.g., colors, images, designs, patterns, etc.) to the 3D bowl, and the view generator 170 can render a view of the textured 3D bowl. Figure 3E 、 Figure 3G 、 Figure 3I and Figure 3K An example of a bowl visualization that can be generated by modeling the surrounding environment as a 3D bowl is shown in FIG.

[0050] In some implementations and / or scenarios, the environment modeling pipeline 113 can implement a detected 3D surface topology modeling pipeline 117 to model the surrounding environment as a detected 3D surface topology. For example, as described in more detail in U.S. Provisional Application No. 63 / 565,885, the contents of which are incorporated herein by reference, to visualize the environment around the self-machine, the detected 3D surface topology modeling pipeline 117 can extract one or more depth maps from the sensor data (e.g., image data 105) and convert the depth maps into a detected 3D surface topology of the surrounding environment, and the texture mapping component 160 can use the sensor data to texture the detected 3D surface topology or otherwise apply a graphical representation of a visual element (e.g., color, image, design, pattern, etc.) to the detected 3D surface topology. The detected 3D surface topology generated by the detected 3D surface topology modeling pipeline 117 can model the geometry of the surrounding environment as a 3D surface using a (e.g., truncated) SDF that encodes the distance between each voxel in the 3D mesh and the 3D surface topology extracted from one or more depth maps. The detected 3D surface topology modeling pipeline 117 can extract a 3D surface representation (e.g., a 3D mesh) of the detected 3D surface from the detected 3D surface topology and can smooth the detected 3D surface topology and / or the 3D surface representation using any known technique. Accordingly, the texture mapping component 160 can back-project sensor data (e.g., image data 105) onto a 3D surface representing the detected 3D surface topology (e.g., a 3D surface mesh) using corresponding depth values ​​in each back-projection (e.g., derived from a smoothed or unsmoothed 3D surface representation, rendered using an unstructured light map, etc.) to generate a textured 3D surface. Accordingly, the view generator 170 can generate a surface topology visualization by rendering a view of the textured 3D surface. In some embodiments, the view generator may generate an untextured view of the detected 3D surface topology (or other model of the surrounding environment) (e.g., a 2D image with placeholder pixels of distance or color values), project one or more depth values ​​into the 2D view (e.g., generate a corresponding depth map), and texture the 2D view using the corresponding (e.g., per-pixel) depth values ​​or otherwise apply a graphical representation of a visual element (e.g., a color, image, design, pattern, etc.) to the 2D view. Figure 3F 、 Figure 3H 、 Figure 3J 、 Figure 3L 、 Figure 3M and Figure 3N An example of a surface topology visualization that can be generated by modeling the surrounding environment as the detected 3D surface topology is shown in FIG.

[0051] In some implementations and / or scenarios, the detected 3D surface topology modeling pipeline 117 can represent the detected 3D surface topology as a (e.g., characterized) point cloud, and the view generator 170 can use the detected 3D surface topology to model the environment to generate a visualization of the environment (e.g., a surface topology visualization). For example, the detected 3D surface topology modeling pipeline 117 can apply a representation of sensor data (e.g., image data 105) to one or more neural networks to extract a corresponding depth map, and the detected 3D surface topology modeling pipeline 117 can featurize the depth map (e.g., by assigning each pixel in the depth map a color from a corresponding pixel in the input image). The detected 3D surface topology modeling pipeline 117 can enhance the featurized depth map to generate and / or featurize a (e.g., sparse) point cloud representing the detected 3D surface topology of the surrounding environment (e.g., where each point in the featurized point cloud carries a color value from a corresponding pixel in the image data). Thus, the view generator 170 can generate a surface topology visualization by rendering a view of the characterized point cloud and / or applying the resulting (e.g., sparse) image to a neural renderer to densify and generate a photo-realistic image. The view generator 170 can use any known neural rendering technique to densify the image. In some embodiments, the surface topology modeling pipeline 117 and / or the 3D bowl modeling pipeline 113 can use any known hole filling technique to identify and fill holes in the corresponding 3D model of the surrounding environment.

[0052] In some embodiments, the environment modeling pipeline 113 may additionally or alternatively support other environment modeling techniques, such as any known view synthesis or view generation techniques, to generate corresponding visualizations based on corresponding supported environment modeling pipelines.

[0053] The texture mapping component 160 can project and / or texture the 3D representation of the environment generated by the environment modeling pipeline 113. In some embodiments, to perform texturing, the texture mapping component 160 back-projects 2D points in pixel coordinates to 3D points using corresponding depth values ​​obtained from a detected 3D surface topology (e.g., an SDF, an extracted surface mesh), from a sensed or extracted depth map (e.g., generated using a LiDAR or RADAR sensor, using one or more neural networks, etc.), and / or otherwise. With image data assigned to corresponding 3D points, the texture mapping component 160 can project the 3D points onto a 3D surface representation, such as the 3D surface topology, to generate a textured 3D surface topology. Thus, in some embodiments, the texture mapping component 160 can back-project sensor data (e.g., image data 105) from the sensor 101 onto a 3D surface representation of the detected 3D surface topology using corresponding depth values ​​(e.g., represented by a smoothed or unsmoothed 3D surface representation, rendered using an unstructured photometric map, etc.) during each back-projection to generate a textured 3D surface. In some embodiments (e.g., where depth values ​​are extracted from the current and / or previous time slices), the texture can be back-warped from corresponding image data frames from any number of sensors 101 and / or corresponding time slices. In some implementations, areas not covered by the texture can be determined to be outdated or stale. Thus, the texture mapping component 160 can cover the detected outdated or stale areas and / or can use blurring to represent such areas. Additionally or alternatively, the texture mapping component 160 can texture the 3D surface topology using image data frames representing one or more previous time slices, and can represent potentially outdated or stale areas by reducing color saturation, applying tint, and / or any other texturing technique.

[0054] The view generator 170 can render a view of a (e.g., textured) 3D model of the surrounding environment generated by the environment modeling pipeline 113 (e.g., a detected 3D surface topology model generated by the detected 3D surface topology modeling pipeline 117, a 3D bowl model generated by the 3D bowl modeling pipeline 115) to generate a corresponding visualization (e.g., a surface topology visualization, a bowl visualization, etc.). For example, the view generator 170 can position and orient a virtual camera in the 3D scene using the 3D model of the surrounding environment and render a view of the 3D model from the perspective of the virtual camera through a corresponding viewport. In some embodiments, the viewport can be selected based on the driving scene (e.g., orienting the viewport in the direction of the ego's motion), based on a detected salient event (e.g., orienting the viewport toward a detected salient event), based on an in-vehicle command (e.g., orienting the viewport in a direction indicated by a command issued by an operator or occupant of the ego machine), based on a remote command (orienting the viewport in a direction indicated by a remote command), and / or other means. Thus, the view generator 170 may output a visualization of the surrounding environment (e.g., a surround view visualization, a third person view, a cabin view, an edge guard view, a side view, a top-down view, a forward view, a rearward view, a view below the self-machine, and / or other), for example, via a display 180 (e.g., a monitor visible to an occupant or operator of the self-machine).

[0055] In some embodiments, the environment modeling pipeline selection component 120 selects or switches between supported environment modeling techniques (e.g., those implemented by the 3D bowl modeling pipeline 115 and the detected 3D surface topology modeling pipeline 117). In some embodiments, the environment modeling pipeline selection component 120 includes an environment model switching state machine 150 that implements decision logic for selecting and / or switching between supported environment modeling techniques. The switching and / or selection can be based on any number of detectable factors, such as the state of the ego machine determined by the ego machine state detection component 140 (e.g., proximity to detected pedestrians; proximity to detected objects; one or more detected features indicating a specified operating scenario, such as a parking scenario or an off-road scenario; speed of ego motion; and / or other factors); the estimated or predicted image quality of the corresponding visualization determined by the estimated image quality module 146 (e.g., which can be based on the confidence of the corresponding depth estimate used to generate the visualization); the selections and / or preferences of the ego machine operator as retrieved from a corresponding configuration file by the selection / preference module 147; and / or other factors.

[0056] In some embodiments, the environment model switching state machine 150 implements a decision tree that determines whether to use the 3D bowl modeling pipeline 115 , the detected 3D surface topology component 116 , and / or some other modeling technique. Figure 2is a schematic diagram 200 of an example environment model switching state machine 201, which may correspond to Figure 1 The environment model switching state machine 150. The environment model switching state machine 201 includes an example driving scenario 205 based on determining that the self-machine is in motion, based on a display (e.g., Figure 1 The environment model switching state machine 201 may be enabled by default based on the 3D bowl model of the surrounding environment (e.g., the bowl model of the surrounding environment) or a specific display interface being active, activated by default based on selecting an option for viewing a specific visualization, and / or other. Figure 1 One or more bowl visualizations supported by the 3D bowl modeling pipeline 115 of FIG. 1 . In this example scenario, at block 210, the ego velocity is determined (e.g., via Figure 1 If the ego speed is below the lower speed threshold (e.g., less than 8-10 km / hr), then at block 215, the environment model switching state machine 201 determines or accesses a state machine associated with the nearest object (e.g., via Figure 1 If the distance to the nearest object is within a certain threshold proximity (e.g., from 1 m to 3 m), then at block 220, the environment model switching state machine 201 determines an estimated or predicted image quality of the corresponding surface topology visualization (e.g., via Figure 1 If the estimated or predicted image quality of the corresponding surface topology visualization is above a threshold quality, then at block 225 the environment model switching state machine 201 enables generating the surface topology visualization (e.g., using Figure 1 Detected 3D surface topology modeling pipeline 117).

[0057] Returning to block 220, if the estimated or predicted image quality of the corresponding surface topology visualization is below a threshold quality, then at block 230, the environment model switching state machine 201 disables generating the surface topology visualization in favor of the bowl visualization (e.g., using Figure 1 ). Returning to block 215, if the distance to the nearest object is above a certain threshold proximity (e.g., above 3 m), then at block 230 the environment model switching state machine 201 disables generating surface topology visualization in favor of bowl visualization. Continuing to block 215, if the distance to the nearest object is below a threshold proximity (e.g., less than 1 m), then at block 240 the environment model switching state machine 201 disables generating surface topology visualization and / or generating bowl visualization in favor of raw image data (e.g., an image / video feed generated using a camera of the ego machine, such as Figure 3A 、 Figure 3B 、 Figure 3C and Figure 3D). Returning to block 210, if the ego speed is within a certain speed threshold (e.g., from 8-10 km / hr to 35 km / hr), then at block 230 the environment model switching state machine 201 disables generating the surface topology visualization in favor of the bowl visualization. Continuing with block 210, if the ego speed is above a certain speed threshold (e.g., above 35 km / hr), then at block 235 the environment model switching state machine 201 disables generating the surface topology visualization and / or generating the bowl visualization in favor of a simulated visualization (e.g., having simulated representations of nearby lanes and a visualization of simulated assets corresponding to detected objects (e.g., cars, trucks, motorcycles, pedestrians, or other)), which is enabled for display via the display 180.

[0058] In some embodiments, the environment model switching state machine 201 includes a driving scenario 245 based on determining that the speed of the ego machine is within a certain speed threshold (e.g., from 8-10 km / hr to 35 km / hr), based on the display (e.g., Figure 1 The environment model switching state machine 201 may be enabled by default based on the 3D bowl model of the surrounding environment (e.g., the bowl model of the surrounding environment) or the specific display interface being active, based on selecting an option for viewing a specific visualization, and / or other. Figure 1 In this example scenario, at block 250, the environment model switching state machine 201 determines whether a specified operating scenario exists (e.g., via Figure 1 If a designated operating scenario for the surface topology visualization is detected (e.g., parking or off-road scenario), then at block 255, the environment model switching state machine 201 determines a measure of estimated or predicted image quality of the corresponding surface topology visualization, as explained in more detail below. If the measure of estimated or predicted image quality of the surface topology visualization is above a threshold, then at block 260, the environment model switching state machine 201 enables generation of the surface topology visualization (e.g., using Figure 1 Detected 3D surface topology modeling pipeline 117). Returning to block 255, if the measured value of the estimated or predicted image quality of the corresponding surface topology visualization is below the threshold quality, then at block 265 the environment model switching state machine 201 disables generating the surface topology visualization in favor of the bowl visualization. Returning to block 250, if a designated operating scenario for bowl visualization is detected (e.g., a non-parking scenario, driving within a threshold speed range, etc.), then at block 265 the environment model switching state machine 201 enables the bowl visualization.

[0059] Back to Figure 1In an example data flow, the environment visualization generator 111 (e.g., its view generator 170) can use sensor data (e.g., image data 105) from the sensor 101 (e.g., one or more cameras of the ego machine) to generate and / or enable one or more visualizations of the environment surrounding the ego machine from any suitable viewing direction. For example, the view generator 170 can generate a visualization (e.g., a textured 3D model of the surrounding environment) from the perspective of a virtual camera positioned and / or oriented in the corresponding 3D scene with specified different viewing directions (e.g., a side view relative to the ego machine, a top-down view, a forward view relative to the ego machine, a rearward view relative to the ego machine, a view from below the ego machine, and / or other), and / or the view generator 170 can cause the display 180 to present the visualization or present some indication of which views are available.

[0060] Figures 3E-3F Shown are examples of different types of visualizations for different viewing directions generated using the example environment modeling pipeline. Figure 3E An example bowl visualization 300E from a third person (eg, perspective) view is shown in FIG. Figure 1 3D bowl modeling pipeline 115 is generated). Figure 3F An example surface topology visualization 300F (eg, which may be generated using the detected surface topology modeling pipeline 117) with a third person (eg, perspective) view is shown in FIG. Figure 3F The surface topology visualization of 300F provides less distortion in certain areas (e.g. pedestrians, curbs, cars, etc.) Figure 3E The Bowl Visualizer 300E.

[0061] Figure 3G An example bowl visualization from a cabin view is shown in 300G (e.g., which can be used Figure 1 Bowl visualization modeling pipeline 115 generated). Figure 3H An example surface topology visualization 300H from a cabin view is shown in FIG. Figure 1 The detected surface topology modeling pipeline 117 generates (in this example, Figure 3H The surface topology visualization of 300H provides less distortion in certain areas (e.g. pedestrians, curbs, cars, etc.) Figure 3G Bowl visualization 300G.

[0062] Figure 3I An example bowl visualization 300I from an edge protection view is shown in FIG. Figure 1 Bowl visualization modeling pipeline 115 generated). Figure 3JAn example surface topology visualization 300J from an edge-protected view is shown in FIG. Figure 1 The detected surface topology modeling pipeline 117 generates (in this example, Figure 3J The surface topology visualization of 300J provides less distortion in certain areas (e.g. pedestrians, curbs, cars, etc.) Figure 3I Bowl Visualizer 300i.

[0063] Figure 3K An example bowl visualization 300K (e.g., which can be used to Figure 1 Bowl visualization modeling pipeline 115 generated). Figure 3L An example surface topology visualization 300L (e.g., which can be used to visualize the view of a cabin in a parking space) is shown in FIG. Figure 1 The detected surface topology modeling pipeline 117 generates (in this example, Figure 3L The surface topology visualization of 300L provides less distortion in certain areas (e.g. parking spaces, cars, etc.) Figure 3K Bowl visualization 300K.

[0064] Figures 3M-3N Additional examples of surface topology visualization are shown in . Figure 3M An example surface topology visualization 300M from a third person view is shown in Figure 1 The detected surface topology modeling pipeline 117 generates (in this example, Figure 3M The surface topology visualization of 300M should provide less distortion of pedestrians than the corresponding bowl visualization. Figure 3N An example surface topology visualization 300N from a top-down view is shown in FIG. Figure 1 The detected surface topology modeling pipeline 117 generates (in this example, Figure 3N The surface topology visualization 300N should provide less distortion of surrounding vehicles in the parking space and the parking lot than a corresponding bowl visualization. These are just a few examples of where a surface topology visualization can provide better visualization than a corresponding bowl visualization (e.g., due to inaccuracies in depth values ​​represented by the bowl visualization, e.g., due to a mismatch between the flat ground assumption used by the bowl visualization and the surrounding environment). Other examples include: a curb view, where a virtual camera can be placed low to the ground (e.g., below a transparent model of the vehicle) and oriented (e.g., toward the curb side of the vehicle) to focus on the curb and assist in parking or maneuvering in tight spaces; an off-road view, where a virtual camera can be placed low to the ground (e.g., below a transparent model of the vehicle) and oriented (e.g., toward the direction of travel or detected obstacles) to focus on the terrain of the road and assist in off-road maneuvering, and / or other examples.

[0065] Back to Figure 1 In some implementations and / or detected scenarios (e.g., detected by the environment modeling pipeline selection component 120), visualizations from one or more viewing directions may be generated (or selectable) as surface topology visualizations (e.g., using the detected 3D surface topology modeling pipeline 117), and in some implementations and / or detected scenarios (e.g., detected by the environment modeling pipeline selection component 120), visualizations from one or more viewing directions may be generated (or selectable) as bowl visualizations (e.g., using the 3D bowl modeling pipeline 117).

[0066] In an example embodiment, a bowl visualization generated by the view generator 170 (e.g., generated using the 3D bowl modeling pipeline 117) can be automatically displayed or selected as a default visualization for one or more viewing directions (e.g., or all available viewing directions) via the display 180. Thus, the environment modeling pipeline selection component 120 can detect one or more specified scenarios (e.g., a parking scenario, an estimated or predicted image quality of a surface topology visualization above a threshold), and can use the environment model switching state machine 150 to switch the enablement of one or more visualizations supported by the surface topology visualization (e.g., generated by the detected 3D surface topology modeling pipeline 117) based on the one or more specified scenarios.

[0067] Examples of scenarios for which the environment modeling pipeline selection component 120 can enable surface topology visualization to be generated, displayed, and / or selected via the display 180 include: detecting a specified class of objects (e.g., pedestrians) via the perception module 141; detecting objects (e.g., or a class of objects (such as cars), objects greater than a threshold size (e.g., curbs greater than a threshold height), and / or other) within a threshold distance of the ego machine (e.g., within two or three meters of the ego machine) via the perception module 141; detecting off-road scenarios via the perception module 141 (e.g., based on an enabled driving mode, such as inferred based on detection of an object located on the driving surface and having a detected height within a threshold height range, which may indicate a large obstacle that can be driven over); detecting or predicting parking scenarios, such as entering or exiting a parking space, parallel parking, etc., via the perception module 141. parking, etc. (e.g., based on operator input, such as commanding an automatic parking maneuver or selecting a parking space on a display; detecting a speed of ego-motion below a specified threshold (e.g., 8-10 km / hr), predicting a driver's parking intention, detecting a trajectory, detecting the presence of a vacant parking space, detecting one or more driver actions (e.g., signaling or shifting gears), detecting that the ego-machine is reversing, some combination thereof, etc.); indicating some measure of estimated or predicted image quality of the resulting visualization (e.g., less than a certain number, amount, or coverage of occlusions, at least some threshold confidence level of extracted depth values ​​represented by a 3D model of the surrounding environment) via the estimated image quality module 146; making selections and / or preferences of the operator of the ego-machine (e.g., for surface topology visualization) via the selection / preference module 147; and / or other (e.g., such as a combination thereof).

[0068] Examples of scenarios in which the environment modeling pipeline selection component 120 may enable the bowl visualization to be generated, displayed, and / or selected via the display 180 include: detection of an object (e.g., or certain types of objects, such as a car or an object exceeding a threshold size) beyond a certain threshold distance (e.g., two or three meters) from the ego machine by default via the perception module 141; detection of a driving scene on a road via the perception module 141 (e.g., based on an enabled driving mode, such as inferred based on not detecting an object exceeding a threshold height on the driving surface); detection of an ego-motion speed exceeding a specified threshold (e.g., 8-10 km / hr) via the perception module 141; an indication via the estimated image quality module 146 that the surface topology visualization does not provide an improved visualization (e.g., more than the number, amount, or coverage of occlusions, less than a certain threshold confidence level of the extracted depth values); selections and / or preferences (e.g., for the bowl visualization) by the ego machine operator via the selection / preference module 147; and / or other (e.g., such as a combination thereof). In some embodiments, the bowl visualization is always available for selection for display via the display 180.

[0069] The environment modeling pipeline selection component 120 may include an ego-machine state detection component 140 that generates or receives a representation of the ego-machine state (e.g., a detected proximity to a detected object; one or more detected features indicating a specified operating scenario in which the ego-machine is currently operating or predicted to begin operating; a velocity of the ego-machine's motion; and / or other) using any known technique. The ego-machine state detection component 140 may detect the ego-machine's state or otherwise receive a representation of the ego-machine's detected state, such as from one or more upstream components (e.g., one or more neural networks). For example, in some embodiments, one or more sensors 101 of the ego-machine (e.g., a camera, a RADAR sensor, a LiDAR sensor, an ultrasonic sensor) may be used to generate sensor data representing the environment surrounding the ego-machine, and any known object detection and / or tracking technique may be used (e.g., by the object detector 142 of the perception module 141, by one or more upstream components (e.g., one or more neural networks)) to detect objects (e.g., moving obstacles or subclasses thereof, such as vehicles, pedestrians, and / or other) and estimate their positions in the environment (e.g., relative to the ego-machine) based on the sensor data. In some embodiments, object detector 142 may be a component of perception module 141 , and perception module 141 may be used for object detection and / or tracking.

[0070] The perception module 141 may use any known perception technology, such as any known autonomous vehicle (AV) perception technology. In some embodiments, the perception module 141 may detect or predict depth values ​​corresponding to an input image generated using a camera of the ego machine to determine the proximity of the ego machine to a detection object represented in the image.

[0071] The speed detector 143 of the perception module 141 can use any known technology to determine the speed of the self-motion. For example, the speed detector 143 can use a vehicle speed sensor, a gyroscope, an accelerometer, an IMU and / or other methods to determine the self-motion.

[0072] The operating scenario detector 144 of the perception module 141 can use any known techniques to detect a designated operating scenario, such as a designated parking or off-roading scenario. In some embodiments, the operating scenario detector 144 can be a component of the perception module 141 (or can receive one or more signals or other data generated by the perception module 141). Generally, the operating scenario detector 144 can use any known techniques to detect one or more features indicative of a designated operating scenario. For example, the operating scenario detector 144 can determine that a parking scenario (e.g., entering or exiting a parking space, parallel parking) is occurring or is about to begin based on detecting one or more features of one or more current operating or operator conditions, such as the presence of a stop line or parking space, nearby stationary vehicles, proximity to a building or structure, reduced traffic flow, reduced speed or deceleration of the ego machine, detected operator intent (e.g., detected by the operator intent module 145 based on eye movements, gaze direction, facial expression, steering pattern, and / or other factors), and / or other factors. In some embodiments, the operating scenario detector 144 may determine that an off-road scenario is occurring or about to begin based on detecting one or more characteristics of one or more current operating or operator conditions, which characteristics may be indicative of an off-road driving scenario, such as the absence of road markings, rough terrain, uneven surfaces, vegetation, obstacles (e.g., rocks or trees), engagement of an off-road driving mode, suspension adjustments, and / or the like. These are merely a few examples, and any detectable characteristics indicative of any specified operating scenario may be used within the scope of the present disclosure.

[0073] In some embodiments, operator intent module 145 can predict the ego-machine operator's intent to engage in a specified operating scenario (e.g., a parking scenario). For example, operator intent module 145 can be part of (or in communication with) a driver monitoring system (DMS) that uses any known technique to predict the operator's intent to engage in a specified operating scenario (e.g., a parking scenario), such as by tracking eye movements, gaze direction, head orientation, hand gestures, facial expressions, steering patterns, and / or other techniques. Continuing with the example of detecting parking intent, operator intent module 145 (and / or a corresponding DMS) can infer parking intent based on detecting the operator looking at a potential parking space, turning their head to examine the surroundings, and / or reaching for the gear selector or parking brake. Thus, one or more monitored operator characteristics and / or some encoded representation of sensor data representing the environment can be used (e.g., by a neural network) to predict intent to engage in a specified operating scenario based on various internal cues (e.g., observed operator characteristics) and / or external cues (e.g., nearby parking spaces).

[0074] The estimate image quality module 146 may use any known techniques and / or any suitable metrics to determine a measure of estimated or predicted image quality of the visualization (e.g., the number, amount, and / or coverage of occlusions; confidence in the corresponding depth estimate; and / or other). For example, if the estimate image quality module 146 determines that a measure of the number, size, or (e.g., percentage) coverage of occluded regions in the image data used for textured surface topology visualization or in the generated surface topology visualization is above a specified threshold, the environment modeling pipeline selection component 120 may disable or determine not to enable the surface topology visualization.

[0075] In some embodiments, view generator 170 (or some other component) may insert a simulated representation of a detected object (e.g., a 3D model representing a typical pedestrian) into the 3D representation of the surrounding environment and / or the corresponding 2D visualization in place of real image data, for example, based on a determination that the detected object is occluded, located between the ego machine and the virtual camera, and / or otherwise. For example, if an object (e.g., a pedestrian) is detected and can be rendered in a photorealistic manner (e.g., the estimated image quality module 146 determines that a measure of the quality of the object rendering is above a threshold), view generator 170 may render the object in a photorealistic manner (e.g., based on real image data showing the pedestrian's face). If the object cannot be rendered in a photorealistic manner (e.g., the quality of the object rendering is below a threshold, for example, because the estimated image quality module 146 determines that only half of the pedestrian is depicted in the input image), view generator 170 (or some other component) may insert a simulated representation of the object into the 3D representation of the surrounding environment and / or the corresponding 2D visualization instead of rendering the object based on real image data. For example, view generator 170 may insert a simulated asset (e.g., a 3D mannequin of a person, a 3D model of a car, or other simulated model) into a 3D model of an environment or a corresponding 2D visualization to replace an object in the visualization. As another example, view generator 170 may insert content generated by a generative artificial intelligence model into a 3D model of an environment or a corresponding 2D visualization to replace an object in the visualization. The quality of the object rendering (e.g., the confidence that the object is rendered in a photorealistic manner) may be determined using any known techniques and / or using any suitable metric (e.g., via estimate image quality module 146).

[0076] In some embodiments, the environment modeling pipeline selection component 120 can enable visualization (e.g., of surface topology) corresponding to a specified viewing direction in certain scenarios before other viewing directions are enabled. For example, after receiving a certain amount of historical image data, a top-down view of the ego machine's surface topology visualization can be enabled to provide a visualization with above-threshold image quality (e.g., below a threshold amount of occlusion in the visualization). As another example, an edge view visualization of the ego machine's surface topology visualization can be enabled for display immediately after detecting a specified operating scenario or being selected by an operator of the ego machine, because the edge view visualization can generate a visualization with above-threshold image quality without requiring historical image data.

[0077] In some embodiments, one or more components of the environment visualization generator 111 (e.g., the perception module 141, the estimate image quality module 146, etc.) and / or other components may be implemented using a neural network (e.g., a convolutional neural network (CNN)), but this is not intended to be limiting. For example, but not limitation, the components of the environment visualization generator 111 may include any type of multiple different networks or machine learning models, such as machine learning models using linear regression, logistic regression, decision trees, support vector machines (SVMs), naive Bayes, k-nearest neighbors (Knn), K-means clustering, random forests, dimensionality reduction algorithms, gradient boosting algorithms, neural networks (e.g., autoencoders, convolutions, transformers, loops, perceptrons, long / short term memory (LSTM), large language models (LLM), Hopfield, Boltzmann, deep belief, deconvolution, generative adversarial, liquid machines, etc.), and / or other types of machine learning models.

[0078] Now refer to Figure 4 Each block of the method 400 described herein includes a computing process that can be performed using any combination of hardware, firmware, and / or software. For example, the various functions can be performed by a processor executing instructions stored in a memory. The methods can also be embodied as computer-usable instructions stored on a computer storage medium. The methods can be provided by a standalone application, a service, or a hosted service (standalone or in combination with another hosted service), or a plug-in to another product, to name a few. In addition, the method 400 is about Figure 1 The environment visualization pipeline 100 is described by way of example. However, these methods may additionally or alternatively be performed by any one system or any combination of systems, including but not limited to the systems described herein.

[0079] Figure 4is a flow chart illustrating a method 400 for switching between visualizations according to some embodiments of the present disclosure. The method 400 includes, at block B402, generating two or more image frames based on at least sensor data corresponding to a first time slice and generated using one or more sensors of an ego machine in an environment. For example, regarding Figure 1 The environment visualization pipeline 100 may include any number and type of sensors 101 , such as one or more cameras, which may be used to generate sensor data (eg, image data 105 ) representing the surrounding environment.

[0080] The method 400 includes, at block B404, switching from generating a first visualization of the environment using a three-dimensional (3D) bowl model of the environment to generating a second visualization of the environment using a 3D surface topology model of the environment based at least on the state of the ego machine. Figure 1 In some embodiments, the environment modeling pipeline selection component 120 can select or switch between supported environment modeling techniques (e.g., implemented by the 3D bowl modeling pipeline 115 and the detected 3D surface topology modeling pipeline 117). In some embodiments, the environment modeling pipeline selection component 120 includes an environment model switching state machine 150 that implements decision logic for selecting and / or switching between supported environment modeling techniques. The switching and / or selection can be based on any number of detectable factors, such as the state of the ego machine determined by the ego machine state detection component 140 (e.g., proximity to detected pedestrians; proximity to detected objects; one or more detected features indicating a specified operating scenario, such as a parking scenario or an off-road scenario; speed of ego motion; and / or other factors); the estimated image quality of the corresponding visualization determined by the estimated image quality module 146 (e.g., which can be based on the confidence of the corresponding depth estimate used to generate the visualization); the selections and / or preferences of the ego machine operator as retrieved by the selection / preference module 147; and / or other factors.

[0081] Method 400 includes, at block B406, presenting a second visualization of the environment based on at least two or more image frames. Figure 1 , the detected 3D surface topology modeling pipeline 117 can generate visualizations using an environment modeling pipeline that models the surrounding environment as a detected 3D surface topology (e.g., in certain scenarios, such as low ego-motion speed, objects detected within a threshold distance, etc.), and / or the 3D bowl modeling pipeline 115 can generate visualizations using an environment modeling pipeline that models the surrounding environment as a 3D bowl (e.g., in other scenarios, such as high ego-motion speed, no objects detected within a threshold distance, etc.).

[0082] The systems and methods described herein may be used by, but are not limited to, non-autonomous vehicles, semi-autonomous vehicles (e.g., in one or more adaptive driver assistance systems (ADAS)), manned and unmanned 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, construction vehicles, trains, submarines, remotely operated vehicles such as drones, and / or other vehicle types. In addition, the systems and methods described herein may be used for a variety of purposes, such as, but not limited to, 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, light transport simulation (e.g., ray tracing, path tracing, etc.), collaborative content creation of 3D assets, cloud computing, generative AI, and / or any other suitable application.

[0083] Embodiments of the present disclosure include 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, aviation 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, systems implementing one or more language models such as one or more large language models (LLMs), 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.

[0084] Example autonomous vehicle

[0085] Figure 5Ais an illustration of an example autonomous vehicle 500, according to some embodiments of the present disclosure. Autonomous vehicle 500 (alternatively referred to herein as "vehicle 500") 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, a construction vehicle, a submarine, a robotic vehicle, a drone, an airplane, a vehicle coupled to a trailer (e.g., a semi-tractor-trailer truck for hauling cargo), and / or other types of vehicles (e.g., unmanned and / or capable of accommodating one or more occupants). Autonomous vehicles are generally described in terms of levels of automation as 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. J3016-201806, issued on June 15, 2018, Standard No. J3016-201609, issued on September 30, 2016, and prior and future versions of such standards). The vehicle 500 may be capable of implementing one or more functions consistent with autonomous driving levels 3-5. The vehicle 500 may be capable of implementing one or more functions consistent with autonomous driving levels 3-5. For example, depending on the embodiment, the vehicle 500 may be capable of driver assistance (Level 1), semi-automation (Level 2), conditional automation (Level 3), high automation (Level 4), and / or full automation (Level 5). The term “autonomous” as used herein may include any and / or all types of autonomy of a vehicle 500 or other machine, such as fully autonomous, highly autonomous, conditionally autonomous, partially autonomous, with assistive autonomy, semi-autonomous, primarily autonomous, or other designations.

[0086] Vehicle 500 may include components such as a chassis, a body, wheels (e.g., 2, 4, 6, 8, 18, etc.), tires, axles, and other vehicle components. Vehicle 500 may include a propulsion system 550, such as an internal combustion engine, a hybrid power plant, an all-electric engine, and / or another type of propulsion system. Propulsion system 550 may be connected to a drivetrain of vehicle 500, which may include a transmission, to achieve propulsion of vehicle 500. Propulsion system 550 may be controlled in response to receiving a signal from throttle / accelerator 552.

[0087] A steering system 554, which may include a steering wheel, may be used to steer vehicle 500 (e.g., along a desired path or route) when propulsion system 550 is operating (e.g., when the vehicle is in motion). Steering system 554 may receive signals from steering actuator 556. For fully automated (Level 5) functionality, a steering wheel may be optional.

[0088] Brake sensor system 546 may be used to operate vehicle brakes in response to receiving signals from brake actuator 548 and / or brake sensors.

[0089] May include one or more system on chip (SoC) 504 ( Figure 5C ) and / or one or more GPUs can provide signals (e.g., representing commands) to one or more components and / or systems of the vehicle 500. For example, the one or more controllers can send signals to operate the vehicle brakes via one or more brake actuators 548, to operate the steering system 554 via one or more steering actuators 556, and to operate the propulsion system 550 via one or more throttles / accelerators 552. The one or more controllers 536 can 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 500. The one or more controllers 536 can include a first controller 536 for autonomous driving functions, a second controller 536 for functional safety functions, a third controller 536 for artificial intelligence functions (e.g., computer vision), a fourth controller 536 for infotainment functions, a fifth controller 536 for redundancy in emergency situations, and / or other controllers. In some examples, a single controller 536 may handle two or more of the above functions, two or more controllers 536 may handle a single function, and / or any combination thereof.

[0090] The one or more controllers 536 may provide signals for controlling one or more components and / or systems of the vehicle 500 in response to sensor data (eg, sensor input) received from one or more sensors. Sensor data may be received from, for example and without limitation, global navigation satellite system sensors (“GNSS”) 558 (e.g., global positioning system sensors), RADAR sensors 560 , ultrasonic sensors 562 , LiDAR sensors 564 , inertial measurement unit (IMU) sensors 566 (e.g., accelerometers, gyroscopes, magnetic compasses, magnetometers, etc.), microphones 596 , stereo cameras 568 , wide-angle cameras 570 (e.g., fisheye cameras), infrared cameras 572 , surround cameras 574 (e.g., 360-degree cameras), long-range and / or mid-range cameras 598 , speed sensors 544 (e.g., for measuring the velocity of the vehicle 500 ), vibration sensors 542 , steering sensors 540 , brake sensors (e.g., as part of a brake sensor system 546 ), one or more occupant monitoring system (OMS) sensors 501 (e.g., one or more interior cameras), and / or other sensor types.

[0091] One or more of the controllers 536 may receive input (e.g., represented by input data) from the instrument cluster 532 of the vehicle 500 and provide output (e.g., represented by output data, display data, etc.) via a human machine interface (HMI) display 534, an audible annunciator, a speaker, and / or via other components of the vehicle 500. These outputs may include information such as vehicle speed, velocity, time, map data (e.g., Figure 5C The HMI display 534 may include information such as a high-definition ("HD") map 522 of the vehicle 500, location data (e.g., the location of the vehicle 500 on the map), directions, the locations of other vehicles (e.g., an occupancy grid), information about objects and object states as sensed by the controller 536, and the like. For example, the HMI display 534 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.).

[0092] The vehicle 500 further includes a network interface 524 that can communicate over one or more networks using one or more wireless antennas 526 and / or a modem. For example, the network interface 524 can be capable of communicating over Long Term Evolution ("LTE"), Wideband Code Division Multiple Access ("WCDMA"), Universal Mobile Telecommunications System ("UMTS"), Global System for Mobile Communications ("GSM"), IMT-CDMA Multi-Carrier ("CDMA2000"), etc. The one or more wireless antennas 526 can 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 Low Energy ("LE"), Z-Wave, ZigBee, etc., and / or one or more low power wide area networks (LPWANs) such as LoRaWAN, SigFox, etc.

[0093] Figure 5B For use according to some embodiments of the present disclosure Figure 5A An example of camera positions and fields of view for an autonomous vehicle 500 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 alternative cameras may be included and / or located at different locations on the vehicle 500.

[0094] The camera type used 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 500. The camera may operate under Automotive Safety Integrity Level (ASIL) B and / or under another ASIL. The camera type may have any image capture rate, such as 60 frames per second (fps), 120fps, 240fps, and the like, 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.

[0095] 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).

[0096] One or more of the cameras can be mounted in a mounting assembly, such as a custom-designed (three-dimensional "3D" printed) assembly, to cut out stray light and reflections from within the vehicle (e.g., reflections from the dashboard reflected in the windshield mirror) that may interfere with the camera's ability to capture image data. With respect to the wing mirror mounting assembly, the wing mirror assembly can be custom 3D printed so that the camera mounting plate matches the shape of the wing mirror. In some examples, one or more cameras can be integrated into the wing mirror. For side-view cameras, one or more cameras can also be integrated into the four pillars at each corner of the cab.

[0097] A camera with a field of view that includes a portion of the environment in front of the vehicle 500 (e.g., a front-facing camera) can be used for surround vision 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 536 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.

[0098] A variety of cameras may be used in the front-facing configuration, including, for example, a monocular camera platform including a complementary metal oxide semiconductor ("CMOS") color imager. Another example may be a wide-angle camera 570, which may be used to sense objects entering the field of view from the periphery (e.g., pedestrians, intersection traffic, or bicycles). Although Figure 5B The figure shows only one wide-angle camera, but there can be any number (including zero) of wide-angle cameras 570 on the vehicle 500. In addition, any number of long-range cameras 598 (e.g., a pair of long-view stereo cameras) can be used for depth-based object detection, especially for objects for which neural networks have not yet been trained. Long-range cameras 598 can also be used for object detection and classification and basic object tracking.

[0099] One or more stereo cameras 568 may also be included in the front configuration. In at least one embodiment, one or more of the stereo cameras 568 may include an integrated control unit that includes a scalable processing unit that can provide a multi-core microprocessor and programmable logic ("FPGA") with an integrated controller area network ("CAN") or Ethernet interface on a single chip. Such a unit can be used to generate a 3D map of the vehicle environment, including distance estimates for all points in the image. Alternative stereo cameras 568 may include a compact stereo vision sensor that may include two camera lenses (one on the left and one on the right) and an image processing chip that can 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. Other types of stereo cameras 568 may be used in addition to or alternatively to those described herein.

[0100] Cameras with a field of view that includes portions of the environment to the sides of the vehicle 500 (e.g., side view cameras) can be used for surround vision, providing information used to create and update occupancy grids and generate side impact collision warnings. For example, surround cameras 574 (e.g., Figure 5B Four surround cameras 574 (shown in FIG) can be placed on the vehicle 500. The surround cameras 574 can include a wide-angle camera 570, a fisheye camera, a 360-degree camera, and / or the like. For example, the four fisheye cameras can be placed on the front, rear, and sides of the vehicle. In an alternative arrangement, the vehicle can use three surround cameras 574 (e.g., left, right, and rear), and can utilize one or more other cameras (e.g., a forward-facing camera) as a fourth surround-view camera.

[0101] A camera having a field of view that includes a portion of the environment behind the vehicle 500 (e.g., a rearview camera) can be used to assist with parking, surround view, rear collision warning, and creating and updating occupancy grids. 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 598, stereo cameras 568, infrared cameras 572, etc.).

[0102] A camera (e.g., one or more OMS sensors 501) whose field of view includes a portion of the interior environment within the cabin of the vehicle 500 can be used as part of an occupant monitoring system (OMS), such as, but not limited to, a driver monitoring system (DMS). For example, an OMS sensor (e.g., OMS sensor 501) can be used (e.g., by a controller 536) to track the gaze direction, head posture, and / or blinking of an occupant and / or driver. This gaze information can be used to determine the occupant's or driver's attention level (e.g., to detect drowsiness, fatigue, and / or distraction) and / or take responsive action to prevent harm to the occupant or operator. In some embodiments, data from the OMS sensor can be used to implement gaze-controlled operations triggered by the driver and / or non-driver occupants, such as, but not limited to, adjusting cabin temperature and / or airflow, opening and closing windows, controlling cabin lighting, controlling the entertainment system, adjusting rearview mirrors, adjusting seat position, and / or other operations. In some embodiments, the OMS can be used for applications such as determining when objects and / or occupants remain in the cabin (e.g., by detecting the presence of occupants after the driver has exited the vehicle).

[0103] Figure 5C For use according to some embodiments of the present disclosure Figure 5A 5. Block diagram of an example system architecture for an example autonomous vehicle 500. 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 conjunction with other components, and in any appropriate combination and location. The various functions described herein as being performed by entities 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.

[0104] Figure 5C Each of the components, features, and systems of the vehicle 500 is illustrated as being connected via a bus 502. The bus 502 may include a controller area network (CAN) data interface (alternatively, referred to herein as a "CAN bus"). The CAN may be a network internal to the vehicle 500 that assists in controlling various features and functions of the vehicle 500, 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., a CAN ID). The CAN bus may be read to find steering wheel angle, ground speed, engine revolutions per minute (RPM), button positions, and / or other vehicle status indicators. The CAN bus may be ASIL B compliant.

[0105] Although bus 502 is described here as a CAN bus, this is not intended to be limiting. For example, FlexRay and / or Ethernet may be used in addition to or in lieu of a CAN bus. Furthermore, although bus 502 is represented by a single line, this is not intended to be limiting. For example, there may be any number of buses 502, including 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 502 may be used to perform different functions and / or may be used for redundancy. For example, a first bus 502 may be used for collision avoidance functionality, and a second bus 502 may be used for drive control. In any example, each bus 502 may communicate with any component of vehicle 500, and two or more buses 502 may communicate with the same component. In some examples, each SoC 504, each controller 536, and / or each computer within the vehicle may have access to the same input data (e.g., input from sensors on vehicle 500) and may be connected to a common bus such as a CAN bus.

[0106] The vehicle 500 may include one or more controllers 536, such as those described herein. Figure 5A Controller 536 may be used for a variety of functions. Controller 536 may be coupled to any of the various other components and systems of vehicle 500 and may be used for control of vehicle 500, artificial intelligence of vehicle 500, infotainment for vehicle 500, and / or the like.

[0107] The vehicle 500 may include one or more system on a chip (SoC) 504. The SoC 504 may include a CPU 506, a GPU 508, a processor 510, a cache 512, an accelerator 514, a data store 516, and / or other components and features not shown. The SoC 504 may be used to control the vehicle 500 in a variety of platforms and systems. For example, the one or more SoCs 504 may be combined with an HD map 522 in a system (e.g., a system of the vehicle 500), which may be downloaded from one or more servers (e.g., a server) via a network interface 524. Figure 5D one or more servers 578) to obtain map refreshes and / or updates.

[0108] The CPU 506 may include a CPU cluster or CPU complex (alternatively, referred to herein as a "CCPLEX"). The CPU 506 may include multiple cores and / or L2 caches. For example, in some embodiments, the CPU 506 may include eight cores in a coherent multiprocessor configuration. In some embodiments, the CPU 506 may include four dual-core clusters, each with a dedicated L2 cache (e.g., a 2MB L2 cache). The CPU 506 (e.g., CCPLEX) may be configured to support simultaneous cluster operations such that any combination of CPU 506 clusters can be active at any given time.

[0109] The CPU 506 may implement power management capabilities including one or more of the following features: each hardware block may be automatically clock gated when idle to conserve 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. The CPU 506 may further implement an enhanced algorithm for managing power states, wherein allowed power states and expected wakeup 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 this work being offloaded to the microcode.

[0110] The GPU 508 may include an integrated GPU (alternatively referred to herein as an "iGPU"). The GPU 508 may be programmable and efficient for parallel workloads. In some examples, the GPU 508 may use an enhanced tensor instruction set. The GPU 508 may include one or more streaming microprocessors, each of which may include an L1 cache (e.g., an L1 cache with at least 96KB of storage capacity), and two or more of these streaming microprocessors may share an L2 cache (e.g., an L2 cache with 512KB of storage capacity). In some embodiments, the GPU 508 may include at least eight streaming microprocessors. The GPU 508 may use a computing application programming interface (API). In addition, the GPU 508 may use one or more parallel computing platforms and / or programming models (e.g., NVIDIA's CUDA).

[0111] In the case of automotive and embedded use cases, GPU 508 can be power optimized to achieve optimal performance. For example, GPU 508 can be manufactured on fin field effect transistors (FinFETs). However, this is not intended to be limiting, and GPU 508 can be manufactured using other semiconductor manufacturing processes. Each streaming microprocessor can incorporate 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, 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 computation and addressing calculations. The streaming microprocessor can 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.

[0112] The GPU 508 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.

[0113] The GPU 508 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 services (ATS) support may be used to allow the GPU 508 to directly access the CPU 506 page tables. In such an example, when the GPU 508 memory management unit (MMU) experiences a miss, an address translation request may be transmitted to the CPU 506. In response, the CPU 506 may look up the virtual-to-physical mapping for the address in its page table and transmit the translation back to the GPU 508. In this way, unified memory technology may allow a single unified virtual address space to be used for memory of both the CPU 506 and the GPU 508, thereby simplifying GPU 508 programming and porting of applications to the GPU 508.

[0114] In addition, GPU 508 can include access counters that can track how often GPU 508 accesses the memory of other processors. The access counters can help ensure that memory pages are moved to the physical memory of the processor that accesses them most frequently.

[0115] SoC 504 may include any number of caches 512, including those described herein. For example, cache 512 may include an L3 cache available to both CPU 506 and GPU 508 (e.g., connected to both CPU 506 and GPU 508). Cache 512 may include a write-back cache that can track the state of lines, for example, 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.

[0116] The SoC 504 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 related to the vehicle 500, such as processing a DNN. Furthermore, the SoC 504 may include a floating point unit (FPU) or other math coprocessor or digital coprocessor type for performing mathematical operations within the system. For example, the SoC 504 may include one or more FPUs integrated as execution units within the CPU 506 and / or GPU 508.

[0117] SoC 504 may include one or more accelerators 514 (e.g., hardware accelerators, software accelerators, or a combination thereof). For example, SoC 504 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 508 and offload some tasks of GPU 508 (e.g., freeing up more cycles of GPU 508 to perform other tasks). As an example, accelerator 514 may be used for targeted workloads (e.g., perception, convolutional neural networks (CNNs), etc.) that are stable enough to easily control 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).

[0118] The accelerator 514 (e.g., a hardware acceleration cluster) may include a deep learning accelerator (DLA). The DLA may include one or more tensor processing units (TPUs) that can be configured to provide an additional 10 trillion operations per second for deep learning applications and reasoning. The TPU may be an accelerator configured to perform image processing functions (e.g., for CNN, RCNN, etc.) and optimized for performing image processing functions. The DLA may be further optimized for a specific set of neural network types and floating-point operations and reasoning. The design of the DLA may provide higher performance per millimeter than a general-purpose GPU and far exceed the performance of the CPU. The TPU may perform several functions, including a single-instance convolution function, support for INT8, INT16, and FP16 data types for both features and weights, and post-processor functions.

[0119] 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: CNNs for object recognition and detection using data from camera sensors; CNNs for distance estimation using data from camera sensors; CNNs for emergency vehicle detection and recognition and detection using data from microphones; CNNs for facial recognition and vehicle owner identification using data from camera sensors; and / or CNNs for safety and / or security-related events.

[0120] The DLA can perform any function of the GPU 508, and by using an inference accelerator, for example, the designer can target any function to either the DLA or the GPU 508. For example, the designer can focus the processing of CNNs and floating-point operations on the DLA and leave other functions to the GPU 508 and / or other accelerators 514.

[0121] The accelerator 514 (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.

[0122] 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 memory devices. For example, the RISC core can include an instruction cache and / or tightly coupled RAM.

[0123] The DMA can enable components of the PVA to access system memory independently of the CPU 506. The DMA can 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 can support addressing in up to six or more dimensions, which can include block width, block height, block depth, horizontal block stepping, vertical block stepping, and / or depth stepping.

[0124] A vector processor can be a programmable processor that can be designed to efficiently and flexibly execute 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 speed.

[0125] Each of the vector processors can include an instruction cache and can be coupled to dedicated memory. As a result, in some examples, each of the vector processors can be configured to execute independently of the other vector processors. In other examples, the vector processors included in a particular PVA can be configured to employ data parallelism. For example, in some embodiments, multiple vector processors included in a single PVA can execute the same computer vision algorithm, but on different regions of an image. In other examples, the vector processors included in a particular PVA can execute different computer vision algorithms simultaneously on the same image, or even execute different algorithms on sequential images or portions of images. Among other things, any number of PVAs can be included in a hardware acceleration cluster, and any number of vector processors can be included in each of these PVAs. In addition, the PVAs can include additional error correction code (ECC) memory to enhance overall system security.

[0126] The accelerator 514 (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 514. In some examples, the on-chip memory may include at least 4MB of SRAM consisting of, for example and without limitation, eight field-configurable memory blocks that can be accessed by both the PVA and the DLA. Each pair of memory blocks may include an advanced peripheral bus (APB) interface, configuration circuitry, 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).

[0127] The on-chip computer vision network can include an interface that ensures that both the PVA and DLA provide ready and valid signals before transmitting any control signals / addresses / data. Such an interface can provide separate phases and separate channels for transmitting control signals / addresses / data, as well as burst-based communication for continuous data transmission. This type of interface can comply with ISO 26262 or IEC 61508 standards, but other standards and protocols can also be used.

[0128] In some examples, the SoC 504 may include a real-time ray tracing hardware accelerator such as that described in U.S. patent application Ser. No. 16 / 101,232 filed on Aug. 10, 2018. The real-time ray tracing hardware accelerator may be used to quickly and efficiently determine the position and extent of objects (e.g., within a world model) in order to generate real-time visual simulations for use in RADAR signal interpretation, for sound propagation synthesis and / or analysis, for SONAR system simulation, for general wave propagation simulation, for comparison with LiDAR data for positioning and / or other functions, and / or for other uses. In some embodiments, one or more tree traversal units (TTUs) may be used to perform one or more ray tracing related operations.

[0129] The accelerator 514 (e.g., a hardware accelerator cluster) has a wide range of uses in autonomous driving. The PVA can be a programmable vision 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. Therefore, the PVA performs well on semi-intensive or intensive 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 operations.

[0130] For example, according to one embodiment of the technology, PVA is used to perform computer stereo vision. In some examples, a semi-global matching-based algorithm can be used, but this is not intended to be limiting. Many applications for Level 3-5 autonomous driving require on-the-fly motion estimation / stereo matching (e.g., structure from motion, pedestrian recognition, lane detection, etc.). PVA can perform computer stereo vision functions on input from two monocular cameras.

[0131] In some examples, PVA can be used to perform dense optical flow, by processing raw RADAR data (e.g., using a 4D Fast Fourier Transform) to provide processed RADAR. In other examples, PVA is used for time-of-flight depth processing, by processing raw time-of-flight data to provide processed time-of-flight data.

[0132] The 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 confidence values ​​can be interpreted as probabilities, 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 positives versus false positives. For example, the system can set a threshold for confidence and only consider detections that exceed the threshold as true positives. In an automatic emergency braking (AEB) system, a false positive detection could cause the vehicle to automatically apply emergency braking, which is clearly undesirable. Therefore, only the most confident detections should be considered triggers for AEB. The DLA can run a neural network to regress the confidence value. This neural network can take as its input at least some subset of parameters, such as bounding box dimensions, a ground plane estimate obtained (e.g., from another subsystem), inertial measurement unit (IMU) sensor 566 output related to the vehicle 500's orientation and distance, and 3D position estimates of objects obtained from the neural network and / or other sensors (e.g., LiDAR sensor 564 or RADAR sensor 560).

[0133] SoC 504 may include one or more data stores 516 (e.g., memory). Data stores 516 may be on-chip memory of SoC 504 that may store neural networks to be executed on the GPU and / or DLA. In some examples, for redundancy and safety, data stores 516 may be large enough to store multiple instances of the neural network. Data stores 516 may include L2 or L3 cache 512. References to data stores 516 may include references to memory associated with the PVA, DLA, and / or other accelerators 514 as described herein.

[0134] SoC 504 may include one or more processors 510 (e.g., embedded processors). Processors 510 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 504 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 504 thermal and temperature sensor management, and / or SoC 504 power state management. Each temperature sensor may be implemented as a ring oscillator whose output frequency is proportional to temperature, and SoC 504 may use the ring oscillator to detect the temperature of CPU 506, GPU 508, and / or accelerator 514. 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 504 in a lower power state and / or place vehicle 500 in a driver safety parking mode (e.g., to safely park vehicle 500).

[0135] The processor 510 may further include a set of embedded processors that can be used as an audio processing engine. The audio processing engine can 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.

[0136] The processor 510 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 (such as timers and interrupt controllers), various I / O controller peripherals, and routing logic.

[0137] Processor 510 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 (such as timers, interrupt controllers, etc.), and / or routing logic. In safety mode, the two or more cores may operate in lockstep mode and function as a single core with comparison logic to detect any differences between their operations.

[0138] Processor 510 may further include a real-time camera engine, which may include a dedicated processor subsystem for handling real-time camera management.

[0139] Processor 510 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 a camera processing pipeline.

[0140] The processor 510 may include a video image compositer, which may be a processing block (e.g., implemented on a microprocessor) that implements the video post-processing functions required by the video playback application to produce the final image for the player window. The video image compositer may perform lens distortion correction for the wide-angle camera 570, the surround camera 574, and / or for the in-cab monitoring camera sensor. The in-cab monitoring 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 calls, dictate emails, change vehicle destinations, 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.

[0141] The video image compositer can include enhanced temporal noise reduction for both spatial and temporal noise reduction. For example, in the presence of motion in the video, the noise reduction appropriately weights spatial information and downweights information provided by neighboring 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 can use information from previous images to reduce noise in the current image.

[0142] The video image compositor can also be configured to perform stereo rectification on the input stereo footage frames. The video image compositor can further be used for user interface composition when the operating system desktop is in use and the GPU 508 does not need to continuously render new surfaces. Even when the GPU 508 is powered on and active for 3D rendering, the video image compositor can be used to offload the GPU 508 to improve performance and responsiveness.

[0143] The SoC 504 may further include a mobile industry processor interface (MIPI) camera serial interface, a high-speed interface for receiving video and input from a camera, and / or a video input block that may be used for camera and related pixel input functions. The SoC 504 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 assigned to a specific role.

[0144] SoC 504 may further include a wide range of peripheral interfaces to enable communication with peripherals, audio codecs, power management, and / or other devices. SoC 504 may be used to process data from cameras (connected via Gigabit multimedia serial links and Ethernet), sensors (e.g., LiDAR sensor 564, RADAR sensor 560, etc., which may be connected via Ethernet), data from bus 502 (e.g., vehicle 500 speed, steering wheel position, etc.), and data from GNSS sensor 558 (connected via Ethernet or CAN bus). SoC 504 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 506 from routine data management tasks.

[0145] SoC 504 can be an end-to-end platform with a flexible architecture that spans levels 3-5 of automation, providing a comprehensive functional safety architecture that leverages and efficiently uses computer vision and ADAS technologies for diversity and redundancy, along with deep learning tools to provide a platform for a flexible and reliable driving software stack. SoC 504 can be faster, more reliable, and even more energy- and space-efficient than conventional systems. For example, when combined with CPU 506, GPU 508, and data storage 516, accelerator 514 can provide a fast and efficient platform for Level 3-5 autonomous vehicles.

[0146] This technology therefore provides capabilities and functionality not achievable with conventional systems. For example, computer vision algorithms can be executed on CPUs, which can be configured using high-level programming languages ​​such as C 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 execution time and power consumption. In particular, many CPUs are unable to execute complex object detection algorithms in real time, a requirement for in-vehicle ADAS applications and practical Level 3-5 autonomous vehicles.

[0147] In contrast to conventional systems, by providing a CPU complex, a GPU complex, and a hardware acceleration cluster, the technology described herein allows for multiple neural networks to be executed simultaneously and / or sequentially, and the results to be combined to achieve Level 3-5 autonomous driving capabilities. For example, a CNN executed on a DLA or dGPU (e.g., GPU 520) can 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 can 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.

[0148] As another example, as required for Level 3, 4, or 5 driving, multiple neural networks can be run simultaneously. For example, a warning sign consisting of "Caution: Flashing lights indicate icing conditions" along with a light can be interpreted by several neural networks, either independently or collectively. The sign itself can be identified as a traffic sign by a first neural network deployed (e.g., a trained neural network), and the text "Flashing lights indicate icing conditions" can be interpreted by a second neural network deployed, which 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 can be identified by operating a third neural network deployed over multiple frames, which informs the vehicle's path planning software of the presence (or absence) of the flashing lights. All three neural networks can run simultaneously, for example, within the DLA and / or on GPU 508.

[0149] In some examples, a CNN for facial recognition and owner recognition can use data from a camera sensor to identify the presence of an authorized driver and / or owner of the vehicle 500. 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 504 provides security against theft and / or carjacking.

[0150] In another example, a CNN for emergency vehicle detection and identification can use data from microphone 596 to detect and identify emergency vehicle sirens. In contrast to conventional systems that use general classifiers to detect sirens and manually extract features, SoC 504 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 speed of emergency vehicles (for example, 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 558. 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 562, the control program can be used to execute emergency vehicle safety routines, slowing the vehicle, pulling to the side of the road, stopping the vehicle, and / or idling the vehicle until the emergency vehicle passes.

[0151] The vehicle may include a CPU 518 (e.g., a discrete CPU or dCPU) that may be coupled to the SoC 504 via a high-speed interconnect (e.g., PCIe). The CPU 518 may include, for example, an X86 processor. The CPU 518 may be used to perform any of a variety of functions, including, for example, arbitrating potentially inconsistent results between ADAS sensors and the SoC 504, and / or monitoring the status and health of the controller 536 and / or the infotainment SoC 530.

[0152] The vehicle 500 may include a GPU 520 (e.g., a discrete GPU or dGPU) that may be coupled to the SoC 504 via a high-speed interconnect (e.g., NVIDIA's NVLINK). The GPU 520 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 input (e.g., sensor data) from sensors of the vehicle 500.

[0153] The vehicle 500 may further include a network interface 524, which may include one or more wireless antennas 526 (e.g., one or more wireless antennas for different communication protocols, such as a cellular antenna, a Bluetooth antenna, etc.). The network interface 524 can be used to enable wireless connections to the cloud (e.g., to a server 578 and / or other network devices), to other vehicles, and / or to 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 500 with information about vehicles approaching the vehicle 500 (e.g., vehicles in front of, to the side of, and / or behind the vehicle 500). This functionality can be part of the cooperative adaptive cruise control functionality of the vehicle 500.

[0154] The network interface 524 may include a SoC that provides modulation and demodulation functions and enables the controller 536 to communicate over a wireless network. The network interface 524 may include an RF 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 well-known process and / or may be performed using a super-heterodyne process. In some examples, the RF front-end function may be provided by a separate chip. The network interface may include wireless functions for communicating over LTE, WCDMA, UMTS, GSM, CDMA2000, Bluetooth, Bluetooth LE, Wi-Fi, Z-wave, ZigBee, LoRaWAN, and / or other wireless protocols.

[0155] The vehicle 500 may further include data storage 528, which may include off-chip storage (e.g., outside the SoC 504). The data storage 528 may include one or more storage elements, including RAM, SRAM, DRAM, VRAM, flash memory, a hard disk, and / or other components and / or devices that can store at least one bit of data.

[0156] The vehicle 500 may further include a GNSS sensor 558. The GNSS sensor 558 (e.g., GPS, assisted GPS sensor, differential GPS (DGPS) sensor, etc.) is used to assist with mapping, perception, occupancy grid generation, and / or path planning functions. Any number of GNSS sensors 558 may be used, including, for example and without limitation, GPS using a USB connector with an Ethernet to serial (RS-232) bridge.

[0157] The vehicle 500 may further include a RADAR sensor 560. The RADAR sensor 560 may be used by the vehicle 500 for remote vehicle detection even in darkness and / or in adverse weather conditions. The RADAR functional safety level may be ASIL B. The RADAR sensor 560 may use CAN and / or bus 502 (e.g., to transmit data generated using the RADAR sensor 560) for control and access to object tracking data, and in some examples access Ethernet to access raw data. A variety of RADAR sensor types may be used. For example and without limitation, the RADAR sensor 560 may be suitable for front, rear, and side RADAR use. In some examples, a pulsed Doppler RADAR sensor is used.

[0158] The RADAR sensor 560 can 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 so on. In some examples, the long-range RADAR can be used for adaptive cruise control functions. The long-range RADAR system can provide a wide field of view (e.g., within 250m) achieved through two or more independent browsing. The RADAR sensor 560 can help distinguish between static objects and moving objects and can be used by the ADAS system for emergency braking assistance and forward collision warning. 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 the example with six antennas, the central four antennas can create a focused beam pattern that is designed to record the surroundings of the vehicle 500 at a higher rate with minimal traffic interference from adjacent lanes. The other two antennas can expand the field of view, making it possible to quickly detect vehicles entering or leaving the lane of the vehicle 500.

[0159] As an example, a medium-range RADAR system may include a range of up to 560m (front) or 80m (rear) and a field of view of up to 42 degrees (front) or 550 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 the rear bumper. When mounted on both ends of the rear bumper, such a RADAR sensor system can create two beams that continuously monitor the blind spots behind and beside the vehicle.

[0160] Short-range RADAR systems can be used in ADAS systems for blind spot detection and / or lane change assistance.

[0161] Vehicle 500 may further include ultrasonic sensors 562. Ultrasonic sensors 562, which may be located on the front, rear, and / or sides of vehicle 500, may be used for parking assistance and / or for creating and updating an occupancy grid. A variety of ultrasonic sensors 562 may be used, and different ultrasonic sensors 562 may have different detection ranges (e.g., 2.5 m, 4 m). Ultrasonic sensors 562 may operate at functional safety level ASIL B.

[0162] Vehicle 500 may include a LiDAR sensor 564. LiDAR sensor 564 may be used for object and pedestrian detection, emergency braking, collision avoidance, and / or other functions. LiDAR sensor 564 may be ASIL B functional safety level. In some examples, vehicle 500 may include multiple LiDAR sensors 564 (e.g., two, four, six, etc.) that may use Ethernet (e.g., to provide data to a Gigabit Ethernet switch).

[0163] In some examples, the LiDAR sensor 564 may be capable of providing a list of objects and their distances for a 360-degree field of view. Commercially available LiDAR sensors 564 may have, for example, an advertised range of approximately 500 meters, an accuracy of 2-3 cm, and support for 500 Mbps Ethernet connections. In some examples, one or more non-obtrusive LiDAR sensors 564 may be used. In such examples, the LiDAR sensor 564 may be implemented as a small device that can be embedded in the front, back, sides, and / or corners of the vehicle 500. In such an example, the LiDAR sensor 564 may provide a field of view of up to 120 degrees horizontally and 35 degrees vertically, with a range of 200 meters, even for low-reflectivity objects. The front-mounted LiDAR sensor 564 may be configured for a horizontal field of view between 45 and 135 degrees.

[0164] 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 200 m. The flash LiDAR unit includes a receiver that records the laser pulse transmission time and the reflected light on each pixel, which in turn corresponds to the range from the vehicle to the object. Flash LiDAR can allow a highly accurate and distortion-free image 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 500. 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 a fan. 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, the LiDAR sensor 564 may be less susceptible to motion blur, vibration, and / or shock.

[0165] The vehicle may further include an IMU sensor 566. In some examples, the IMU sensor 566 may be located at the center of the rear axle of the vehicle 500. The IMU sensor 566 may include, for example and without limitation, an accelerometer, a magnetometer, a gyroscope, a magnetic compass, and / or other sensor types. In some examples, such as in a six-axis application, the IMU sensor 566 may include an accelerometer and a gyroscope, while in a nine-axis application, the IMU sensor 566 may include an accelerometer, a gyroscope, and a magnetometer.

[0166] In some embodiments, the IMU sensor 566 can be implemented as a miniature, high-performance GPS-assisted inertial navigation system (GPS / INS) that combines micro-electromechanical systems (MEMS) inertial sensors, a high-sensitivity GPS receiver, and advanced Kalman filtering algorithms to provide estimates of position, velocity, and attitude. Thus, in some examples, the IMU sensor 566 can enable the vehicle 500 to estimate heading without input from a magnetic sensor by directly observing and correlating velocity changes from the GPS to the IMU sensor 566. In some examples, the IMU sensor 566 and the GNSS sensor 558 can be combined into a single integrated unit.

[0167] The vehicle may include microphones 596 positioned in and / or around the vehicle 500. The microphones 596 may be used for, among other things, emergency vehicle detection and identification.

[0168] The vehicle may further include any number of camera types, including stereo cameras 568, wide angle cameras 570, infrared cameras 572, surround cameras 574, long and / or medium range cameras 598, and / or other camera types. These cameras may be used to capture image data around the entire periphery of the vehicle 500. The type of camera used depends on the embodiment and the requirements of the vehicle 500, and any combination of camera types may be used to provide the necessary coverage around the vehicle 500. Additionally, the number of cameras may 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 may support Gigabit Multimedia Serial Link (GMSL) and / or Gigabit Ethernet. Each of the cameras described herein may include a 10GbE GPIO (100GbE) ... Figure 5A and Figure 5B Described in more detail.

[0169] Vehicle 500 may further include a vibration sensor 542. Vibration sensor 542 can measure vibrations of vehicle components, such as axles. For example, changes in vibration can indicate changes in the road surface. In another example, when two or more vibration sensors 542 are used, the difference between the vibrations can be used to determine friction or slippage of the road surface (e.g., when there is a vibration difference between a powered drive shaft and a freely rotating shaft).

[0170] The vehicle 500 may include an ADAS system 538. In some examples, the ADAS system 538 may include a SoC. The ADAS system 538 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.

[0171] The ACC system can utilize RADAR sensor 560, LiDAR sensor 564, and / or a camera. The ACC system can include longitudinal ACC and / or lateral ACC. Longitudinal ACC monitors and controls the distance to the vehicle immediately in front of vehicle 500, automatically adjusting the vehicle speed to maintain a safe distance from the vehicle in front. Lateral ACC maintains distance and, when necessary, recommends that vehicle 500 change lanes. Lateral ACC is related to other ADAS applications such as LCA and CWS.

[0172] CACC uses information from other vehicles, which can be received from other vehicles indirectly via a wireless link or through a network connection (e.g., through the Internet) via the network interface 524 and / or wireless antenna 526. 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 500 and in the same lane as it), while the I2V communication concept provides information about traffic further ahead. A CACC system can include either or both I2V and V2V information sources. Given information about the vehicle ahead of the vehicle 500, CACC can be more reliable, and it has the potential to improve the smoothness of traffic flow and reduce road congestion.

[0173] 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-facing camera and / or RADAR sensor 560 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 component. The FCW system can provide warnings in the form of, for example, audible, visual warnings, vibrations, and / or rapid brake pulses.

[0174] The AEB system detects an impending forward 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-facing camera and / or RADAR sensor 560 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 can include technologies such as dynamic brake support and / or collision approach braking.

[0175] The LDW system provides visual, audible, and / or tactile warnings, such as steering wheel or seat vibrations, to alert the driver when the vehicle 500 crosses a lane marking. When the driver indicates an intention to leave the lane by activating a turn signal, the LDW system is deactivated. The LDW system may utilize 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 vibration components.

[0176] The LKA system is a variation of the LDW system. If the vehicle 500 begins to leave the lane, the LKA system provides steering input or braking to correct the vehicle 500.

[0177] The BSW system detects and warns the driver of vehicles in the car's blind spot. The BSW system can provide visual, audible, and / or tactile alerts to indicate that merging or changing lanes is unsafe. The system can provide additional warnings when the driver uses a turn signal. The BSW system can use a rear-facing camera and / or RADAR sensor 560 coupled to a dedicated processor, DSP, FPGA, and / or ASIC that is electrically coupled to driver feedback such as a display, speaker, and / or vibration component.

[0178] The RCTW system can provide visual, audible, and / or tactile notifications when an object is detected outside the range of the rear-mounted camera while the vehicle 500 is in reverse. Some RCTW systems include AEB to ensure that the vehicle brakes are applied to avoid a collision. The RCTW system can use one or more rear-mounted RADAR sensors 560 coupled to a dedicated processor, DSP, FPGA, and / or ASIC that is electrically coupled to driver feedback such as a display, speaker, and / or vibration component.

[0179] Conventional ADAS systems may be prone to false positive results, which may be annoying and distracting to the driver, but are typically not catastrophic because the ADAS system alerts the driver and allows the driver to decide whether a safety condition actually exists and take action accordingly. However, in the autonomous vehicle 500, in the event of conflicting results, the vehicle 500 itself must decide whether to pay attention to the results from the main computer or the auxiliary computer (e.g., the first controller 536 or the second controller 536). For example, in some embodiments, the ADAS system 538 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 failures in perception and dynamic driving tasks. The output from the ADAS system 538 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 reconcile the conflict to ensure safe operation.

[0180] In some examples, the primary computer can be configured to provide a confidence score to the supervisory MCU, indicating the primary computer's confidence in the selected result. If the confidence score exceeds a threshold, the supervisory MCU can follow the primary computer's direction, regardless of whether the secondary computer provides conflicting or inconsistent results. In the event that the confidence score does not meet the threshold and the primary and secondary computers indicate different results (e.g., a conflict), the supervisory MCU can arbitrate between these computers to determine the appropriate result.

[0181] The supervisory MCU can be configured to run a neural network that is trained and configured to determine, based on outputs from the primary and secondary computers, conditions under which the secondary computer provides a false alarm. Thus, the neural network in the supervisory MCU can learn when the output of the secondary computer can be trusted and when it cannot. For example, when the secondary computer is a RADAR-based FCW system, the neural network in the supervisory MCU can learn when the FCW system is identifying a metal object that is not actually a danger, such as a drain grate or manhole cover that triggers an alarm. Similarly, when the secondary computer is a camera-based LDW system, the neural network in the supervisory MCU can learn to disregard the LDW when a cyclist or pedestrian is present and lane departure is actually the safest strategy. In embodiments that include a neural network running on the supervisory MCU, the supervisory MCU can include at least one of a DLA or a GPU suitable for running the neural network with associated memory. In preferred embodiments, the supervisory MCU can include and / or be included as a component of the SoC 504.

[0182] In other examples, the ADAS system 538 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 with respect to failures caused by software (or software-hardware interface) functions. For example, if there is a software vulnerability or bug 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 did not cause a substantial error.

[0183] In some examples, the output of ADAS system 538 can be fed into the primary computer's perception block and / or the primary computer's dynamic driving task block. For example, if ADAS system 538 indicates a forward collision warning due to an object immediately ahead, the perception block can use this information when identifying the object. In other examples, the secondary computer can have its own neural network that is trained and thus reduces the risk of false positives as described herein.

[0184] The vehicle 500 may further include an infotainment SoC 530 (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 separate components. The infotainment SoC 530 may include a combination of hardware and software that can 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., a 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 500. For example, the infotainment SoC 530 may include a radio, a disc player, a navigation system, a video player, USB and Bluetooth connectivity, an onboard computer, in-vehicle entertainment, Wi-Fi, steering wheel audio controls, hands-free voice controls, a head-up display (HUD), an HMI display 534, 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 530 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 538, 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.

[0185] The infotainment SoC 530 may include GPU functionality. The infotainment SoC 530 may communicate with other devices, systems, and / or components of the vehicle 500 via a bus 502 (e.g., a CAN bus, Ethernet, etc.). In some examples, the infotainment SoC 530 may be coupled to a supervisory MCU so that in the event of a failure of a primary controller 536 (e.g., a primary and / or backup computer of the vehicle 500), the infotainment system's GPU may perform some self-driving functions. In such an example, the infotainment SoC 530 may place the vehicle 500 in a driver-safe parking mode as described herein.

[0186] The vehicle 500 may further include an instrument cluster 532 (e.g., a digital instrument panel, an electronic instrument cluster, a digital instrument panel, etc.). The instrument cluster 532 may include a controller and / or a supercomputer (e.g., a separate controller or a supercomputer). The instrument cluster 532 may include a set of instruments, such as a speedometer, fuel level, 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 check light, 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 530 and the instrument cluster 532. Thus, the instrument cluster 532 may be included as part of the infotainment SoC 530, or vice versa.

[0187] Figure 5D For cloud-based servers and Figure 5A Schematic diagram of a system for communicating between an example autonomous vehicle 500. System 576 may include a server 578, a network 590, and a vehicle including vehicle 500. Server 578 may include multiple GPUs 584(A)-584(H) (collectively referred to herein as GPUs 584), PCIe switches 582(A)-582(D) (collectively referred to herein as PCIe switches 582), and / or CPUs 580(A)-580(B) (collectively referred to herein as CPUs 580). GPUs 584, CPUs 580, and PCIe switches may be interconnected with a high-speed interconnect and / or PCIe connection 586, such as, for example and without limitation, an NVLink interface 588 developed by NVIDIA. In some examples, GPUs 584 are connected via NVLink and / or NVSwitch SoCs, and GPUs 584 and PCIe switches 582 are connected via a PCIe interconnect. Although eight GPUs 584, two CPUs 580, and two PCIe switches are illustrated, this is not intended to be limiting. Depending on the embodiment, each of the servers 578 can include any number of GPUs 584, CPUs 580, and / or PCIe switches. For example, each of the servers 578 can include eight, sixteen, thirty-two, and / or more GPUs 584.

[0188] Server 578 can receive image data from a vehicle via network 590, the image data representing images showing unexpected or changed road conditions, such as recently begun road construction. Server 578 can transmit neural network 592, updated neural network 592, and / or map information 594, including information about traffic and road conditions, via network 590 and to the vehicle. Updates to map information 594 can include updates to HD map 522, such as information about construction sites, potholes, curves, flooding, or other obstacles. In some examples, neural network 592, updated neural network 592, and / or map information 594 can be generated from new training and / or data received from any number of vehicles in the environment and / or based on experience with training performed at a data center (e.g., using server 578 and / or other servers).

[0189] Server 578 can be used to train a machine learning model (e.g., a neural network) based on training data. The training data can be generated using the vehicle and / or can be generated in simulation (e.g., using a game engine). In some examples, the training data is labeled (e.g., in cases where 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., in cases where 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: supervised training, semi-supervised training, unsupervised training, self-learning, reinforcement learning, federated learning, transfer learning, feature learning (including principal component and cluster analysis), multilinear subspace learning, manifold learning, representation learning (including alternative 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 590), and / or the machine learning model can be used by server 578 to remotely monitor the vehicle.

[0190] In some examples, server 578 can receive data from the vehicle and apply the data to the latest real-time neural network for real-time intelligent reasoning. Server 578 can include a deep learning supercomputer and / or a dedicated AI computer powered by GPU 584, such as the DGX and DGX Station machines developed by NVIDIA. However, in some examples, server 578 can include the deep learning infrastructure of a data center using only CPU power.

[0191] The deep learning infrastructure of server 578 may be capable of rapid real-time reasoning and may use this capability to assess and verify the health of the processors, software, and / or associated hardware in vehicle 500. For example, the deep learning infrastructure may receive periodic updates from vehicle 500, such as a sequence of images and / or objects that vehicle 500 has located in the sequence of images (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 500, and if the results do not match and the infrastructure concludes that the AI ​​in vehicle 500 has malfunctioned, server 578 may transmit a signal to vehicle 500 instructing the fail-safe computer of vehicle 500 to take control, notify passengers, and complete a safe parking maneuver.

[0192] For inference, the server 578 may include a GPU 584 and one or more programmable inference accelerators (e.g., NVIDIA's TensorRT 3). The combination of GPU-powered servers and inference acceleration can enable real-time responses. In other examples, such as where performance is less important, CPU, FPGA, and other processor-powered servers can be used for inference.

[0193] Example computing device

[0194] Figure 6 FIG6 is a block diagram of an example computing device 600 suitable for implementing some embodiments of the present disclosure. Computing device 600 may include an interconnect system 602 that directly or indirectly couples the following devices: memory 604, one or more central processing units (CPUs) 606, one or more graphics processing units (GPUs) 608, a communication interface 610, input / output (I / O) ports 612, an I / O component 614, a power supply 616, one or more presentation components 618 (e.g., a display), and one or more logic units 620. In at least one embodiment, computing device 600 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, one or more GPUs 608 may include one or more vGPUs, one or more CPUs 606 may include one or more vCPUs, and / or one or more logic units 620 may include one or more virtual logic units. Thus, computing device 600 may include discrete components (eg, a complete GPU dedicated to computing device 600 ), virtual components (eg, a portion of a GPU dedicated to computing device 600 ), or a combination thereof.

[0195] although Figure 6The various blocks of are shown as being connected via an interconnect system 602 having wires, but this is not intended to be limiting and is provided for clarity only. For example, in some embodiments, a presentation component 618 such as a display device may be considered an I / O component 614 (e.g., if the display is a touch screen). As another example, the CPU 606 and / or the GPU 608 may include memory (e.g., the memory 604 may represent a storage device in addition to the memory of the GPU 608, the CPU 606, and / or the other components). Thus, Figure 6 The term computing device is illustrative only. No distinction is made between categories such as "workstation," "server," "laptop," "desktop," "tablet," "client device," "mobile device," "handheld device," "game console," "electronic control unit (ECU)," "virtual reality system," and / or other device or system types, as all are considered within the Figure 6 within the range of computing devices.

[0196] Interconnect system 602 can represent one or more links or buses, such as an address bus, a data bus, a control bus, or a combination thereof. Interconnect system 602 can 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, CPU 606 can be directly connected to memory 604. In addition, CPU 606 can be directly connected to GPU 608. In the case where there is a direct or point-to-point connection between components, interconnect system 602 can include a PCIe link to perform the connection. In these examples, it is not necessary to include a PCI bus in computing device 600.

[0197] Memory 604 may include any of a variety of computer-readable media. Computer-readable media can be any available media that can be accessed by computing device 600. Computer-readable media can include volatile and non-volatile media and removable and non-removable media. By way of example and not limitation, computer-readable media can include computer storage media and communication media.

[0198] Computer storage media may include volatile and non-volatile 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 604 may store computer-readable instructions (e.g., representing programs and / or program elements, such as an operating system). Computer storage media may include, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage devices, magnetic cassettes, magnetic tape, magnetic disk storage devices or other magnetic storage devices, or any other medium that can be used to store the desired information and that can be accessed by the computing device 600. As used herein, computer storage media does not include signals themselves.

[0199] Computer storage media may include 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 transport 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 manner as to encode information 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 acoustic, RF, infrared, and other wireless media. Any combination of the above should also be included within the scope of computer-readable media.

[0200] The CPU 606 can be configured to execute at least some of the computer-readable instructions to control one or more components of the computing device 600 to perform one or more of the methods and / or processes described herein. Each of the CPUs 606 can 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 606 can include any type of processor and can include different types of processors, depending on the type of computing device 600 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 600, the processor can be an Advanced RISC (ARM) processor implemented using Reduced Instruction Set Computing (RISC) or an x86 processor implemented using Complex Instruction Set Computing (CISC). The computing device 600 can also include one or more CPUs 606 in addition to one or more microprocessors or supplementary coprocessors such as math coprocessors.

[0201] In addition to or in place of the CPU 606, the GPU 608 may also be configured to execute at least some computer-readable instructions to control one or more components of the computing device 600 to perform one or more of the methods and / or processes described herein. One or more GPUs 608 may be integrated GPUs (e.g., with one or more CPUs 606) and / or one or more GPUs 608 may be discrete GPUs. In embodiments, one or more GPUs 608 may be coprocessors for one or more CPUs 606. The computing device 600 may use the GPU 608 to render graphics (e.g., 3D graphics) or perform general-purpose computing. For example, the GPU 608 may be used for general-purpose computing on a GPU (GPGPU). The GPU 608 may include hundreds or thousands of cores capable of processing hundreds or thousands of software threads simultaneously. The GPU 608 may generate pixel data for outputting an image in response to a rendering command (e.g., a rendering command received from the CPU 606 via a host interface). The GPU 608 may include graphics memory, such as display memory, for storing pixel data or any other suitable data (e.g., GPGPU data). Display memory can be included as part of memory 604. GPU 608 can include two or more GPUs operating in parallel (e.g., via a link). The link can connect the GPUs directly (e.g., using NVLINK) or through a switch (e.g., using NVSwitch). When combined, each GPU 608 can generate pixel data or GPGPU data for different portions or different outputs (e.g., a first GPU for a first image and a second GPU for a second image). Each GPU can include its own memory or can share memory with other GPUs.

[0202] In addition to or in lieu of the CPU 606 and / or GPU 608, the logic unit 620 may be configured to execute at least some computer-readable instructions to control one or more components of the computing device 600 to perform one or more methods and / or processes described herein. In embodiments, the CPU 606, GPU 608, and / or logic unit 620 may perform any combination of methods, processes, and / or portions thereof, either separately or in conjunction. The one or more logic units 620 may be part of and / or integrated within the one or more CPUs 606 and / or the one or more GPUs 608, and / or the one or more logic units 620 may be discrete components of or otherwise external to the CPU 606 and / or GPU 608. In embodiments, the one or more logic units 620 may be processors of the one or more CPUs 606 and / or the one or more GPUs 608.

[0203] Examples of logic unit 620 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.

[0204] The communication interface 610 may include one or more receivers, transmitters, and / or transceivers that enable the computing device 600 to communicate with other computing devices via an electronic communication network, including wired and / or wireless communications. The communication interface 610 may include components and functionality 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 one or more logic units 620 and / or the communication interface 610 may include one or more data processing units (DPUs) for transmitting data received over the network and / or interconnect system 602 directly to one or more GPUs 608 (e.g., memory of a GPU 608).

[0205] The I / O ports 612 can enable the computing device 600 to be logically coupled to other devices including I / O components 614, presentation components 618, and / or other components, some of which can be built into (e.g., integrated into) the computing device 600. Illustrative I / O components 614 include a microphone, a mouse, a keyboard, a joystick, a game pad, a game controller, a satellite dish, a browser, a printer, a wireless device, and the like. The I / O components 614 can provide a natural user interface (NUI) that processes user-generated air gestures, voice, or other physiological input. 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 600 (as described in more detail below). The computing device 600 can include a depth camera such as a stereo camera system, an infrared camera system, an RGB camera system, touch screen technology, and combinations thereof for gesture detection and recognition. Additionally, computing device 600 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 600 to render immersive augmented or virtual reality.

[0206] The power supply 616 may include a hardwired power supply, a battery power supply, or a combination thereof. The power supply 616 may provide power to the computing device 600 to enable the components of the computing device 600 to operate.

[0207] The presentation component 618 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), speakers, and / or other presentation components. The presentation component 618 may receive data from other components (e.g., the GPU 608, the CPU 606, the DPU, etc.) and output the data (e.g., as images, video, sound, etc.).

[0208] Sample Data Center

[0209] Figure 7 An example data center 700 is shown, which may be used in at least one embodiment of the present disclosure. The data center 700 may include a data center infrastructure layer 710, a framework layer 720, a software layer 730, and an application layer 740.

[0210] like Figure 7As shown, the data center infrastructure layer 710 may include a resource coordinator 712, grouped computing resources 714, and node computing resources ("node CRs") 716(1)-716(N), where "N" represents any complete positive integer. In at least one embodiment, the node CRs 716(1)-716(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 memories), 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 716(1)-716(N) may correspond to a server having one or more of the above-mentioned computing resources. Furthermore, in some embodiments, nodes CR 716 ( 1 )- 716 (N) may include one or more virtual components, such as vGPUs, vCPUs, etc., and / or one or more of nodes CR 716 ( 1 )- 716 (N) may correspond to virtual machines (VMs).

[0211] In at least one embodiment, the computing resources 714 of the grouping can include a separate grouping (not shown) of the node CR716 housed in one or more racks, or a lot of racks (also not shown) housed in the data center of each geographical location. The separate grouping of the node CR716 in the computing resources 714 of the grouping can 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 CR716 comprising CPU, GPU, DPU and / or other processors can be grouped in one or more racks to provide computing resources to support one or more workloads. One or more racks can also include any number of power modules, cooling modules and / or network switches in any combination.

[0212] Resource coordinator 712 may configure or otherwise control one or more nodes CR 716(1)-716(N) and / or grouped computing resources 714. In at least one embodiment, resource coordinator 712 may comprise a software design infrastructure ("SDI") management entity for data center 700. Resource coordinator 712 may comprise hardware, software, or some combination thereof.

[0213] In at least one embodiment, Figure 7As shown, the framework layer 720 may include a job scheduler 733, a configuration manager 734, a resource manager 736, and a distributed file system 738. The framework layer 720 may include a framework that supports the software 732 of the software layer 730 and / or one or more applications 742 of the application layer 740. The software 732 or the application 742 may include web-based service software or applications, such as those provided by Amazon Web Services, Google Cloud, and Microsoft Azure. The framework layer 720 may be, but is not limited to, a free and open source software network application framework, such as Apache Spark, which can utilize the distributed file system 738 for large-scale data processing (e.g., "big data"). TM (hereinafter referred to as "Spark"). In at least one embodiment, the job scheduler 733 may include a Spark driver to facilitate scheduling of workloads supported by the various layers of the data center 700. In at least one embodiment, the configuration manager 734 may be capable of configuring different layers, such as the software layer 730 and the framework layer 720 including Spark and a distributed file system 738 for supporting large-scale data processing. The resource manager 736 may be capable of managing the mapping or allocation of clustered or grouped computing resources to support the distributed file system 738 and the job scheduler 733. In at least one embodiment, the clustered or grouped computing resources may include the grouped computing resources 714 at the data center infrastructure layer 710. The resource manager 736 may coordinate with the resource coordinator 712 to manage these mapped or allocated computing resources.

[0214] In at least one embodiment, the software 732 included in the software layer 730 may include software used by at least a portion of the node CRs 716(1)-716(N), the grouped computing resources 714, and / or the distributed file system 738 of the framework layer 720. The 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.

[0215] In at least one embodiment, the one or more applications 742 included in the application layer 740 may include one or more types of applications used by at least a portion of the node CRs 716(1)-716(N), the grouped computing resources 714, and / or the distributed file system 738 of the framework layer 720. 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.

[0216] In at least one embodiment, any of the configuration manager 734, resource manager 736, and resource coordinator 712 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 the data center 700 from making potentially poor configuration decisions and can avoid underutilized and / or poorly performing portions of the data center.

[0217] The data center 700 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 above with respect to the data center 700. In at least one embodiment, using the weight parameters calculated by one or more training techniques, the resources described above with respect to the data center 700 may be used to infer or predict information using a trained machine learning model corresponding to one or more neural networks, such as but not limited to those described herein.

[0218] In at least one embodiment, the data center 700 may use a CPU, an application-specific integrated circuit (ASIC), a GPU, an FPGA, and / or other hardware (or corresponding virtual computing resources) to perform training and / or reasoning using the aforementioned resources. In addition, one or more of the aforementioned software and / or hardware resources may 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.

[0219] Sample network environment

[0220] 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: Figure 6 The backend device 700 may be implemented on one or more instances of the computing device 600 of the embodiment of the present invention—for example, each device may include similar components, features and / or functions of the computing device 600. 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 700, examples of which are described herein with respect to FIG. Figure 7 Describe in more detail.

[0221] 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 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 (e.g., 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.

[0222] 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.

[0223] 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. The 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 that may, for example, use a distributed file system for large-scale data processing (e.g., "big data").

[0224] 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 described herein (or one or more portions thereof). 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 a state, region, country, global, 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).

[0225] Client devices may include Figure 6 The client device 600 may be embodied as a personal computer (PC), a laptop computer, a mobile device, a smartphone, a tablet computer, a smartwatch, 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 watercraft, 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.

[0226] The present disclosure can be described in the general context of machine-usable instructions or computer code 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. Generally, program modules including routines, programs, objects, components, data structures, etc. refer to code that performs a specific task or implements a specific abstract data type. The present disclosure can be practiced in a variety of system configurations, including handheld devices, consumer electronics, general-purpose computers, more specialized computing devices, etc. The present disclosure can also be practiced in a distributed computing environment where tasks are performed by remote processing devices linked through a communication network.

[0227] As used herein, the statement "and / or" with respect to two or more elements should be interpreted as referring to only one element or 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.

[0228] 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 that are 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.

[0229] Sample text support

[0230] In an example embodiment, one or more processors include processing circuitry configured to: generate two or more image frames based at least on sensor data corresponding to a first time slice and generated using one or more sensors of an ego machine in an environment; switch from enabling a first environment modeling pipeline that models the environment as a three-dimensional (3D) bowl model to enabling a second environment modeling pipeline that models the environment as a detected 3D surface topology based at least on a state of the ego machine; and generate a visualization of the environment based at least on applying the two or more image frames to the second environment modeling pipeline.

[0231] In some embodiments, the second environment modeling pipeline models the environment as a detected 3D surface topology based at least on a point cloud.

[0232] In some embodiments, the second environment modeling pipeline models the environment as a detected 3D surface topology based at least on a truncated signed distance function (TSDF).

[0233] In some embodiments, the processing circuit is further configured to switch from enabling the first environment modeling pipeline to enabling the second environment modeling pipeline based at least on detecting that the state of the ego machine includes a detected proximity to a detected pedestrian.

[0234] In some embodiments, the processing circuit is further configured to switch from enabling the first environment modeling pipeline to enabling the second environment modeling pipeline based at least on detecting that the state of the ego machine includes a detected depth to an object of an unsupported class.

[0235] In some embodiments, the processing circuit is further configured to: switch from enabling the first environment modeling pipeline to enabling the second environment modeling pipeline based on detecting that the state of the self-machine is operating in a parking scenario or is predicted to start operating in a parking scenario.

[0236] In some embodiments, the processing circuit is further configured to: present an indication that one or more visualizations associated with the second environment pipeline are available based at least on detecting that a measurement of an estimated image quality of the visualization is above a specified threshold, wherein the processing circuit is further configured to: generate the visualization in response to a selection of a visualization from the one or more visualizations by an operator of the self-machine.

[0237] In some embodiments, the processing circuit is further configured to: present an indication that one or more visualizations associated with the second environmental pipeline are available based at least on detecting that the state of the self-machine includes a speed of self-motion below a threshold speed, wherein the processing circuit is further configured to: generate the visualization in response to a selection of a visualization from the one or more visualizations by an operator of the self-machine.

[0238] In some embodiments, the processing circuit is further configured to generate a visualization using a simulated representation of the detected object based at least on detecting at least one of: the detected object being occluded, or the detected object being located between the self-machine and a virtual camera used to generate the visualization.

[0239] In some embodiments, the processing circuitry is included in at least one of: a control system for an autonomous or semi-autonomous machine; a perception system for an autonomous or semi-autonomous machine; a system for performing simulation operations; a system for performing digital twin operations; a system for performing light transport simulations; a system for performing collaborative content creation of 3D assets; a system for performing deep learning operations; a system for performing remote operations; a system for performing real-time streaming; a system for generating or presenting one or more of augmented reality content, virtual reality content, or mixed reality content; a system implemented using an edge device; a system implemented using a robot; a system for performing conversational AI operations; a system implementing one or more language models; a system implementing one or more large language models (LLMs); a system for generating synthetic data; a system for generating synthetic data using AI; a system comprising one or more virtual machines (VMs); a system implemented at least in part in a data center; or a system implemented at least in part using cloud computing resources.

[0240] In an example embodiment, a system includes one or more processors configured to switch between enabling a first environment modeling pipeline that models the environment as a three-dimensional (3D) bowl and enabling a second environment modeling pipeline that models the environment as a detected 3D surface topology based at least on a detected state of an ego machine in the environment.

[0241] In some embodiments, the second environment modeling pipeline models the environment as a detected 3D surface topology based at least on a point cloud.

[0242] In some embodiments, the second environment modeling pipeline models the environment as a detected 3D surface topology based at least on a truncated signed distance function (TSDF).

[0243] In some embodiments, the processing circuit is further configured to switch between enabling the first environment modeling pipeline and enabling the second environment modeling pipeline based at least on determining that the detected state of the ego machine indicates a specified operating scenario.

[0244] In some embodiments, the one or more processors are further configured to present an indication that one or more visualizations associated with the second environment modeling pipeline are available based at least on detecting that a measurement of an estimated image quality of the visualization associated with the second environment modeling pipeline is above a specified threshold, wherein the processing circuit is further configured to generate the visualization in response to a selection of a visualization from the one or more visualizations by an operator of the self-machine.

[0245] In some embodiments, the one or more processors are further configured to: present an indication that one or more visualizations associated with the second environmental pipeline are available based at least on a determination that the detected state of the self-machine includes a self-motion speed below a threshold speed, wherein the processing circuit is further configured to: generate a visualization associated with the second environmental pipeline in response to a selection of a visualization from the one or more visualizations by an operator of the self-machine.

[0246] In some embodiments, the one or more processors are further configured to generate a visualization associated with the second environment modeling pipeline using a simulated representation of the detected object based at least on detecting at least one of: the detected object being occluded or the detected object being located between the self-machine and a virtual camera used to generate the visualization.

[0247] In some embodiments, the system is included in at least one of: a control system for an autonomous or semi-autonomous machine; a perception system for an autonomous or semi-autonomous machine; a system for performing simulation operations; a system for performing digital twin operations; a system for performing light transport simulations; a system for performing collaborative content creation of 3D assets; a system for performing deep learning operations; a system for performing remote operations; a system for performing real-time streaming; a system for generating or presenting one or more of augmented reality content, virtual reality content, or mixed reality content; a system implemented using an edge device; a system implemented using a robot; a system for performing conversational AI operations; a system implementing one or more language models; a system implementing one or more large language models (LLMs); a system for generating synthetic data; a system for generating synthetic data using AI; a system comprising one or more virtual machines (VMs); a system implemented at least in part in a data center; or a system implemented at least in part using cloud computing resources.

[0248] In one example embodiment, a method includes: determining, based at least on a detected state of an ego machine in an environment, whether to enable a first supported environment modeling pipeline that models the environment as a three-dimensional (3D) bowl or a second supported environment modeling pipeline that models the environment without a 3D bowl; and performing an operation based on at least the first supported environment modeling pipeline or the second supported environment modeling pipeline.

[0249] In some embodiments, the method is performed by at least one of: a control system for an autonomous or semi-autonomous machine; a perception system for an autonomous or semi-autonomous machine; a system for performing simulation operations; a system for performing digital twin operations; a system for performing light transport simulations; a system for performing collaborative content creation of 3D assets; a system for performing deep learning operations; a system for performing remote operations; a system for performing real-time streaming; a system for generating or presenting one or more of augmented reality content, virtual reality content, or mixed reality content; a system implemented using an edge device; a system implemented using a robot; a system for performing conversational AI operations; a system implementing one or more language models; a system implementing one or more large language models (LLMs); a system for generating synthetic data; a system for generating synthetic data using AI; a system comprising one or more virtual machines (VMs); a system implemented at least in part in a data center; or a system implemented at least in part using cloud computing resources.

Claims

1. One or more processors, including processing circuitry configured to: generating two or more image frames based at least on sensor data corresponding to a first time slice and generated using one or more sensors of the ego-machine in the environment; Based at least on the state of the ego machine, switching from generating a first visualization of the environment using a three-dimensional 3D bowl model of the environment to generating a second visualization of the environment using a 3D surface topology model of the environment; and The second visualization of the environment is rendered based at least on the two or more image frames.

2. The one or more processors of claim 1 , wherein the second visualization of the environment is generated by modeling the environment as a 3D surface topology of one or more objects in the environment based at least on a point cloud corresponding to the one or more objects.

3. The one or more processors of claim 1 , wherein the second visualization of the environment is generated by modeling the environment as a 3D surface topology of one or more objects in the environment based at least on a truncated signed distance function (TSDF) representing a distance between the ego machine and the one or more objects in the environment.

4. The one or more processors of claim 1 , the processing circuitry further configured to: switch from generating the first visualization to generating the second visualization based at least on detecting that the state of the ego machine includes a detected proximity to a detected vulnerable road user (VRU).

5. The one or more processors of claim 1 , the processing circuitry further configured to switch from generating the first visualization to generating the second visualization based at least on detecting that the state of the ego-machine includes a detected depth to an object of an unsupported category.

6. The one or more processors according to claim 1, the processing circuit being further configured to: switch from generating the first visualization to generating the second visualization based on detecting that the state of the self-machine is operating in a parking scenario or is predicted to begin operating in a parking scenario.

7. The one or more processors of claim 1, the processing circuitry further configured to cause presentation of an indication that the second visualization is available based at least on detecting that a measure of estimated image quality of the second visualization is above a specified threshold.

8. The one or more processors of claim 1, the processing circuitry further configured to cause presentation of an indication that the second visualization is available based at least on detecting that the state of the ego-machine includes a speed of ego-motion below a threshold speed.

9. The one or more processors of claim 1 , the processing circuitry further configured to generate the second visualization using a simulated representation of the detected object based at least on detecting at least one of: the detected object being occluded, or the detected object being located between the ego machine and a virtual camera used to generate the second visualization.

10. The one or more processors of claim 1 , wherein the one or more processors are included in at least one of: 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; systems for performing remote operations; Systems for performing real-time streaming; Systems for generating or presenting one or more of augmented reality content, virtual reality content, or mixed reality content; Systems implemented using edge devices; Systems implemented using robots; Systems for performing conversational AI operations; A system implementing one or more language models; A system implementing one or more large language models (LLMs); Systems for generating synthetic data; Systems for generating synthetic data using AI; 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.

11. A system comprising one or more processors configured to switch between generating a first visualization of the environment using a three-dimensional 3D bowl model of the environment and generating a second visualization of the environment using a 3D surface topology model of the environment based at least on a detected state of a self-machine in the environment. 12 . The system of claim 11 , wherein the second visualization of the environment is generated by modeling the environment as a detected 3D surface topology based at least on a point cloud. 13 . The system of claim 11 , wherein the second visualization of the environment is generated by modeling the environment as a detected 3D surface topology based at least on a truncated signed distance function (TSDF).

14. The system of claim 11, the one or more processors further configured to switch between generating the first visualization and generating the second visualization based at least on determining that the detected state of the ego machine indicates a specified operating scenario.

15. The system of claim 11, the one or more processors further configured to cause presentation of an indication that the second visualization is available based at least on detecting that a measure of estimated image quality of a visualization associated with a second environment modeling pipeline is above a specified threshold.

16. The system of claim 11, the one or more processors further configured to cause presentation of an indication that the second visualization is available based at least on a determination that the detected state of the self-machine includes a speed of self-motion below a threshold speed.

17. The system of claim 11, wherein the one or more processors are further configured to generate the second visualization using a simulated representation of the detected object based at least on detecting at least one of: the detected object being occluded, or the detected object being located between the self-machine and a virtual camera used to generate the second visualization.

18. The system of claim 11, 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; systems for performing remote operations; A system for performing real-time streaming; Systems for generating or presenting one or more of augmented reality content, virtual reality content, or mixed reality content; Systems implemented using edge devices; Systems implemented using robots; Systems for performing conversational AI operations; A system implementing one or more language models; A system implementing one or more large language models (LLMs); Systems for generating synthetic data; Systems for generating synthetic data using AI; 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.

19. A method comprising: determining, based at least on a detected state of the ego-machine in an environment, whether to enable generating a first visualization of the environment using a three-dimensional (3D) bowl model of the environment or to enable generating a second visualization of the environment using a 3D surface topology model of the environment; and An operation is performed based on at least the presentation of the first visualization or the second visualization.

20. The method of claim 19, wherein the method is performed by at least one of: 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; systems for performing remote operations; A system for performing real-time streaming; Systems for generating or presenting one or more of augmented reality content, virtual reality content, or mixed reality content; Systems implemented using edge devices; Systems implemented using robots; Systems for performing conversational AI operations; A system implementing one or more language models; A system implementing one or more large language models (LLMs); Systems for generating synthetic data; Systems for generating synthetic data using AI; 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.

Images