Object detection using sensor fusion for autonomous systems and applications
By using machine learning models to process and fuse sensor data, the problems of manual tuning and high resource consumption required for sensor data fusion in existing systems have been solved, achieving more efficient and accurate data fusion.
Patent Information
- Application Number
- CN202410562208.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-05-09
- Filing Date
- 2024-05-08
- Publication Date
- 2024-11-12
AI Technical Summary
Existing vehicle sensor data fusion systems require a large number of heuristics and manual tuning parameters, which results in significant time and computational resources being spent when configuration changes occur, and they are difficult to adapt to the addition, removal or updating of sensors.
By employing machine learning models, especially deep neural networks, different types of sensor data are processed and information is automatically fused, reducing reliance on heuristics and manually designed components, and the model is automatically updated through training data.
It improves the accuracy and efficiency of sensor data fusion, reduces the time and computing resource requirements, and eliminates the need for additional updates when adapting to sensor changes.
Smart Images

Figure CN118940202A_ABST
Abstract
Description
Background Art
[0001] A vehicle or machine (e.g., an autonomous vehicle or machine, a semi-autonomous vehicle or machine, etc.) may use various types of sensors, such as image sensors, radar (RADAR) sensors, laser radar (LiDAR) sensors, ultrasonic sensors, etc., to perceive the environment around the vehicle or machine. For example, the vehicle may use a first processing pipeline to process first sensor data generated using a first type of sensor to determine first information associated with an object around the vehicle. The vehicle may also use a second processing pipeline to process second sensor data generated using a second type of sensor to determine second information associated with an object around the environment. In order to create a more robust perception system, the vehicle may then fuse the first information with the second information to determine fused information associated with the object.
[0002] For example, a vehicle may use sensor data fusion to track objects located around the vehicle. For example, whenever an object is detected using one of the sensors, the vehicle (e.g., a component of the vehicle, such as a multi-sensor fusion module) may attempt to associate the detected object with an existing object track. In some examples, this association is performed by calculating a cost, such as based on a measured distance between the detected object and an existing object track. For example, if the cost is high (e.g., above a threshold cost), the vehicle may generate a new object track for the detected object. However, if the cost is low (e.g., below a threshold cost), the vehicle may associate the detected object with an existing object track.
[0003] When performing such a process, the vehicle may associate objects detected using multiple sensors with the same existing object track. For example, and using the above example, if first information generated using a first type of sensor and second information generated using a second type of sensor are associated with the same object, the first information is fused with the second information to generate fused information of the object. The vehicle may then use the fused information to update the track associated with the object, such as updating the position of the object, the velocity of the object, the orientation of the object, and / or any other information associated with the object.
[0004] However, current systems used by vehicles perform a multi-step process in order to fuse information generated using different sensors. This may require a large number of heuristics to perform the association, along with manually tuned parameters to set thresholds for creating, updating, and / or terminating tracks. In addition, when the configuration of the fusion process changes, and / or when sensors of the vehicle are added, removed, and / or updated, the heuristics and / or parameters may need to be updated with other systems of the vehicle. Determining such heuristics and / or parameters and updating such heuristics and / or parameters may require the developer of the system to spend a lot of time and a lot of computing resources to implement. Summary of the invention
[0005] Embodiments of the present disclosure relate to sensor fusion for autonomous or semi-autonomous systems and applications. A system and method for using a data pipeline to process sensor data generated using sensors of different types or modalities (e.g., image sensors, RADAR sensors, LiDAR sensors, ultrasonic sensors, etc.) to generate first data representing information associated with objects around a vehicle is disclosed. For example, the information may represent the value of a parameter associated with an object, such as the position of the object, the size of the object, the speed of the object, the orientation or posture of the object, the classification of the object, and / or any other parameter. The system and method may then process the first data using one or more machine learning models (e.g., one or more deep neural networks), which are trained to fuse the information (e.g., the value of the parameter) and output second data representing the final information associated with the object. In some examples, the first data is processed before being input to one or more machine learning models. For example, the first data may be processed to generate a grid representing the parameter (e.g., a bird's-eye view grid), where the grid is then input to one or more machine learning models and processed by one or more machine learning models.
[0006] Compared to conventional systems (such as those described above), in some embodiments, the current system is able to perform data fusion without manually designing components (e.g., multi-sensor fusion modules) with heuristics, algorithms, and parameters. For example, designing such components may take many iterations to get the heuristics, algorithms, and parameters correct. In addition, whenever there is an update to the vehicle, such as an update to the system that uses the fused data and / or an update to one of the sensors that generate the data to be fused, the heuristics, algorithms, and / or parameters may also need to be updated. In contrast, the current system uses training data to train one or more machine learning models, one or more of which are then used to fuse the data. Training such a machine learning model to perform data fusion may require less time and / or computing resources. In addition, the trained machine learning model may not need to be updated with other systems and / or when sensors are added or removed. In addition, the trained machine learning model can be more accurate and precise when performing sensor data fusion than using manually designed or hard-coded components. BRIEF DESCRIPTION OF THE DRAWINGS
[0007] The present system and method for sensor fusion for autonomous or semi-autonomous systems and applications is described in detail below with reference to the accompanying drawings, in which:
[0008] Figure 1An example data flow diagram illustrating a process of performing data fusion using one or more machine learning models according to some embodiments of the present disclosure;
[0009] Figure 2 shows an example of a vehicle navigating around an environment according to some embodiments of the present disclosure;
[0010] FIG. 3A to FIG. 3B shows an example of a visualization of output associated with a processing pipeline according to some embodiments of the present invention, wherein the output indicates information associated with an object;
[0011] 4A to 4D shows an example visualization of generating inputs for one or more machine learning models according to some embodiments of the present disclosure;
[0012] FIG. 5A to FIG. 5B shows example visualizations of outputs generated using one or more machine learning models according to some embodiments of the present disclosure;
[0013] Figure 6 A data flow diagram illustrating a process for training one or more machine learning models to perform data fusion according to some embodiments of the present disclosure is shown;
[0014] Figure 7 A flowchart illustrating a method for performing data fusion using one or more machine learning models according to some embodiments of the present disclosure is shown;
[0015] Figure 8 A flowchart illustrating a method for generating input data for one or more machine learning models according to some embodiments of the present disclosure is shown;
[0016] Fig.9A is an illustration of an example autonomous vehicle according to some embodiments of the present disclosure;
[0017] Fig. 9B According to some embodiments of the present disclosure Fig.9A Examples of autonomous vehicle camera positions and fields of view;
[0018] Fig. 9C According to some embodiments of the present disclosure Fig.9A a block diagram of an example system architecture for an example autonomous vehicle;
[0019] Fig.9D is a cloud-based server and Fig.9A System diagram of an example of communication between autonomous vehicles;
[0020] Fig.10 is a block diagram of an example computing device suitable for implementing some embodiments of the present disclosure; and
[0021] Fig.11 is a block diagram of an example data center suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION
[0022] Systems and methods related to sensor fusion for autonomous or semi-autonomous systems and applications are disclosed. Although the present invention may be described with respect to an example autonomous or semi-autonomous vehicle or machine 900 (alternatively referred to herein as "vehicle 900" or "ego-machine 900"), the present invention may be described with respect to an example autonomous or semi-autonomous vehicle or machine 900 (alternatively referred to herein as "vehicle 900" or "ego-machine 900"). 9A to 9D The present disclosure may be described with reference to a plurality of examples thereof, 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 adaptive driver assistance systems (ADAS)), non-autonomous vehicles or machines, driven and undriven robots or robotic platforms, warehouse vehicles, off-road vehicles, vehicles coupled to one or more trailers, flying boats, boats, shuttles, emergency response vehicles, motorcycles, electric or motorized bicycles, airplanes, construction vehicles, underwater vehicles, drones, and / or other vehicle types. Furthermore, while the present disclosure may be described with respect to data fusion in machine applications, 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, security and surveillance, autonomous or semi-autonomous machine applications, and / or any other technology space where data fusion may be used.
[0023] For example, one or more systems may receive, generate and / or obtain sensor data generated using sensors of various types or modalities. As described herein, sensor data may include, but is not limited to, image data generated using one or more image sensors, RADAR data generated using one or more RADAR sensors, LiDAR data generated using one or more LiDAR sensors, ultrasonic data generated using one or more ultrasonic sensors, and / or any other type of sensor data generated using any other type of sensor. The system may then process the sensor data using a processing pipeline to determine information associated with the environment surrounding the vehicle. As described herein, information may include the value of a parameter associated with one or more objects located near the vehicle. For example, parameters associated with an object may include (but are not limited to) the position of the object (e.g., x-coordinate position, y-coordinate position, z-coordinate position, etc.), the size of the object (e.g., height, width, length, etc.), the speed of the object (e.g., speed in the x-coordinate direction, speed in the y-coordinate direction, etc.), the orientation or posture of the object, the classification of the object (e.g., vehicle, pedestrian, animal, structure, etc.) and / or any other parameter.
[0024] For example, the vehicle may use a first processing pipeline to process first sensor data generated using one or more first sensors. Based on the processing, the first processing pipeline may output first data representing first information associated with one or more first objects around the vehicle. The vehicle may also use a second processing pipeline to process second sensor data generated using one or more second sensors. Based on the processing, the second processing pipeline may output second data representing second information associated with one or more second objects around the vehicle. In some examples, one or more first objects are the same as one or more second objects. In some examples, at least one object from one or more first objects may be different from at least one object from one or more second objects. In any example, the sensor fusion process may include any number of sensor processing pipelines from any number and type of sensors.
[0025] One or more systems may then use the data output from the processing pipeline to generate input data. In some examples, the input data may represent one or more grids, such as one or more bird's eye view (BEV) or top view grids, that represent parameters associated with one or more objects. For example, and using the above example, if the first information represents the values of ten parameters associated with one or more first objects, one or more systems may use the first information to generate ten grids, each of which represents a corresponding type of parameter. In addition, if the second information also represents the values of ten parameters associated with one or more second objects, one or more systems may use the second information to generate ten additional grids, each of which again represents a corresponding type of parameter. Although this example describes information representing ten different types of parameters and a system that generates a corresponding grid for each type of parameter, in other examples, the information may represent any number of parameters and / or the system may generate a corresponding grid for more than one type of parameter (e.g., a combined parameter grid).
[0026] One or more systems can use one or more locations of one or more objects within the environment to generate a grid. For example, and for a grid, the grid can be divided into multiple parts (e.g., cells), where each part represents a corresponding area within the environment. For a first non-limiting example, if the entire grid represents a 400 meter by 400 meter segment of the environment, then the grid can be separated so that each part represents a 0.5 meter by 0.5 meter area of the environment. For a second non-limiting example, if the entire grid again represents a 400 meter by 400 meter segment of the environment, then the grid can be separated so that each part represents a 1 meter by 1 meter area of the environment. However, in other examples, the grid can represent different sized segments of the environment and / or the parts can each represent different sized areas of the environment.
[0027] One or more systems may then use one or more positions of one or more objects to input values of parameters within the grid. For example, and for an object, one or more systems may use the position of the object to determine that the object is associated with a portion of a grid (e.g., a portion representing an area of the environment in which the object is located). One or more systems may then input values of the parameters of the object into portions of the grid. For example, one or more systems may input a first value of a first parameter (e.g., an x-coordinate position) into a portion of a first grid, a second value of a second parameter (e.g., a y-coordinate position) into a corresponding portion of a second grid, a third value of a third parameter (e.g., a z-coordinate position) into a corresponding portion of a third grid, and so on. One or more systems may then perform a similar process to input values of parameters of one or more other objects (e.g., each object) into the grid.
[0028] In some examples, because portions of a grid represent areas of an environment, a single portion of a grid may be associated with the locations of multiple objects (which, in some examples, are referred to as "overlapping objects") located within an area of the environment. As such, one or more systems may perform one or more processes to represent overlapping objects within an environment. For example, and using the example of one or more systems generating ten grids for ten different parameters of an object as described above, one or more systems may generate a first set of ten grids for representing values of parameters associated with one or more first objects located within the environment and a second set of ten grids for representing values of parameters associated with one or more second objects located within the environment. Thus, if a portion is associated with two overlapping objects, the portion of the first set of grids represents the value of the parameter of the first object in the overlapping object, and the corresponding portion of the second set of grids represents the value of the parameter of the second object in the overlapping object.
[0029] The one or more systems may then input the input data into one or more machine learning models trained to perform data fusion. For example, the one or more machine learning models may process the input data and, based at least on the processing, output data representing fusion information associated with one or more objects located within the environment. In some examples, the data output by the one or more machine learning models may be similar to the data input into the one or more machine learning models. For example, the data output by the one or more machine learning models may represent multiple grids, where individual grids are associated with types of parameters. For example, and for an object, a portion of a first grid may represent a first value of a first parameter associated with the object (e.g., an x-coordinate position), a corresponding portion of a second grid may represent a second value of a second parameter associated with the object (e.g., a y-coordinate position), a corresponding portion of a third grid may represent a third value of a third parameter associated with the object (e.g., a z-coordinate position), and so on.
[0030] In some examples, one or more systems (and / or additional one or more systems) can train one or more machine learning models to perform data fusion. For example, and as described in more detail herein, one or more systems can train one or more machine learning models using at least training sensor data generated by different types of sensors and truth data representing actual values of parameters associated with objects represented by the training sensor data. In addition, during training, one or more systems can generate and input data (e.g., input data representing a grid) that is similar to data that is input into one or more machine learning models when used by a vehicle or machine.
[0031] 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)), autonomous vehicles, driven and undriven robots or robotic platforms, warehouse vehicles, off-road vehicles, vehicles coupled to one or more trailers, flying boats, boats, shuttles, emergency response vehicles, motorcycles, electric or motorized bicycles, airplanes, construction vehicles, underwater vehicles, drones, and / or other vehicle types. Further, the systems and methods described herein may be used for a variety of purposes, by way of example and not limitation, for machine control, machine motion, machine driving, synthetic data generation, model training, perception, augmented reality, virtual reality, mixed reality, robotics, security 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 for 3D assets, cloud computing, and / or any other suitable application.
[0032] The disclosed embodiments may be included in a variety of different systems, such as automotive systems (e.g., control systems for autonomous or semi-autonomous machines, perception systems for autonomous or semi-autonomous machines), systems implemented using robots, aerial systems, medical systems, boating systems, smart area monitoring systems, systems for performing deep learning operations, systems for performing simulation operations, systems for performing data fusion, 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 large language models (LLMs) that process text, audio, and / or sensor data, 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.
[0033] Figure 1An example data flow diagram of a process 100 for performing data fusion using one or more machine learning models 102 in accordance with some embodiments of the present disclosure is shown. It should be understood that this and other arrangements described herein are set forth by way of example only. Other arrangements and elements (e.g., machines, interfaces, functions, orders, 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. Further, many of the elements described herein are functional entities that may be implemented as discrete or distributed components or in combination with other components, and in any suitable combination and location. The various functions described herein as being performed by an entity may be performed by hardware, firmware, and / or software. For example, the various functions may be implemented by a processor executing instructions stored in a memory. In some embodiments, the systems, methods, and processes described herein may be implemented using 9A to 9D Example of autonomous vehicle 900, Fig.10 The example computing device 1000 and / or Fig.11 The example data center 1100 may be performed using similar components, features, and / or functions to those of the example data center 1100.
[0034] The process 100 may include receiving first sensor data 104 generated using one or more first sensors 106 and receiving second sensor data 108 generated using one or more second sensors 110. As described herein, the one or more first sensors 106 may include a first type of sensor, and the one or more second sensors 110 may include a different second type of sensor. For a first example, the first sensor 106 may include an image sensor and the second sensor 110 may include a RADAR sensor. For a second example, the first sensor 106 may again include an image sensor and the second sensor 110 may include a LiDAR sensor. While these are only a few examples of sensor types associated with the first sensor 106 and the second sensor 110, in other examples, the first sensor 106 and / or the second sensor 110 may include any type of sensor.
[0035] For example, Figure 2 An example of a vehicle 202 navigating around an environment 204 is shown in accordance with some embodiments of the present disclosure. Figure 2 In the example of , while navigating around environment 204, vehicle 202 may use sensors (e.g., one or more first sensors 106, one or more second sensors 110, etc.) to generate sensor data (e.g., sensor data 104, sensor data 108, etc.) representing multiple objects 206(1)-(5) (also referred to as "object 206" in the singular or "multiple objects 206" in the plural) around vehicle 202. Although Figure 2The example of 206(1)-(3) is shown as including other vehicles and objects 206(4)-(5) are shown as including pedestrians, but in other examples, objects 206 may include any other type of object (e.g., buildings, street signs, animals, etc.). In addition, although Figure 2 The example of 202 describes sensor data as representing five objects 206, but in other examples, vehicle 202 may generate sensor data representing any number of objects (e.g., one object, ten objects, fifty objects, one hundred objects, one thousand objects, etc.). Further, while described as detecting, tracking, or determining parameters of objects, objects will extend to static objects, dynamic objects, fluctuations / disturbances in a surface (such as a driving surface), cavities or holes in a surface, and / or features within the environment (road markings, etc.).
[0036] Return to reference Figure 1 In an example of a process 100, the process 100 may include a first processing pipeline 112 that processes the first sensor data 104 and outputs first object data 114 based on the processing. For example, the first processing pipeline 112 may include one or more models, one or more neural networks, one or more algorithms, and / or any other components that process the first sensor data 104 to determine first information associated with one or more first objects located within the environment, wherein the first object data 114 represents the first information. In some examples, the first information may include a first value of a first parameter associated with the first object. As described herein, the parameters associated with the object may include, but are not limited to, an x-coordinate position, a y-coordinate position, a z-coordinate position, a height, a width, a length, a speed in the x-direction, a speed in the y-direction, an orientation, a posture, a classification, and / or any other parameter.
[0037] The process 100 may include a second processing pipeline 116 that processes the second sensor data 108 and outputs second object data 118 based on the processing. For example, the second processing pipeline 116 may include one or more models, one or more neural networks, one or more algorithms, and / or any other components that process the second sensor data 108 to determine second information associated with one or more second objects located within the environment, wherein the second object data 118 represents the second information. In some examples, the second information may include a second value of a second parameter associated with the second object. In some examples, the first object associated with the first information may be the same as the second object associated with the second information. In some examples, one or more objects from the first object associated with the first information may be different from one or more objects from the second object associated with the second information.
[0038] In some examples, first object data 114 generated using first processing pipeline 112 and / or second output data 118 generated using second processing pipeline 116 may not represent one or more values of one or more of the parameters described herein based on the types of sensors 106 and 110 used to generate sensor data 104 and 108. For non-limiting example, the second information may not include one or more values of one or more heights associated with one or more second objects if the values of such parameters are difficult to determine based on the sensor modality that generated the data.
[0039] For example, FIG. 3A to FIG. 3B An example of an output associated with a processing pipeline according to some embodiments of the present disclosure is shown, wherein the output indicates information associated with an object. Figure 3A In an example of , first processing pipeline 112 may have processed first sensor data (e.g., first sensor data 104) and, based at least on the processing, output first data (e.g., first object data 114) representing first information 302 of one or more objects 206. In some examples, first processing pipeline 112 may be associated with processing image data generated using an image sensor. As such, and as shown, information 302 may include at least bounding shapes 304 (1)-(5) (also referred to as "bounding shapes 304" in the singular or "bounding shapes 304" in the plural) indicating the location of object 206. Information 302 may then be used to determine a first value for a first parameter associated with object 206.
[0040] For example, and for object 206, the center of bounding shape 304 can be used to determine a value for an x-coordinate position and a value for a y-coordinate position associated with object 206. In some examples, the x-coordinate position and the y-coordinate position are relative to a vehicle located at the center of the illustration of information 302 (e.g., in an ego-coordinate system centered about an origin located on the ego-vehicle, such as, but not limited to, at the center of the ego-vehicle's rear axle). The length and width of bounding shape 304 can also be used to determine a value for the width and a value for the length, respectively, associated with object 206. Additionally, while Figure 3A The example of FIG. 30 shows information 302 as including a two-dimensional (2D) bounding shape 304 associated with object 206, but in other examples, information 302 may include a three-dimensional (3D) bounding shape (e.g., a cuboid) associated with object 206. In such an example, the center of the 3D bounding shape may be used to determine a value for the z-coordinate position of the object, and the height of the 3D bounding shape may be used to determine a value for the height of the object.
[0041] Boundary shape 304 can also be used to determine the value of the orientation associated with the object. For example, an object 206 that is directly oriented in the negative x direction can be associated with an orientation of 0 degrees. Then, the angle associated with the orientation can increase in the counterclockwise direction. For example, an object 206 that is directly oriented in the negative y direction can be associated with an orientation of 90 degrees, an object 206 that is directly oriented in the positive x direction can be associated with an orientation of 180 degrees, and an object 206 that is oriented in the positive y direction can be associated with an orientation of 270 degrees (and / or -90 degrees). Although this is merely an example technique for determining a value (e.g., angle) associated with the orientation of object 206, in other examples, additional and / or alternative techniques can be used to determine the value (e.g., angle can increase in the clockwise direction, the positive x direction can be associated with an orientation of 0 degrees, etc.).
[0042] The first processing pipeline 112 may additionally output values of additional parameters associated with the object 206. For example, and for the object, the first processing pipeline 112 may output a value of velocity in the x-direction and a value of velocity in the y-direction (and / or a value of velocity in the z-direction). In such examples, the first processing pipeline 112 may use a single representation of the sensor data (e.g., a single frame), or, for example, to improve accuracy, the first processing pipeline 112 may use multiple representations of the sensor data (e.g., multiple frames generated over a period of time). The first processing pipeline 112 may also output a value associated with a classification of the object (which is described in more detail herein).
[0043] exist Figure 3B In the example of , the second processing pipeline 116 may have processed second sensor data (e.g., second sensor data 108) and based on the processing output second data (e.g., second object data 118) representing second information 306 of one or more objects 206. In some examples, the second processing pipeline 116 may be associated with processing RADAR data generated using a RADAR sensor and / or LiDAR data generated using a LiDAR sensor. Thus, and as shown, the information 306 may include at least points 308(1)-(4) associated with objects 206(1)-(4) (also referred to as "point 308" in the singular or "points 308" in the plural). However, the information 306 may not include points associated with object 206(5) (e.g., object 206(5) may have been blocked by object 206(4) from being sensed by the sensor). The information 302 may then be used to determine a second value of a second parameter associated with the object 206.
[0044] For example, the second processing pipeline 116 (and / or another component) may use the points 308 to determine a bounding shape 310 (1)-(4) (also referred to singularly as a "bounding shape 310" or plurally as "plural bounding shapes 310"). For example, and for the bounding shape 310, an x-coordinate value and a y-coordinate value of a centroid are calculated, and then the dimensions (e.g., length and width) of the bounding shape 310 are also calculated. In some examples, to perform these calculations, a minimum deviation and a maximum deviation of the points 308 associated with the object 206 are calculated. The minimum deviation and the maximum deviation give the left and right extremes of the bounding shape 308. The length and width are then calculated by subtracting the minimum deviation from the maximum deviation. Additionally, the offset to the tracking point provides the centroid of the bounding shape 310. While this is merely one example technique for determining the bounding shape 310 of the object 206, in other examples, additional and / or alternative techniques may be used to determine the bounding shape 310 of the object.
[0045] Boundary shape 310 may then be used to determine a second parameter of object 206 (1)-(4). For example, and for object 206, the center of boundary shape 310 may be used to determine a value for an x-coordinate position and a value for a y-coordinate position associated with object 206. In some examples, the x-coordinate position and the y-coordinate position are again relative to a vehicle located at the center of the illustration of information 306. The width and length of boundary shape 310 may be used to determine a value for the width and a value for the length, respectively, associated with object 206. Additionally, second processing pipeline 116 may output additional information representing a value for speed in the x-direction, a value for speed in the y-direction, a value for orientation, and / or a value for a classification associated with object 206. Additionally, a similar process may be used to determine a value for a second parameter associated with one or more (e.g., each) of other objects 206 (1)-(4).
[0046] As described herein, second processing pipeline 116 may not determine one or more values of one or more of the parameters determined by first processing pipeline 112 because the second sensor data is different from the first sensor data. For example, second processing pipeline 116 may not be able to determine at least a value of a z-coordinate location and a value of a height associated with objects 206(1)-(4). However, and as described in more detail herein, one or more machine learning models 102 may still be able to fuse the data even without this information.
[0047] Return to reference Figure 1In an example of, process 100 may include first input component 120 processing first object data 114 and outputting first input data 122 associated with one or more first sensors 106 based at least on the processing. Process 100 may further include second input component 124 processing second object data 118 and outputting second input data 122 associated with second sensor data 108 based at least on the processing. Figure 1 The example of FIG. 1 shows the first input component 120 as being separate from the second input component 124 , but in other examples, the first input component 120 and the second input component 124 may be combined into a single input component.
[0048] As described herein, in some examples, the input data 122 may represent one or more grids, such as one or more BEV grids, that represent values of parameters associated with an object. In some examples, the number of grids generated by the first input component 120 and / or the second input component 124 may correspond to the number of parameters associated with the one or more objects. For the first example, if the first object data 114 represents values of ten parameters associated with one or more first objects, the first input component 120 may generate ten grids, each of which is associated with a corresponding parameter. For the second example, if the second object data 118 represents values of eight parameters, but values of two additional parameters cannot be determined based on the second sensor data 108, the second input component 124 may again generate ten grids, eight of which are associated with eight determined parameters, and two of which are associated with two undetermined parameters (which are described in more detail herein).
[0049] For example, and using the ten parameter examples above, the first input component 120 may generate: a first grid representing one or more values of one or more x-coordinate positions of one or more first objects, a second grid representing values of y-coordinate positions of the first objects, a third grid representing values of z-coordinate positions of the first objects, a fourth grid representing values of the width of the first objects, a fifth grid representing values of the length of the first objects, a sixth grid representing values of the height of the first objects, a seventh grid representing values of the x-direction velocity of the first objects, an eighth grid representing values of the y-direction velocity of the first objects, a ninth grid representing values of the orientation of the first objects, and a tenth grid representing the values of the classification associated with the first objects. Additionally, the second input component 124 may perform a similar process to generate a grid representing similar parameters of a second object.
[0050] In some examples, the grid may be divided into sections (e.g., cells), each of which is associated with an area of the environment surrounding the vehicle. For a first example, if the grid represents a 400 meter by 400 meter segment of the environment, the grid may be divided into 640,000 sections, each of which is associated with a 0.5 meter by 0.5 meter area of the environment. In such an example, the input data 122 for the grid may include a size of 800x800. Additionally, if each sensor type is associated with ten parameters (e.g., ten channels), where there are two different sensor types, the input data 122 may include 800x800x20. For a second example, and again, if the grid represents a 400 meter by 400 meter segment of the environment, the grid may be decomposed into 160,000 sections, each of which is associated with a 1 meter by 1 meter area of the environment. In such an example, the input data 122 for the grid may include a size of 400x400. Additionally, if each sensor type is associated with ten parameters (eg, ten channels), where there are two different sensor types, the input data 122 may include 400x400x20.
[0051] Then, when generating input data 122, the first input component 120 and / or the second input component 124 can use the positions of these objects together with these parts. For example, and for a grid, the first input component 120 (and / or similarly, the second input component 124) can determine a portion of the grid associated with the position of the object. In some examples, the first input component 120 uses the value of the x-coordinate position and the value of the y-coordinate position associated with the object to determine the portion of the grid (this is described in more detail herein). The first input component 120 can then enter the value associated with the parameter into the portion of the grid. In addition, the first input component 120 may perform a similar process to enter one or more values associated with the parameters of one or more additional objects. In addition, the first input component 120 may perform a similar process to enter the values of the remaining parameters of the first object into other grids.
[0052] In some examples, the first input component 120 (and / or similarly, the second input component 124) can use different techniques to input the value of the parameter. For example, the first input component 120 can determine the distance between the center of a portion of the grid associated with the object and the value of the x-coordinate position of the object. The first input component 120 can then enter the value associated with the distance into the portion of the grid associated with the x-coordinate position parameter. The first input component 120 can also determine the distance between the center of a portion of the grid associated with the object and the value of the y-coordinate position of the object. The first input component 120 can then enter the value associated with the distance into the portion of the grid associated with the y-coordinate position parameter.
[0053] In addition, the first input component 120 can then enter the values of the z-coordinate position, height, width, length, speed in the x-direction, speed in the y-direction, and orientation into the corresponding portion of the grid associated with those parameters. In addition, the first input component 120 can determine the value associated with the classification of the object. In some examples, the value can be associated with a range, such as a range between 0 and 1, where different classifications are associated with different values. For example, a vehicle can be associated with a value of 0.1, a pedestrian can be associated with a value of 0.3, an animal can be associated with a value of 0.9, and so on. The first input component 120 can then enter the value into the corresponding portion of the grid associated with the classification parameter. Using a similar process, the first input component 120 can enter the values of the parameters associated with other first objects into the grid.
[0054] In some examples, more than one object (e.g., overlapping objects) may be associated with the same portion of these grids. For example, if the portion of the grid represents a 1 meter by 1 meter area of the environment, then more than one object may be located within the 1 meter by 1 meter area. In such an example, the first input component 120 (and / or similarly, the second input component 124) may use one or more techniques to represent two overlapping objects through the input data 122. For example, the first input component 120 may generate a first set of grids (e.g., ten grids) representing the values of the parameters of the first object in the overlapping objects and also generate a second set of grids (e.g., ten grids) representing the values of the parameters of the second object in the overlapping objects. Although this example describes generating two sets of grids for two overlapping objects, in some examples, the first input component 120 may generate any number of sets of grids for any number of overlapping objects.
[0055] For example, 4A to 4D An example visualization of generating inputs to one or more machine learning models 102 according to some embodiments of the present disclosure is shown. Figure 4AAs shown in the example of , the first input component 120 may generate a first set of inputs 402 (although only one is labeled for clarity) representing values of parameters associated with objects 206 ( 1 )-( 4 ), as determined by the first processing pipeline 112 . For example, the first input 402 may represent the value of the x-coordinate position of the object 206(1)-(4), the second input 402 may represent the value of the y-coordinate position of the object 206(1)-(4), the third input 402 may represent the value of the z-coordinate position of the object 206(1)-(4), the fourth input 402 may represent the value of the width of the object 206(1)-(4), the fifth input 402 may represent the value of the length of the object 206(1)-(4), the sixth input 402 may represent the value of the height of the object 206(1)-(4), the seventh input 402 may represent the value of the x-direction speed of the object 206(1)-(4), the eighth input 402 may represent the value of the y-direction speed of the object 206(1)-(4), the ninth input 402 may represent the value of the orientation of the object 206(1)-(4), and the tenth input 402 may represent the value of the classification associated with the object 206(1)-(4). Figure 4A Examples (and FIG. 4B to FIG. 4D ) represents the set of inputs 402 as including ten inputs 402 associated with ten different parameters, but in other examples, the set of inputs may include any other number of inputs associated with any number of parameters.
[0056] exist Figure 4A Examples (and FIG. 4B to FIG. 4D ), input 402 includes a grid (e.g., a BEV grid) representing an environment 204 that vehicle 202 is navigating. For example, the grid may represent a rectangular cross-section of environment 204 surrounding vehicle 202 (e.g., the vehicle may be located in the middle of the rectangular cross-section), where a first dimension of the rectangle represents a first distance within environment 204, and a second dimension of the rectangle represents a second distance within environment 204. However, in other examples, the grid may include any other shape (e.g., a circle, a triangle, a pentagon, a hexagon, etc.). Further, the grid may include a plurality of portions 404 (although only a few are labeled for clarity) (also referred to as "portion 404" in the singular or "portions 404" in the plural), where each portion 404 represents an area of environment 204. Although Figure 4A The example of illustrative embodiments shows the grid as including one hundred portions 404, but in other examples, the grid may include any other number of portions.
[0057] First input component 120 may then use the positions of objects 206(1)-(4) to input values associated with the parameters into input 402. For example, and with respect to object 206(1), first input component 120 may use the position of object 206(1) (e.g., the value of the x-coordinate position and the value of the y-coordinate position) to determine that object 206(1) is associated with portion 404(1) of the grid. In other words, portion 404(1) of the grid may represent an area of environment 204 where object 206(1) is located. As such, first input component 120 may input values of the parameters associated with object 206(1) into portion 404(1) of the grid. Additionally, first input component 120 may use a similar process to input additional values of additional parameters associated with object 206(1) into corresponding portions 404(1) of other grids.
[0058] If Figure 4A As further illustrated in the example of , first input component 120 may perform a similar process to determine that object 206(2) is associated with portion 404(2) of the grid. First input component 120 may then input the value of the parameter associated with object 206(2) into portion 402(2) of the grid. Additionally, first input component 120 may perform a similar process to determine that object 206(3) is associated with portion 404(3) of the grid. First input component 120 may then input the value of the parameter associated with object 206(3) into portion 404(3) of the grid. Additionally, first input component 120 may perform a similar process to determine that object 206(4) is associated with portion 404(4) of the grid. First input component 120 may then input the value of the parameter associated with object 206(4) into portion 404(4) of the grid.
[0059] like Figure 4B , first input component 120 may generate a second set of inputs 406 (although only one is labeled for clarity) representing values of parameters associated with object 206(5) determined by first processing pipeline 112. In some examples, based on objects 206(4)-(5) being associated with the same portion of the grid (e.g., objects 206(4)-(5) are overlapping), first input component 120 uses first set of inputs 402 for parameters associated with object 206(4) and uses second set of inputs 406 for parameters associated with object 206(5). For example, input 406 includes portion 408 that corresponds to portion 404 of input 402. Additionally, portion 408(1) of input 406 associated with object 206(5) corresponds to portion 404(4) of input 402 associated with object 206(4). As such, first input component 120 may input values of parameters of object 206(5) within portion 408(1) of the grid.
[0060] like Figure 4C As shown in the example of , the second input component 124 can generate a third set of inputs 410 (although only one is labeled for clarity) representing values of parameters associated with objects 206 ( 1 )-( 4 ) as determined by the second processing pipeline 116 . For example, the first input 410 may represent the value of the x-coordinate position of the object 206(1)-(4), the second input 410 may represent the value of the y-coordinate position of the object 206(1)-(4), the third input 410 may represent the value of the z-coordinate position of the object 206(1)-(4), the fourth input 410 may represent the value of the width of the object 206(1)-(4), the fifth input 410 may represent the value of the length of the object 206(1)-(4), the sixth input 410 may represent the value of the height of the object 206(1)-(4), the seventh input 410 may represent the value of the x-direction speed of the object 206(1)-(4), the eighth input 410 may represent the value of the y-direction speed of the object 206(1)-(4), the ninth input 410 may represent the value of the orientation of the object 206(1)-(4), and the tenth input 410 may represent the value of the classification associated with the object 206(1)-(4).
[0061] As such, second input component 124 may perform a similar process as first input component 120 to input values of parameters of objects 206(1)-(4) into input 410. For example, second input component 124 may perform the process described herein to determine that object 206(1) is associated with portion 412(1) of the grid. Second input component 124 may then input values of parameters associated with object 206(1) into portion 412(1) of the grid. Additionally, second input component 124 may perform the process described herein to determine that object 206(2) is associated with portion 412(2) of the grid. Second input component 124 may then input values of parameters associated with object 206(2) into portion 412(2) of the grid. Additionally, second input component 124 may perform the process described herein to determine that object 206(3) is associated with portion 412(3) of the grid. Second input component 124 may then input values of parameters associated with object 206(3) into portion 412(3) of the grid. Additionally, second input component 124 may perform the processes described herein to determine that object 206(4) is associated with portion 412(4) of the grid. Second input component 124 may then input the value of the parameter associated with object 206(4) into portion 412(4) of the grid.
[0062] like Figure 4DAs shown in the example of , similar to second set of inputs 406, second input component 124 may generate a fourth set of inputs 414 associated with overlapping objects (although only one is labeled for clarity). However, because second processing pipeline 116 has not determined information 306 associated with object 206(5), second input component 124 may not have any values for the parameters entered into inputs 414. Therefore, Figure 4D The example of 414 shows portion 416 of input 414 as not including any input values.
[0063] exist Figures 4A-4D In the example of FIG. 1 , many of the portions 404, 408, 412, and 416 of the inputs 402, 406, 410, and 414 do not include values associated with parameters of the object 206 (which are indicated by white squares). As such, in some examples, the first input component 120 and / or the second input component 124 may input set values into those portions 404, 408, 412, and 416. In such examples, the set values may include values that cannot be associated with any parameter, such as a negative value, or the set values may include any other value. Additionally, as described herein, in some examples, the second object data 118 may not include one or more values for one or more parameters, such as the z-coordinate position of the object 206 (1)-(4), the height of the object 206 (1)-(4), and / or the like. Again, in such examples, the second input component 124 may input set values into the portions 412 and 416 of those inputs 410 and 414 that are associated with those parameters. The set value may include a value that cannot be associated with any parameter, such as a negative value, or the set value may include any other value. In some examples, by using such a set value, one or more machine learning models 102 can learn to "ignore" those values when generating outputs.
[0064] Return to see Figure 1In an example of, the process 100 may include inputting the input data 122 into one or more machine learning models 102, wherein the one or more machine learning models 102 are trained to process the input data 122 so as to fuse the first object data 114 with the second object data 118. The one or more machine learning models 102 may include any type of machine learning model, such as a machine learning model using linear regression, logistic regression, decision tree, support vector machine (SVM), naive Bayes, k-nearest neighbor (Knn), K-means clustering, random forest, dimensionality reduction algorithm, gradient boosting algorithm, neural network (e.g., autoencoder, convolution, recurrent, perceptron, long / short term memory (LSTM), Hopfield, Boltzmann, deep belief, deconvolution, generative adversarial, liquid machine, etc.), large language model (LLM), and / or other types of machine learning models. For example, the one or more machine learning models 102 may use one or more neural networks to process the input data 122. The one or more neural networks may include, but are not limited to, one or more deep neural networks, one or more convolutional neural networks, one or more recurrent neural networks, one or more generative neural networks, one or more modular neural networks, and / or any other type of one or more neural networks.
[0065] For example, one or more machine learning models 102 may include any number and / or type of layers. For example, one or more machine learning models 102 may include one or more convolutional layers. In some examples, a convolutional layer may compute the output of neurons connected to local regions in an input layer, where each neuron computes a dot product between their weights and the small region they are connected to in the input volume. The result of the convolutional layer may be another volume with one of the dimensions based on the number of filters applied (e.g., width, height, and number of filters, such as 32x32x12 if 12 is the number of filters).
[0066] One or more of these layers may include an input layer. The input layer may hold values associated with the input data 122. One or more of the layers may include a rectified linear unit (ReLU) layer. The ReLU layer may apply an element-wise activation function, such as max(0, x), with a threshold of zero. The volume of the resulting ReLU layer may be the same as the volume of the input to the ReLU layer.
[0067] One or more of these layers may include a pooling layer. The pooling layer may perform a downsampling operation along the spatial dimensions (e.g., height and width), which may result in a smaller volume than the input to the pooling layer (e.g., 16x16x12 from a 32x32x12 input volume). In some examples, one or more machine learning models 102 may not include any pooling layers. In such examples, other types of convolutional layers may be used instead of pooling layers. In some examples, one or more feature extractor layers may include alternating convolutional layers and pooling layers.
[0068] One or more of these layers may include a fully connected layer. Each neuron in a fully connected layer may be connected to each neuron in a previous volume. The fully connected layer may calculate category scores, and the resulting volume may be 1x1xN (where N is the number of categories). In some examples, one or more feature extractor layers may include fully connected layers, while in other examples, the fully connected layers of one or more machine learning models 102 may be fully connected layers separated from one or more feature extractor layers. In some examples, one or more feature extractor layers and / or one or more machine learning models 102 as a whole may not use fully connected layers in an effort to increase processing time and reduce computing resource requirements. In such an example, without using fully connected layers, one or more machine learning models 102 may include a fully convolutional network.
[0069] In some examples, one or more of these layers may include one or more deconvolution layers. However, the use of the term deconvolution may be misleading and is not intended to be limiting. For example, a deconvolution layer may alternatively be referred to as a transposed convolution layer or a fractionally striped convolution layer. A deconvolution layer may be used to perform upsampling on the output of a previous layer.
[0070] Although input layers, convolutional layers, pooling layers, ReLU layers, deconvolutional layers, and fully connected layers are discussed herein with respect to the machine learning model 102, this is not intended to be limiting. For example, one or more machine learning models 102 may include additional or alternative layers, such as normalization layers, SoftMax layers, and / or other layer types.
[0071] In some examples, one or more of the layers may include parameters (e.g., weights and / or biases), while other layers may not include parameters, such as, for example, ReLU layers and pooling layers may not include parameters. In some examples, the parameters may be learned by the machine learning model 102 during training. Further, some layers may include additional hyperparameters (e.g., learning rate, stride, epoch, kernel size, number of filters, pooling type for pooling layers, etc.) - such as one or more convolutional layers, one or more deconvolutional layers, and one or more pooling layers - while other layers may not include additional hyperparameters, such as one or more ReLU layers. Various activation functions may be used, including but not limited to ReLU, leaky ReLU, sigmoid, hyperbolic tangent (tan h), exponential linear unit (ELU), etc. Parameters, hyperparameters and / or activation functions are not limited and may vary depending on the embodiment.
[0072] exist Figure 1 In an example of , based at least on processing input data 122, one or more machine learning models 102 may output fused object data 126 representing fused information associated with one or more objects. In some examples, the fused information may include fused values of parameters associated with the objects. As described herein, parameters associated with the fused object data 126 of the objects may include, but are not limited to, x-coordinate position, y-coordinate position, z-coordinate position, height, width, length, speed in the x-direction, speed in the y-direction, orientation, classification, and / or any other parameter.
[0073] In some examples, the fused object data 126 may represent one or more grids, such as similar to the one or more grids represented by the input data 122. For example, if the input data 122 represents a first number of grids associated with a plurality of parameters (e.g., ten parameters), the fused object data 126 may represent a second number of grids associated with the plurality of parameters. For the first example, if the input data 122 represents twenty grids associated with values of ten different parameters determined using two different types of sensors, the fused object data 126 may represent ten grids associated with the ten different parameters (e.g., each grid may be associated with one of the parameters). For the second example, if the input data 122 represents forty grids associated with values of ten different parameters determined using different types of sensors (e.g., each type of sensor is used to generate two different sets of grids), the fused object data 126 may represent twenty grids associated with the ten different parameters. In such examples, the first set of grids may be associated with the first object and the second set of grids may be associated with the second object, such as when there are overlapping objects.
[0074] For example, FIG. 5A to FIG. 5B 1 shows an example of output generated by one or more machine learning models 102 according to some embodiments of the present disclosure. Figure 5A As shown in the example of , one or more machine learning models 102 can generate a first set of outputs 502 (although only one is labeled for clarity) representing fused values of parameters associated with objects 206(1)-(4). For example, the first output 502 may represent fused values of the x-coordinate positions of objects 206(1)-(4), the second output 502 may represent fused values of the y-coordinate positions of objects 206(1)-(4), the third output 502 may represent fused values of the z-coordinate positions of objects 206(1)-(4), the fourth output 502 may represent fused values of the widths of objects 206(1)-(4), the fifth output 502 may represent fused values of the lengths of objects 206(1)-(4), and the The sixth output 502 may represent a fused value of the height of the object 206(1)-(4), the seventh output 502 may represent a fused value of the x-direction velocity of the object 206(1)-(4), the eighth output 502 may represent a fused value of the y-direction velocity of the object 206(1)-(4), the ninth output 502 may represent a fused value of the orientation of the object 206(1)-(4), and the tenth output 502 may represent a fused value of the classification associated with the object 206(1)-(4).
[0075] For example, and as shown, similar to inputs 402, 406, 410, and 412, output 502 may include a grid that is separated into a plurality of portions 504. As such, portion 504(1) of output 502 may indicate a fused value of a parameter associated with object 206(1), portion 504(2) of output 502 may indicate a fused value of a parameter associated with object 206(2), portion 504(3) of output 502 may indicate a fused value of a parameter associated with object 206(3), and portion 504(4) of output 502 may indicate a fused value of a parameter associated with object 206(4). For example, portion 504(1) of first output 502 may indicate a fused value of an x-coordinate position of object 206(1), wherein the fused value is determined based on at least a value from portion 404(1) of first input 402 and a value from portion 412(1) of first input 410. Additionally, portion 504(1) of second output 502 may indicate a fused value for the y-coordinate position of object 206(1), where the fused value is determined based on at least a value of portion 404(1) from second input 402 and a value of portion 412(1) from second input 410. This may be similar for one or more (e.g., each) of the other values of the parameter and / or for one or more (e.g., each) of the other objects 206(2)-(4).
[0076] like Figure 5B As shown in the example of , one or more machine learning models 102 can generate a second set of outputs 506 (although only one is labeled for clarity) that represent fused values of parameters associated with overlapping objects, such as object 206 (5). For example, the first output 506 may represent a fused value of the x-coordinate position of the object 206(5), the second output 506 may represent a fused value of the y-coordinate position of the object 206(5), the third output 506 may represent a fused value of the z-coordinate position of the object 206(5), the fourth output 506 may represent a fused value of the width of the object 206(5), the fifth output 506 may represent a fused value of the length of the object 206(5), the sixth output 506 may represent a fused value of the height of the object 206(5), the seventh output 506 may represent a fused value of the x-direction velocity of the object 206(5), the eighth output 506 may represent a fused value of the y-direction velocity of the object 206(5), the ninth output 506 may represent a fused value of the orientation of the object 206(5), and the tenth output 506 may represent a fused value of a classification associated with the object 206(5).
[0077] For example and as shown, similar to inputs 402, 406, 410, and 412, output 506 may include a grid that is divided into a plurality of portions 508. As such, portion 508(1) of output 506 may indicate fused values of parameters of object 206(5). For example, portion 508(1) of first output 502 may indicate fused values of the x-coordinate position of object 206(1), portion 508(1) of second output 502 may indicate fused values of the y-coordinate position of object 206(5), portion 508(1) of third output 502 may indicate fused values of the z-coordinate position of object 206(5), and so on.
[0078] Return to reference Figure 1In some examples, the process 100 may include one or more additional processing components 128 that process the fused object data 126. In some examples, the one or more processing components 128 may include another fusion component, such as a multi-sensor fusion module, which is also configured to fuse the first sensor data 104 from the one or more first sensors 106 with the second sensor data 108 from the one or more second sensors 110. In such examples, the other fusion components may use the fused object data 126 to supplement the fusion already performed by the other fusion components. In some examples, the one or more processing components 128 may include one or more systems of the vehicle, such as a tracking system configured to create, update and / or terminate tracks associated with objects around the vehicle. While these are just a few examples of what the additional processing components 128 may include, in other examples, the other processing components 128 may include any other type of component, system, algorithm, etc. that utilizes the fused object data 126 to perform operations, tasks, actions, etc.
[0079] For example, the fused output data 126 can be used by an autonomous or semi-autonomous driving software stack (which can be represented by one or more processing components 128) to perform one or more operations by a vehicle (and / or other machine type). For example, the driving stack may include a world model manager that can be used to generate, update and / or define a world model. The world model manager can use information generated by one or more perception components of the driving stack and received from one or more perception components of the driving stack. The perception components may include obstacle sensors, path sensors, waiting sensors, map sensors and / or other perception components. For example, the world model can be defined at least in part based on the available attributes of obstacles, paths and waiting conditions that can be perceived by obstacle sensors, path sensors, waiting sensors and / or map sensors in real time or near real time. The world model manager can continuously update the world model based on inputs (e.g., data) newly generated and / or received from obstacle sensors, path sensors, waiting sensors, map sensors and / or other components of the vehicle. For example, the world model manager and / or the perception component can use the fused output data 126 to perform one or more operations.
[0080] The world model can be used to help inform the planning component, control component, obstacle avoidance component, and / or actuation component of the driving stack. An obstacle sensor can perform obstacle perception, which can be based on where the vehicle is allowed to drive or can drive, and the speed at which the vehicle can drive without colliding with obstacles (e.g., objects such as structures, entities, vehicles, etc.) sensed by the vehicle (and represented, for example, in the fused output data 126).
[0081] The path perception may perform path perception, such as by sensing a nominal path available in a particular situation. In some examples, the path perception may further consider lane changes for path perception. A lane map may represent one or more paths available to a vehicle and may be as simple as a single path on a ramp on a highway. In some examples, a lane map may include a path to a desired lane and / or may indicate available changes along a highway (or other road type), or may include adjacent lanes, lane changes, forks, turns, butterfly interchanges, merges, and / or other information.
[0082] As a result of rules, conventions, and / or practical considerations, the waiting sensor may be responsible for determining constraints on the vehicle. For example, the rules, conventions, and / or practical considerations may be related to traffic lights, multi-way stops, yields, merges, toll booths, gates, police or other emergency personnel, road workers, stopped buses or other vehicles, one-way bridge arbitration, ferry entrances, etc. In some examples, the waiting sensor may be responsible for determining longitudinal constraints on the vehicle, which require the vehicle to wait or slow down until certain conditions are true. In some examples, the waiting condition is caused by a potential obstacle, such as a crossing road at an intersection, which may not be sensed by direct sensing of, for example, an obstacle sensor (e.g., by using sensor data from a sensor because the obstacle may be obscured and not in the sensor's field of view). Therefore, the waiting sensor can provide situational awareness by addressing the dangers of obstacles that are not always immediately perceptible through perceptible and / or learned rules and conventions. Therefore, the waiting sensor can be used to identify potential obstacles and implement one or more controls (e.g., slow down, stop, etc.) that may not rely solely on the obstacle sensor.
[0083] A map sensor may include mechanisms to discern behavior and, in some examples, determine what particular instance of a convention applies in a particular region.
[0084] The planning component may include a route planner, a lane planner, a behavior planner, and a behavior selector, among other components, features, and / or functionality. The route planner may use information from a map sensor, a map manager, and / or a positioning manager, among other information, to generate a planned path that may be comprised of GNSS waypoints (e.g., GPS waypoints). Waypoints may represent a specific distance into the future for the vehicle, such as a number of city blocks, kilometers / miles, meters / feet, etc., which may be used as a target for a lane planner.
[0085] The lane planner may use as input a lane map (e.g., a lane map from a path sensor, which may be generated at least in part using fused output data 126), object poses within the lane map (e.g., according to a localization manager), and / or a target point and direction to a certain distance in the future from a route planner. The target point and direction may be mapped to the best matching drivable point and direction in the lane map (e.g., based on GNSS and / or compass direction). A graph search algorithm may then be performed on the lane map from the current edge in the lane map to find the shortest path to the target point.
[0086] The behavior planner may determine the feasibility of basic behaviors of the vehicle, such as staying in the lane or changing lanes left or right, so that these feasible behaviors may be matched with the most desired behavior output from the lane planner. For example, if the desired behavior is determined to be unsafe and / or unavailable, a default behavior may be selected instead (e.g., when the desired behavior or changing lanes is unsafe, the default behavior may be staying in the lane).
[0087] One or more control components may follow a trajectory or path (lateral and longitudinal) that has been received from a behavior selector of one or more planning components as closely as possible within the capabilities of the vehicle.
[0088] One or more obstacle avoidance components can help the vehicle avoid collisions with objects (e.g., moving objects and stationary objects). In some examples, one or more obstacle avoidance components can be used independently of the components, features, and / or functionalities of the vehicle required to comply with traffic regulations and drive courteously. In such examples, one or more obstacle avoidance components can ignore traffic laws, road rules, and courteous driving norms to ensure that no collision occurs between the vehicle and any object. Therefore, the obstacle avoidance layer can be a layer separate from the rules of the road layer, and the obstacle avoidance layer can ensure that the vehicle performs safe actions only from the perspective of obstacle avoidance. On the other hand, the rules of the road layer can ensure that the vehicle complies with traffic laws and conventions and abides by legal and traditional road rights.
[0089] Although Figure 1The example of shows the use of first sensor data 104 generated using one or more first sensors 106 and second sensor data 108 generated using one or more second sensors 110 to perform data fusion, but in other examples, the process 100 may include performing data fusion using additional sensor data generated using one or more additional sensors. For example, one or more third sensors may be used to generate third sensor data, wherein the one or more third sensors include a third type of sensor that is different from the first type of sensor associated with the one or more first sensors 106 and the second type of sensor associated with the one or more second sensors 110. The third processing pipeline can then process the third sensor data to generate third object data representing a third value of a third parameter associated with one or more third objects. In addition, the third input component (and / or the first input component 120 and / or the second input component 124) can then process the third object data to generate third input data 122 for input into one or more machine learning models 102.
[0090] In addition, in some examples, process 100 may continue to repeat using new sensor data 104 and 108 generated using sensors 106 and 110 in order to continue to determine updated information associated with the environment around the vehicle. For example, if the sensor data 104 and 108 are associated with a frame rate, process 100 may repeat based on the frame rate. For the first example, if the frame rate associated with at least one of the first sensor data 104 or the second sensor data 108 is 30 frames per second, process 100 may be repeated 30 times per second (e.g., every frame). For the second example, if the frame rate associated with at least one of the first sensor data 104 or the second sensor data 108 is again 30 frames per second, process 100 may be repeated 15 times per second (e.g., every other frame).
[0091] See now Figure 6 , Figure 66 is a data flow diagram illustrating a process 600 for training one or more machine learning models 102 to fuse data according to some embodiments of the present invention. As shown, one or more machine learning models 102 may be trained using input data 602 (e.g., training input data). The input data 602 may be generated by a first input component 120 and a second input component 124, similar to the input data 122 (e.g., the input data 702 may represent a grid of values indicating parameters associated with one or more objects). For example, the first input component 120 may process first object data generated using the first processing pipeline 112 (and / or another processing pipeline) using the first type of sensor data (e.g., the first sensor data 104). Based at least on the processing, the first input component 120 may generate a first portion of the input data 602. Additionally, the second input component 124 may process second object data generated using the second processing pipeline 116 (and / or another processing pipeline) using the second type of sensor data (e.g., the second sensor data 108). Based at least on the processing, the second input component 124 may generate a second portion of the input data 602.
[0092] One or more machine learning models 102 may be trained using training input data 602 and corresponding ground truth data 604. Ground truth data 604 may include annotations, labels, masks, and / or the like. For example, in some embodiments, ground truth data 604 may indicate actual values of parameters 606 associated with objects within an environment. For example, and for an object, parameters 606 may include (but are not limited to) x-coordinate position, y-coordinate position, z-coordinate position, height, width, length, speed in the x-direction, speed in the y-direction, orientation, classification, and / or any other parameter. In some examples, ground truth data 604 may be generated within a drawing program (e.g., an annotation program), a computer-aided design (CAD) program, a labeling program, another type of program suitable for generating ground truth data 604, and / or may be drawn manually. In any examples, the truth data 604 can be synthetically generated (e.g., generated from a computer model or rendering), realistically generated (e.g., designed and generated from real-world data), machine automated (e.g., using feature analysis and learning to extract features from the data and then generate labels), human annotated (e.g., a tagger or annotation expert defines the location of the labels), and / or a combination thereof (e.g., a human identifies the vertices of a polyline and a machine generates the polygons using a polygon rasterizer).
[0093] The training engine 608 may use one or more loss functions that measure the loss (e.g., error) in the fused output data 610 (which may be similar to the fused output data 126) generated by the one or more machine learning models 102 compared to the true value data 604. Any type of loss function may be used, such as cross entropy loss, mean squared error, mean absolute error, mean bias error, and / or other loss function types. In some examples, different outputs may have different loss functions. For example, the x-coordinate position may include a first loss, the y-coordinate position may include a second loss, the z-coordinate position may include a third loss, and so on. In such an example, the loss functions may be combined to form a total loss, and the total loss may be used to train one or more machine learning models 102 (e.g., update the parameters of one or more machine learning models 102). In any example, a backward pass calculation may be performed to recursively calculate the gradients of one or more loss functions relative to the training parameters. In some examples, the weights and biases of one or more machine learning models 102 may be used to calculate these gradients.
[0094] See now Figure 7 and Figure 8 , each block of methods 700 and 800 described herein includes a computing process that can be performed using any combination of hardware, firmware, and / or software. For example, different functions can be implemented by a processor executing instructions stored in a memory. Methods 700 and 800 can also be embodied as computer-usable instructions stored on a computer storage medium. Methods 700 and 800 can be provided to another product by a stand-alone application, a service, or a hosted service (standalone or in combination with other hosted services), or a plug-in, to name a few. In addition, by way of example, reference is made to Figure 1 Methods 700 and 800 are described. However, these methods 700 and 800 may additionally or alternatively be performed by any one system or any combination of systems, including but not limited to those described herein.
[0095] Figure 7A flow chart of a method 700 for performing data fusion using a machine learning model according to some embodiments of the present disclosure is shown. At block B702, the method 700 may include determining one or more first values of one or more parameters associated with one or more objects based at least on first sensor data generated using a first type of sensor. For example, the first processing pipeline 112 may process the first sensor data 104 generated using the first sensor 106. Based at least on the processing, the first processing pipeline 112 may generate first object data 114 representing one or more first values of one or more parameters associated with one or more objects. For example, and for an object, the parameters may include (but are not limited to) x-coordinate position, y-coordinate position, z-coordinate position, height, width, length, speed in the x-direction, speed in the y-direction, orientation, classification, and / or any other parameter.
[0096] At block B704, method 700 may include determining one or more second values of the one or more parameters associated with the one or more objects based at least on second sensor data generated using a second type of sensor. For example, second processing pipeline 116 may process second sensor data 108 generated using second sensor 110. Based at least on the processing, second processing pipeline 116 may generate second object data 118 representing one or more second values of one or more parameters associated with one or more objects. In some examples, one or more of the one or more second values are the same as one or more of the one or more first values. In some examples, one or more of the one or more second values are different from one or more of the one or more first values.
[0097] At block B706, the method 700 may include generating input data representing the one or more first values and the one or more second values. For example, the first input component 120 may process the first object data 114 and, based at least on the processing, generate a first portion of the input data 122. Additionally, the second input component 124 may process the second object data 118 and, based at least on the processing, generate a second portion of the input data 122. As described herein, the input data 122 may represent one or more grids indicating one or more first values and one or more second values of one or more parameters.
[0098] At block B708, method 700 may include determining, using one or more machine learning models and based at least on the input data, one or more third values of one or more parameters associated with one or more objects. For example, one or more machine learning models 102 may process input data 122 and, based at least on the processing, generate fused object data 126 representing one or more third values of one or more parameters. In some examples, fused object data 126 may represent one or more grids indicating one or more third values of one or more parameters.
[0099] Figure 8 800 for generating input data for one or more machine learning models according to some embodiments of the present disclosure. At block B802, the method 800 may include: receiving data representing a value of a parameter associated with an object. For example, the first input component 120 may receive first object data 114 representing a value of a parameter associated with an object. As described herein, for an object, the parameter may include (but is not limited to) an x-coordinate position, a y-coordinate position, a z-coordinate position, a height, a width, a length, a speed in the x-direction, a speed in the y-direction, an orientation, a classification, and / or any other parameter.
[0100] At block B804, method 800 may include determining a portion of the input associated with the object based at least on the location of the object. For example, the first input component 120 may use the location of the object to determine the portion of the input associated with the object. For example, the input may include a grid representing a fragment of an environment. The grid may then be divided into different portions, where respective portions of the grid represent regions of the environment. As such, the first input component 120 may use the location of the object to determine the portion of the grid associated with the region of the environment where the object is located.
[0101] At box B806, method 800 may include: inputting values into various parts of the input. For example, the first input component 120 may input values into various parts of the input. For example, the first input component 120 may input a value associated with an x-coordinate position into the portion of the first input, a value associated with a y-coordinate position into the portion of the second input, a value associated with a z-coordinate position into the portion of the third input, and so on. In some examples, the first input component 120 may perform a similar process to input additional values of parameters associated with one or more additional objects. In some examples, the second input component 124 may also perform a similar process using the second object data 118 to generate an input representing additional values of parameters associated with an object.
[0102] Example Autonomous Vehicle
[0103] Fig.9A is an illustration of an example autonomous vehicle 900 according to some embodiments of the present disclosure. Autonomous vehicle 900 (also referred to herein as “vehicle 900”) may include, but is not limited to, a passenger vehicle, such as a car, a truck, a bus, an emergency vehicle, a shuttle, an electric or motorized bicycle, a motorcycle, a fire truck, a police car, an ambulance, a boat, an engineering vehicle, an underwater vehicle, a robotic vehicle, a drone, an airplane, a vehicle connected 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 passengers). Autonomous vehicles are generally described according to the automation levels defined by the National Highway Traffic Safety Administration (NHTSA) of the U.S. Department of Transportation and the Society of Automotive Engineers (SAE) in the “Classification and Definition of Terms Related to Driving Automation Systems for Road Motor Vehicles” (Standard No. J3016-201806 issued on June 15, 2018, Standard No. J3016-201609 issued on September 30, 2016, and previous and future versions of the standard). The vehicle 900 may be capable of having one or more functions according to levels 3-5 of autonomous driving levels. The vehicle 900 may be capable of functioning according to one or more of levels 1-5 of autonomous driving levels. For example, the vehicle 900 may be capable of providing driver assistance (level 1), partial automation (level 2), conditional automation (level 3), high automation (level 4) and / or full automation (level 5), depending on the embodiment. As used herein, the term "autonomy" may include any and / or all types of autonomy of the vehicle 900 or other machines, such as full autonomy, high autonomy, conditional autonomy, partial autonomy, providing assisted autonomy, semi-autonomy, main autonomy or other names.
[0104] The vehicle 900 may include components such as a chassis, a body, wheels (e.g., 2, 4, 6, 8, 18, etc.), tires, axles, and other components of the vehicle. The vehicle 900 may include a propulsion system 950, such as an internal combustion engine, a hybrid power plant, an all-electric engine, and / or another type of propulsion system. The propulsion system 950 may be connected to a drive train of the vehicle 900, which may include a transmission, to achieve propulsion of the vehicle 900. The propulsion system 950 may be controlled in response to receiving a signal from a throttle / accelerator 952.
[0105] A steering system 954, which may include a steering wheel, may be used to steer the vehicle 900 (e.g., along a desired path or route) when the propulsion system 950 is operating (e.g., when the vehicle is in motion). The steering system 954 may receive signals from a steering actuator 956. For fully automated (Level 5) functionality, a steering wheel may be optional.
[0106] Brake sensor system 946 may be used to operate vehicle brakes in response to receiving signals from brake actuator 948 and / or brake sensors.
[0107] May include one or more system on chip (SoC) 904 ( Fig. 9C ) and / or one or more controllers 936 of one or more GPUs may provide signals (e.g., representing commands) to one or more components and / or systems of the vehicle 900. For example, the one or more controllers may send signals to operate vehicle brakes via one or more brake actuators 948, to operate a steering system 954 via one or more steering actuators 956, and to operate a propulsion system 950 via one or more throttles / accelerators 952. The one or more controllers 936 may include one or more onboard (e.g., integrated) computing devices (e.g., supercomputers) that process sensor signals and output operating commands (e.g., signals representing commands) to enable autonomous driving and / or assist a human driver in driving the vehicle 900. The one or more controllers 936 may include a first controller 936 for autonomous driving functions, a second controller 936 for functional safety functions, a third controller 936 for artificial intelligence functions (e.g., computer vision), a fourth controller 936 for infotainment functions, a fifth controller 936 for redundancy in emergency situations, and / or other controllers. In some examples, a single controller 936 may handle two or more of the above functions, two or more controllers 936 may handle a single function, and / or any combination thereof.
[0108] The one or more controllers 936 may provide signals for controlling one or more components and / or systems of the vehicle 900 in response to sensor data (e.g., sensor input) received from one or more sensors. The sensor data may be received from, for example and without limitation, a global navigation satellite system ("GNSS") sensor 958 (e.g., a global positioning system sensor), a RADAR sensor 960, an ultrasonic sensor 962, a LiDAR sensor 964, an inertial measurement unit (IMU) sensor 966 (e.g., an accelerometer, a gyroscope, a magnetic compass, a magnetometer, etc.), a microphone 996, a stereo camera 968, a wide angle camera 970 (e.g., a fisheye camera), an infrared camera 972, a surround camera 974 (e.g., a 360 degree camera), a long-range and / or mid-range camera 998, a speed sensor 944 (e.g., for measuring the velocity of the vehicle 900), a vibration sensor 942, a steering sensor 940, a brake sensor (e.g., as part of a brake sensor system 946), and / or other sensor types.
[0109] One or more of the controllers 936 may receive inputs (e.g., represented by input data) from the instrument cluster 932 of the vehicle 900 and provide outputs (e.g., represented by output data, display data, etc.) via a human machine interface (HMI) display 934, an audible annunciator, a speaker, and / or via other components of the vehicle 900. These outputs may include information such as vehicle speed, velocity, time, map data (e.g., Fig. 9C The HMI display 934 may include information such as a high-definition (“HD”) map 922 of the vehicle 900 , location data (e.g., the location of the vehicle 900 on the map), directions, locations of other vehicles (e.g., an occupancy grid), information about objects and states of objects as sensed by the controller 936 , and the like. For example, the HMI display 934 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.).
[0110] The vehicle 900 further includes a network interface 924 that can communicate via one or more networks using one or more wireless antennas 926 and / or a modem. For example, the network interface 924 may be capable of communicating via 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 926 may also enable communication between objects in the environment (e.g., vehicles, mobile devices, etc.) using one or more local area networks such as Bluetooth, Bluetooth Low Energy (“LE”), Z-wave, ZigBee, etc. and / or one or more low power wide area networks (LPWANs) such as LoRaWAN, SigFox, etc.
[0111] Fig. 9B For use according to some embodiments of the present disclosure Fig.9A An example of camera positions and fields of view for an example autonomous vehicle 900 is shown. The cameras and respective fields of view are an example embodiment and are not intended to be limiting. For example, additional and / or replaceable cameras may be included, and / or the cameras may be located at different locations on the vehicle 900.
[0112] The camera type for the camera may include, but is not limited to, a digital camera that may be suitable for use with components and / or systems of the vehicle 900. The camera may operate at an automotive safety integrity level (ASIL) B and / or at another ASIL. The camera type may have any image capture rate, such as 60 frames per second (fps), 120fps, 240fps, etc., depending on the embodiment. The camera may be capable of using a rolling shutter, a global shutter, another type of shutter, or a combination thereof. In some examples, the color filter array may include a red transparent (RCCC) color filter array, a red transparent blue (RCCB) color filter array, a red blue green transparent (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.
[0113] 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).
[0114] One or more of the cameras may 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 car (e.g., reflections from the dashboard reflected in the windshield mirror) that may interfere with the camera's image data capture capabilities. With respect to the wing mirror mounting assembly, the wing mirror assembly may be custom 3D printed so that the camera mounting plate matches the shape of the wing mirror. In some examples, one or more cameras may be integrated into the wing mirror. For the side view cameras, one or more cameras may also be integrated into the four pillars at each corner of the cab.
[0115] A camera (e.g., a front-facing camera) having a field of view that includes a portion of the environment in front of the vehicle 900 can be used for surround view to help identify the forward path and obstacles, as well as assist in providing information critical to generating an occupancy grid and / or determining a preferred vehicle path with the help of one or more controllers 936 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.
[0116] 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 970, which may be used to sense objects (e.g., pedestrians, intersection traffic, or bicycles) entering the field of view from the periphery. Fig. 9B Only one wide-angle camera is shown in the figure, but there may be any number (including zero) of wide-angle cameras 970 on the vehicle 900. In addition, any number of long-range cameras 998 (e.g., a long-view stereo camera pair) may be used for depth-based object detection, especially for objects for which a neural network has not been trained. Long-range cameras 998 may also be used for object detection and classification and basic object tracking.
[0117] Any number of stereo cameras 968 may also be included in the front configuration. In at least one embodiment, one or more stereo cameras 968 may include an integrated control unit including an expandable processing unit that may provide a multi-core microprocessor and programmable logic ("FPGA") with an integrated 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 968 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 may measure the distance from the vehicle to the target object and use the generated information (e.g., metadata) to activate autonomous emergency braking and lane departure warning functions. Other types of stereo cameras 968 may be used in addition to or alternatively to those described herein.
[0118] A camera having a field of view of a portion of the environment including the sides of the vehicle 900 (e.g., a side view camera) can be used for surround viewing, providing information used to create and update the occupancy grid and generate side impact collision warnings. Fig. 9B The four surround cameras 974 shown in FIG. 9 can be placed on the vehicle 900. The surround cameras 974 can include a wide-angle camera 970, a fisheye camera, a 360-degree camera, and / or the like. For example, the four fisheye cameras can be placed in front, behind, and on the sides of the vehicle. In an alternative arrangement, the vehicle can use three surround cameras 974 (e.g., left, right, and rear), and can utilize one or more other cameras (e.g., a forward-facing camera) as a fourth surround camera.
[0119] A camera having a field of view that includes a portion of the environment behind the vehicle 900 (e.g., a rear view camera) can be used to assist with parking, surround view, rear collision warning, and creating and updating an occupancy grid. A variety of cameras can be used, including but not limited to cameras that are also suitable as front cameras as described herein (e.g., long-range and / or mid-range cameras 998, stereo cameras 968, infrared cameras 972, etc.).
[0120] Fig. 9C For use according to some embodiments of the present disclosure Fig.9A Block diagram of an example system architecture for an example autonomous vehicle 900. It should be understood that this arrangement and other arrangements described herein are set forth merely as examples. Other arrangements and elements (e.g., machines, interfaces, functions, sequences, functional groupings, etc.) may be used in addition to or in place of those shown, and some elements may be omitted entirely. Further, many of the elements described herein are functional entities that may be implemented as discrete or distributed components or in combination with other components, and in any appropriate combination and location. The various functions described herein as being performed by an entity may be implemented by hardware, firmware, and / or software. For example, the various functions may be implemented by a processor executing instructions stored in a memory.
[0121] Fig. 9C Each of the components, features, and systems of the vehicle 900 is illustrated as being connected via a bus 902. The bus 902 may include a controller area network (CAN) data interface (alternatively, referred to herein as a "CAN bus"). The CAN may be a network inside the vehicle 900 that assists in controlling various features and functions of the vehicle 900, such as the 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 speed per minute (RPM), button position, and / or other vehicle status indicators. The CAN bus may be ASIL B compatible.
[0122] Although bus 902 is described as a CAN bus here, this is not intended to be limiting. For example, in addition to or alternatively to the CAN bus, FlexRay and / or Ethernet can be used. In addition, although bus 902 is represented by a single line, this is not intended to be limiting. For example, there can be any number of buses 902, which may include one or more CAN buses, one or more FlexRay buses, one or more Ethernet buses, and / or one or more other types of buses using different protocols. In some examples, two or more buses 902 can be used to perform different functions, and / or can be used for redundancy. For example, a first bus 902 can be used for a collision avoidance function, and a second bus 902 can be used for drive control. In any example, each bus 902 can communicate with any component of the vehicle 900, and two or more buses 902 can communicate with the same component. In some examples, each SoC 904, each controller 936, and / or each computer in the vehicle can have access to the same input data (e.g., input from a sensor of the vehicle 900), and can be connected to a common bus such as a CAN bus.
[0123] The vehicle 900 may include one or more controllers 936, such as those described herein. Fig.9A Those controllers described. Controller 936 can be used for a variety of functions. Controller 936 can be coupled to any other different components and systems of vehicle 900, and can be used for control of vehicle 900, artificial intelligence of vehicle 900, infotainment for vehicle 900, and / or the like.
[0124] The vehicle 900 may include one or more system on chip (SoC) 904. The SoC 904 may include a CPU 906, a GPU 908, a processor 910, a cache 912, an accelerator 914, a data store 916, and / or other components and features not shown. The SoC 904 may be used to control the vehicle 900 in a variety of platforms and systems. For example, one or more SoCs 904 may be combined with an HD map 922 in a system (e.g., a system of the vehicle 900), and the HD map may be downloaded from one or more servers (e.g., a server) via a network interface 924. Fig.9D one or more servers 978) to obtain map refreshes and / or updates.
[0125] The CPU 906 may include a CPU cluster or CPU complex (alternatively, referred to herein as a "CCPLEX"). The CPU 906 may include multiple cores and / or L2 caches. For example, in some embodiments, the CPU 906 may include eight cores in a coherent multiprocessor configuration. In some embodiments, the CPU 906 may include four dual-core clusters, each of which has a dedicated L2 cache (e.g., a 2MB L2 cache). The CPU 906 (e.g., CCPLEX) may be configured to support simultaneous cluster operations, such that any combination of CPU 906 clusters can be active at any given time.
[0126] The CPU 906 may implement power management capabilities including one or more of the following features: each hardware block may be automatically clock gated when idle to save dynamic power; each core clock may be gated when the core is not actively executing instructions due to the execution of WFI / WFE instructions; each core may be independently power gated; each core cluster may be independently clock gated when all cores are clock gated or power gated; and / or each core cluster may be independently power gated when all cores are power gated. The CPU 906 may further implement an enhanced algorithm for managing power states, in which allowed power states and expected wake-up times are specified, and hardware / microcode determines the optimal power state to enter for the core, cluster, and CCPLEX. The processing core may support a simplified power state entry sequence in software, with the work being offloaded to the microcode.
[0127] The GPU 908 may include an integrated GPU (alternatively, referred to herein as an "iGPU"). The GPU 908 may be programmable and efficient for parallel workloads. In some examples, the GPU 908 may use an enhanced tensor instruction set. The GPU 908 may include one or more streaming microprocessors, each of which may include an L1 cache (e.g., an L1 cache with at least 96KB storage capacity), and two or more of these streaming microprocessors may share an L2 cache (e.g., an L2 cache with 512KB storage capacity). In some embodiments, the GPU 908 may include at least eight streaming microprocessors. The GPU 908 may use a computing application programming interface (API). In addition, the GPU 908 may use one or more parallel computing platforms and / or programming models (e.g., NVIDIA's CUDA).
[0128] In the case of automotive and embedded use, GPU 908 can be power optimized to achieve optimal performance. For example, GPU 908 can be manufactured on fin field effect transistors (FinFETs). However, this is not intended to be limiting, and GPU 908 can be manufactured using other semiconductor manufacturing processes. Each streaming microprocessor can merge several mixed precision processing cores divided into multiple blocks. For example and without limitation, 64 PF32 cores and 32 PF64 cores can be divided into four processing blocks. In such an example, each processing block can be allocated 16 FP32 cores, 8 FP64 cores, 16 INT32 cores, two mixed precision NVIDIA tensor cores for deep learning matrix arithmetic, L0 instruction cache, thread beam (warp) scheduler, dispatch unit and / or 64KB register file. In addition, the streaming microprocessor can include independent parallel integer and floating point data paths to provide efficient execution of workloads using a mix of calculations and addressing calculations. The streaming microprocessor may include independent thread scheduling capabilities to allow for finer-grained synchronization and collaboration between parallel threads. A streaming microprocessor may include a combined L1 data cache and shared memory unit to increase performance while simplifying programming.
[0129] GPU 908 may 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), may be used in addition to or in lieu of HBM memory.
[0130] The GPU 908 may include unified memory technology that includes access counters to allow memory pages to be more accurately migrated to the processor that accesses them most frequently, thereby improving the efficiency of memory ranges shared between processors. In some examples, address translation service (ATS) support may be used to allow the GPU 908 to directly access the CPU 906 page table. In such an example, when the GPU 908 memory management unit (MMU) experiences a miss, an address translation request may be transmitted to the CPU 906. In response, the CPU 906 may look up the virtual-physical mapping for the address in its page table and transmit the translation back to the GPU 908. In this way, unified memory technology may allow a single unified virtual address space to be used for memory of both the CPU 906 and the GPU 908, thereby simplifying GPU 908 programming and porting applications to the GPU 908.
[0131] In addition, GPU 908 may include access counters that can track how often GPU 908 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.
[0132] SoC 904 may include any number of caches 912, including those described herein. For example, cache 912 may include an L3 cache available to both CPU 906 and GPU 908 (e.g., connected to both CPU 906 and GPU 908). Cache 912 may include a write-back cache that may track the state of a line, for example, by using a cache coherence protocol (e.g., MEI, MESI, MSI, etc.). Depending on the embodiment, the L3 cache may include 4MB or more, although smaller cache sizes may also be used.
[0133] The SoC 904 may include an arithmetic logic unit (ALU) that may be used to perform processing for any of a variety of tasks or operations with respect to the vehicle 900 - such as processing a DNN. In addition, the SoC 904 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 104 may include one or more FPUs integrated as execution units within the CPU 906 and / or GPU 908.
[0134] SoC 904 may include one or more accelerators 914 (e.g., hardware accelerators, software accelerators, or a combination thereof). For example, SoC 904 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 908 and offload some tasks of GPU 908 (e.g., free up more cycles of GPU 908 for performing other tasks). As an example, accelerator 914 may be used for targeted workloads (e.g., perception, convolutional neural networks (CNNs), etc.) that are sufficiently stable to be easily controlled for acceleration. When used herein, the term "CNN" may include all types of CNNs, including region-based or regional convolutional neural networks (RCNNs) and fast RCNNs (e.g., for object detection).
[0135] Accelerator 914 (e.g., hardware acceleration cluster) may include a deep learning accelerator (DLA). DLA may include one or more tensor processing units (TPUs) that may be configured to provide an additional 10 trillion operations per second for deep learning applications and reasoning. TPU may be an accelerator configured to perform image processing functions (e.g., for CNN, RCNN, etc.) and optimized for performing image processing functions. DLA may be further optimized for a specific set of neural network types and floating point operations and reasoning. The design of DLA may provide higher performance per millimeter than a general-purpose GPU, and far exceeds the performance of a CPU. TPU may perform several functions, including a single instance convolution function, support, for example, INT8, INT16, and FP16 data types for both features and weights, and post-processor functions.
[0136] The DLA can quickly and efficiently execute neural networks, particularly CNNs, on processed or unprocessed data for any of a wide variety of functions, such as, but not limited to: a CNN for object recognition and detection using data from a camera sensor; a CNN for distance estimation using data from a camera sensor; a CNN for emergency vehicle detection and identification and detection using data from a microphone; a CNN for facial recognition and vehicle owner identification using data from a camera sensor; and / or a CNN for safety and / or security-related events.
[0137] The DLA can perform any function of the GPU 908, and by using an inference accelerator, for example, the designer can target any function to either the DLA or the GPU 908. For example, the designer can focus the processing of CNNs and floating point operations on the DLA, and leave other functions to the GPU 908 and / or other accelerators 914.
[0138] The accelerator 914 (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.
[0139] 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 number of memories. Depending on the embodiment, the RISC core can use any of a number of protocols. In some examples, the RISC core can execute a real-time operating system (RTOS). The RISC core can be implemented using one or more integrated circuit devices, application specific integrated circuits (ASICs), and / or storage devices. For example, the RISC core can include an instruction cache and / or a tightly coupled RAM.
[0140] The DMA may enable components of the PVA to access system memory independently of the CPU 906. The DMA may support any number of features used to provide optimizations for the PVA, including, but not limited to, support for multi-dimensional addressing and / or circular addressing. In some examples, the DMA may support addressing in up to six or more dimensions, which may include block width, block height, block depth, horizontal block stepping, vertical block stepping, and / or depth stepping.
[0141] The vector processor can be a programmable processor that can be designed to efficiently and flexibly perform programming for computer vision algorithms and provide signal processing capabilities. In some examples, the PVA can include a PVA core and two vector processing subsystem partitions. The PVA core can include a processor subsystem, one or more DMA engines (e.g., two DMA engines), and / or other peripherals. The vector processing subsystem can operate as the main processing engine of the PVA, and can include a vector processing unit (VPU), an instruction cache, and / or a vector memory (e.g., VMEM). The VPU core can include a digital signal processor, such as, for example, a single instruction multiple data (SIMD), a very long instruction word (VLIW) digital signal processor. The combination of SIMD and VLIW can enhance throughput and rate.
[0142] Each of the vector processors may include an instruction cache and may be coupled to a dedicated memory. As a result, in some examples, each of the vector processors may be configured to execute independently of the other vector processors. In other examples, the vector processors included in a particular PVA may be configured to employ data parallelism. For example, in some embodiments, multiple vector processors included in a single PVA may execute the same computer vision algorithm, but on different regions of an image. In other examples, the vector processors included in a particular PVA may execute different computer vision algorithms simultaneously on the same image, or even execute different algorithms on sequence images or portions of an image. Among other things, any number of PVAs may be included in a hardware acceleration cluster, and any number of vector processors may be included in each of these PVAs. In addition, the PVA may include additional error correction code (ECC) memory to enhance overall system security.
[0143] The accelerator 914 (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 914. In some examples, the on-chip memory may include at least 4MB of SRAM consisting of, for example and not limited to, eight field-configurable memory blocks that can be accessed by both the PVA and the DLA. Each pair of memory blocks may include an advanced peripheral bus (APB) interface, a configuration circuit system, a controller, and a multiplexer. Any type of memory may be used. The PVA and DLA may access the memory via a backbone that provides high-speed memory access to the PVA and DLA. The backbone may include an on-chip computer vision network that interconnects the PVA and DLA to the memory (e.g., using APB).
[0144] The on-chip computer vision network may include an interface that determines that both the PVA and DLA provide ready and valid signals before transmitting any control signals / addresses / data. Such an interface may provide separate phases and separate channels for transmitting control signals / addresses / data, as well as burst communication for continuous data transmission. This type of interface may comply with ISO 26262 or IEC 61508 standards, but other standards and protocols may also be used.
[0145] In some examples, SoC 904 may include a real-time ray tracing hardware accelerator such as described in U.S. Patent Application No. 16 / 101,232 filed on August 10, 2018. The real-time ray tracing hardware accelerator may be used to quickly and efficiently determine the position and range of objects (e.g., within a world model) to generate real-time visualization simulations for RADAR signal interpretation, for sound propagation synthesis and / or analysis, for SONAR system simulations, for general wave propagation simulations, for comparison with LiDAR data for the purpose of positioning and / or other functions, and / or for other purposes. In some embodiments, one or more tree traversal units (TTUs) may be used to perform one or more ray tracing related operations.
[0146] The accelerator 914 (e.g., a hardware accelerator cluster) has a wide range of uses in autonomous driving. The PVA can be a programmable visual accelerator that can be used in key processing stages in ADAS and autonomous vehicles. The capabilities of the PVA are a good match for algorithmic domains that require predictable processing, low power, and low latency. In other words, the PVA performs well on semi-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.
[0147] For example, according to one embodiment of the technology, PVA is used to perform computer stereo vision. In some examples, algorithms based on semi-global matching can be used, but this is not intended to be limiting. Many applications for level 3-5 autonomous driving require instant motion estimation / stereo matching (e.g., structure from motion, pedestrian recognition, lane detection, etc.). PVA can perform computer stereo vision functions on input from two monocular cameras.
[0148] In some examples, PVA can be used to perform dense optical flow. According to the process raw RADAR data (e.g., using 4D fast Fourier transform) to provide processed RADAR. In other examples, PVA is used for time-of-flight depth processing, which is for example by processing raw time-of-flight data to provide processed time-of-flight data.
[0149] DLA can be used to run any type of network to enhance control and driving safety, including, for example, a neural network that outputs a confidence measure for each object detection. Such a confidence value can be interpreted as a probability, or as providing a relative "weight" of each detection compared to other detections. This confidence value enables the system to make further decisions about which detections should be considered true positive detections rather than false positive detections. For example, the system can set a threshold for confidence and only consider detections that exceed the threshold as true positive detections. In an automatic emergency braking (AEB) system, a false positive detection will cause the vehicle to automatically perform emergency braking, which is obviously undesirable. Therefore, only the most confident detection should be considered a trigger for AEB. DLA can run a neural network for regressing confidence values. The neural network can take at least some subset of parameters as its input, such as bounding box dimensions, ground plane estimates obtained (e.g., from another subsystem), inertial measurement unit (IMU) sensor 966 output related to vehicle 900 orientation and distance, 3D position estimates of objects obtained from neural networks and / or other sensors (e.g., LiDAR sensor 964 or RADAR sensor 960), etc.
[0150] SoC 904 may include one or more data stores 916 (e.g., memory). Data store 916 may be on-chip memory of SoC 904 that may store neural networks to be executed on the GPU and / or DLA. In some examples, data store 916 may be large enough to store multiple instances of the neural network for redundancy and safety. Data store 912 may include an L2 or L3 cache 912. References to data store 916 may include references to memory associated with a PVA, DLA, and / or other accelerator 914 as described herein.
[0151] SoC 904 may include one or more processors 910 (e.g., embedded processors). Processor 910 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 904 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 904 thermal and temperature sensor management, and / or SoC 904 power state management. Each temperature sensor may be implemented as a ring oscillator whose output frequency is proportional to the temperature, and SoC 904 may use the ring oscillator to detect the temperature of CPU 906, GPU 908, and / or accelerator 914. 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 904 in a lower power state and / or place the vehicle 900 in a driver safety parking mode (e.g., parking the vehicle 900 safely).
[0152] The processor 910 may further include a set of embedded processors that may be used as an audio processing engine. The audio processing engine may be an audio subsystem that allows for full hardware support for multi-channel audio through multiple interfaces and a wide range of flexible audio I / O interfaces. In some examples, the audio processing engine is a dedicated processor core having a digital signal processor with dedicated RAM.
[0153] The processor 910 may further include an always-on processor engine that may provide the necessary hardware features to support low-power sensor management and wake-up use cases. The always-on processor engine may include a processor core, tightly coupled RAM, supporting peripherals (e.g., timers and interrupt controllers), various I / O controller peripherals, and routing logic.
[0154] The processor 910 may further include a safety cluster engine, which includes a dedicated processor subsystem that handles safety management of automotive applications. The safety cluster engine may include two or more processor cores, tightly coupled RAM, supporting peripherals (e.g., timers, interrupt controllers, etc.), and / or routing logic. In safety mode, the two or more cores may operate in lockstep mode and act as a single core with comparison logic to detect any differences between their operations.
[0155] Processor 910 may further include a real-time camera engine, which may include a dedicated processor subsystem for handling real-time camera management.
[0156] Processor 910 may further include a high dynamic range signal processor, which may include an image signal processor, which is a hardware engine that is part of the camera processing pipeline.
[0157] The processor 910 may include a video image compositer, which may be a processing block (e.g., implemented on a microprocessor) that implements video post-processing functions required for a video playback application to produce a final image for a player window. The video image compositer may perform lens distortion correction for the wide-angle camera 970, the surround camera 974, and / or for an in-cab surveillance camera sensor. The in-cab surveillance camera sensor is preferably monitored by a neural network running on another instance of the advanced SoC, configured to recognize in-cab events and respond accordingly. The in-cab system may perform lip reading to activate mobile phone service and place 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.
[0158] The video image compositer may include enhanced temporal noise reduction for both spatial and temporal noise reduction. For example, in the case of motion in the video, the noise reduction appropriately weights the spatial information and reduces the weight of information provided by adjacent frames. In the case where an image or portion of an image does not include motion, the temporal noise reduction performed by the video image compositer may use information from previous images to reduce noise in the current image.
[0159] The video image compositor may also be configured to perform stereoscopic rectification on the input stereoscopic footage frames. The video image compositor may further be used for user interface composition when the operating system desktop is in use and the GPU 908 does not need to continuously render new surfaces. Even when the GPU 908 is powered on and active for 3D rendering, the video image compositor may be used to offload the GPU 908 to improve performance and responsiveness.
[0160] The SoC 904 may further include a mobile industry processor interface (MIPI) camera serial interface for receiving video and input from a camera, a high-speed interface, and / or a video input block that may be used for camera and related pixel input functions. The SoC 904 may further include an input / output controller that may be controlled by software and may be used to receive I / O signals that are not committed to a specific role.
[0161] SoC 904 may further include a wide range of peripheral interfaces to enable communication with peripherals, audio codecs, power management and / or other devices. SoC 904 may be used to process data from cameras (connected via Gigabit multimedia serial link and Ethernet), sensors (e.g., LiDAR sensor 964, RADAR sensor 960, etc., which may be connected via Ethernet), data from bus 902 (e.g., speed of vehicle 900, steering wheel position, etc.), data from GNSS sensor 958 (connected via Ethernet or CAN bus). SoC 904 may further include dedicated high-performance mass storage controllers, which may include their own DMA engines, and which may be used to release CPU 906 from routine data management tasks.
[0162] SoC 904 can be an end-to-end platform with a flexible architecture that spans automation levels 3-5, thereby providing a comprehensive functional safety architecture that utilizes and efficiently uses computer vision and ADAS technologies to achieve diversity and redundancy, together with deep learning tools to provide a platform for a flexible and reliable driving software stack. SoC 904 can be faster, more reliable, and even more energy efficient and space efficient than conventional systems. For example, when combined with CPU 906, GPU 908, and data storage 916, accelerator 914 can provide a fast and efficient platform for level 3-5 autonomous vehicles.
[0163] The technology thus provides capabilities and functionality that cannot be achieved by conventional systems. For example, computer vision algorithms can be executed on CPUs, which can be configured using high-level programming languages such as the C programming language to execute a wide variety of processing algorithms across a wide variety of visual data. However, CPUs often fail to meet the performance requirements of many computer vision applications, such as those related to, for example, execution time and power consumption. In particular, many CPUs are unable to execute complex object detection algorithms in real time, which is a requirement for in-vehicle ADAS applications and a requirement for practical Level 3-5 autonomous vehicles.
[0164] In contrast to conventional systems, by providing a CPU complex, a GPU complex, and a hardware acceleration cluster, the techniques described herein allow multiple neural networks to be executed simultaneously and / or sequentially, and the results to be combined together to achieve Level 3-5 autonomous driving functions. For example, a CNN executed on a DLA or dGPU (e.g., GPU 920) may include text and word recognition, allowing the supercomputer to read and understand traffic signs, including signs for which a neural network has not been specifically trained. The DLA may further include a neural network capable of recognizing, interpreting, and providing semantic understanding of the signs, and passing that semantic understanding to a path planning module running on the CPU complex.
[0165] As another example, as required for Level 3, 4, or 5 driving, multiple neural networks may be running simultaneously. For example, a warning sign consisting of "Caution: Flashing Lights Indicate Icing Conditions" along with electric lights may be interpreted by several neural networks independently or collectively. The sign itself may be recognized as a traffic sign by a deployed first neural network (e.g., a trained neural network), and the text "Flashing Lights Indicate Icing Conditions" may be interpreted by a deployed second neural network that informs the vehicle's path planning software (preferably executing on a CPU complex) that icing conditions exist when the flashing lights are detected. The flashing lights may be recognized by operating a deployed third neural network over multiple frames that informs the vehicle's path planning software of the presence (or absence) of the flashing lights. All three neural networks may be running simultaneously, for example, within the DLA and / or on the GPU 908.
[0166] In some examples, a CNN for facial recognition and owner recognition can use data from the camera sensor to identify the presence of an authorized driver and / or owner of the vehicle 900. 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 904 provides security against theft and / or carjacking.
[0167] In another example, a CNN for emergency vehicle detection and identification can use data from microphone 996 to detect and identify emergency vehicle sirens. In contrast to conventional systems that use general classifiers to detect sirens and manually extract features, SoC 904 uses CNN to classify environmental and urban sounds and to classify visual data. In a preferred embodiment, the CNN running on the DLA is trained to identify the relative closing rate of emergency vehicles (e.g., by using the Doppler effect). The CNN can also be trained to identify emergency vehicles specific to the local area in which the vehicle is operating, as identified by the GNSS sensor 958. 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 962, the control program can be used to execute emergency vehicle safety routines to slow the vehicle, drive to the side of the road, stop the vehicle, and / or idle the vehicle until the emergency vehicle passes.
[0168] The vehicle may include a CPU 918 (e.g., a discrete CPU or dCPU) that may be coupled to the SoC 904 via a high-speed interconnect (e.g., PCIe). The CPU 918 may include, for example, an X86 processor. The CPU 918 may be used to perform any of a variety of functions, including, for example, arbitrating potentially inconsistent results between ADAS sensors and the SoC 904, and / or monitoring the status and health of the controller 936 and / or the infotainment SoC 930.
[0169] The vehicle 900 may include a GPU 920 (e.g., a discrete GPU or dGPU) that may be coupled to the SoC 904 via a high-speed interconnect (e.g., NVIDIA's NVLINK). The GPU 920 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 from sensors of the vehicle 900 (e.g., sensor data).
[0170] The vehicle 900 may further include a network interface 924, which may include one or more wireless antennas 926 (e.g., one or more wireless antennas for different communication protocols, such as cellular antennas, Bluetooth antennas, etc.). The network interface 924 can be used to enable wireless connection with the cloud (e.g., with a server 978 and / or other network devices), with other vehicles, and / or with a computing device (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 900 with information about vehicles approaching the vehicle 900 (e.g., vehicles in front of, to the side of, and / or behind the vehicle 900). This functionality can be part of the collaborative adaptive cruise control functionality of the vehicle 900.
[0171] The network interface 924 may include a SoC that provides modulation and demodulation functions and enables the controller 936 to communicate via a wireless network. The network interface 924 may include a radio frequency front end for up-conversion from baseband to radio frequency and down-conversion from radio frequency to baseband. Frequency conversion may be performed by a known process, and / or may be performed using a super-heterodyne process. In some examples, the radio frequency front end function may be provided by a separate chip. The network interface may include a wireless function for communicating via LTE, WCDMA, UMTS, GSM, CDMA2000, Bluetooth, Bluetooth LE, Wi-Fi, Z-wave, ZigBee, LoRaWAN, and / or other wireless protocols.
[0172] The vehicle 900 may further include data storage 928, which may include off-chip storage (e.g., outside the SoC 904). The data storage 928 may include one or more storage elements, including RAM, SRAM, DRAM, VRAM, flash memory, hard disk, and / or other components and / or devices that can store at least one bit of data.
[0173] The vehicle 900 may further include a GNSS sensor 958. The GNSS sensor 958 (e.g., GPS, assisted GPS sensor, differential GPS (DGPS) sensor, etc.) is used to assist in mapping, perception, occupancy grid generation, and / or path planning functions. Any number of GNSS sensors 958 may be used, including, for example and without limitation, GPS using a USB connector with an Ethernet to serial (RS-232) bridge.
[0174] The vehicle 900 may further include a RADAR sensor 960. The RADAR sensor 960 may be used by the vehicle 900 for remote vehicle detection even in darkness and / or inclement weather conditions. The RADAR functional safety level may be ASIL B. The RADAR sensor 960 may use CAN and / or bus 902 (e.g., to transmit data generated by the RADAR sensor 960) for control and access to object tracking data, accessing Ethernet in some examples to access raw data. A variety of RADAR sensor types may be used. For example and without limitation, the RADAR sensor 960 may be suitable for front, rear, and side RADAR use. In some examples, a pulse Doppler RADAR sensor is used.
[0175] The RADAR sensor 960 may include different configurations, such as long-range with a narrow field of view, short-range with a wide field of view, short-range side coverage, and the like. In some examples, the long-range RADAR may be used for adaptive cruise control functions. The long-range RADAR system may provide a wide field of view (e.g., within a range of 250m) achieved by two or more independent scans. The RADAR sensor 960 may help distinguish between static objects and moving objects, and may be used by the ADAS system for emergency braking assistance and forward collision warnings. The long-range RADAR sensor may include a single-station multimode RADAR with multiple (e.g., six or more) fixed RADAR antennas and high-speed CAN and FlexRay interfaces. In an example with six antennas, the central four antennas may create a focused beam pattern designed to record the surroundings of the vehicle 900 at a higher rate with minimal traffic interference from adjacent lanes. The other two antennas may expand the field of view, making it possible to quickly detect vehicles entering or leaving the lane of the vehicle 900.
[0176] As an example, a medium-range RADAR system may include a range of up to 960m (front) or 80m (rear) and a field of view of up to 42 degrees (front) or 950 degrees (rear). A short-range RADAR system may include, but is not limited to, a RADAR sensor designed to be mounted on both ends of a rear bumper. When mounted on both ends of a rear bumper, such a RADAR sensor system may create two beams that continuously monitor the blind spots behind and beside the vehicle.
[0177] Short-range RADAR systems can be used in ADAS systems for blind spot detection and / or lane change assistance.
[0178] The vehicle 900 may further include an ultrasonic sensor 962. The ultrasonic sensor 962, which may be placed on the front, rear, and / or sides of the vehicle 900, may be used for parking assistance and / or creating and updating an occupancy grid. A variety of ultrasonic sensors 962 may be used, and different ultrasonic sensors 962 may be used for different detection ranges (e.g., 2.5 m, 4 m). The ultrasonic sensor 962 may operate at a functional safety level of ASIL B.
[0179] The vehicle 900 may include a LiDAR sensor 964. The LiDAR sensor 964 may be used for object and pedestrian detection, emergency braking, collision avoidance, and / or other functions. The LiDAR sensor 964 may be ASIL B for functional safety level. In some examples, the vehicle 900 may include multiple LiDAR sensors 964 (e.g., two, four, six, etc.) that may use Ethernet (e.g., to provide data to a Gigabit Ethernet switch).
[0180] In some examples, the LiDAR sensor 964 may be able to provide a list of objects and their distances for a 360-degree field of view. Commercially available LiDAR sensors 964 may have, for example, an advertised range of approximately 900m, an accuracy of 2cm-3cm, and support for 900Mbps Ethernet connections. In some examples, one or more non-protruding LiDAR sensors 964 may be used. In such examples, the LiDAR sensor 964 may be implemented as a small device that can be embedded in the front, back, side, and / or corner of the vehicle 900. In such examples, the LiDAR sensor 964 may provide a field of view of up to 120 degrees horizontally and 35 degrees vertically, with a range of 200m, even for low reflectivity objects. The front-mounted LiDAR sensor 964 may be configured for a horizontal field of view between 45 and 135 degrees.
[0181] In some examples, LiDAR technologies such as 3D flash LiDAR may also be used. 3D flash LiDAR uses a flash of laser as an emission source to illuminate the vehicle's surroundings up to about 200m. The flash LiDAR unit includes a receiver that records the laser pulse transmission time and the reflected light on each pixel, which in turn corresponds to the range from the vehicle to the object. Flash LiDAR can allow highly accurate and distortion-free images of the surrounding environment to be generated with each laser flash. In some examples, four flash LiDAR sensors can be deployed, one on each side of the vehicle 900. Available 3D flash LiDAR systems include solid-state 3D staring array LiDAR cameras (e.g., non-scanning LiDAR devices) with no moving parts other than fans. The flash LiDAR device can use 5 nanosecond Class I (eye-safe) laser pulses per frame, and can capture the reflected laser light in the form of a 3D range point cloud and co-registered intensity data. By using flash LiDAR, and because flash LiDAR is a solid-state device with no moving parts, the LiDAR sensor 964 may be less susceptible to motion blur, vibration, and / or shock.
[0182] The vehicle may further include an IMU sensor 966. In some examples, the IMU sensor 966 may be located at the center of the rear axle of the vehicle 900. The IMU sensor 966 may include, for example and without limitation, an accelerometer, a magnetometer, a gyroscope, a magnetic compass, and / or other sensor types. In some examples, for example, in a six-axis application, the IMU sensor 966 may include an accelerometer and a gyroscope, and in a nine-axis application, the IMU sensor 966 may include an accelerometer, a gyroscope, and a magnetometer.
[0183] In some embodiments, the IMU sensor 966 can be implemented as a miniature high-performance GPS-assisted inertial navigation system (GPS / INS) that combines a micro-electromechanical system (MEMS) inertial sensor, a high-sensitivity GPS receiver, and an advanced Kalman filter algorithm to provide estimates of position, velocity, and attitude. Thus, in some examples, the IMU sensor 966 can enable the vehicle 900 to estimate heading by directly observing and correlating velocity changes from the GPS to the IMU sensor 966 without the need for input from a magnetic sensor. In some examples, the IMU sensor 966 and the GNSS sensor 958 can be combined into a single integrated unit.
[0184] The vehicle may include a microphone 996 positioned in and / or around the vehicle 900. The microphone 996 may be used for, among other things, emergency vehicle detection and identification.
[0185] The vehicle may further include any number of camera types, including a stereo camera 968, a wide angle camera 970, an infrared camera 972, a surround camera 974, a long-range and / or mid-range camera 998, and / or other camera types. These cameras may be used to capture image data around the entire periphery of the vehicle 900. The type of camera used depends on the embodiment and the requirements of the vehicle 900, and any combination of camera types may be used to provide the necessary coverage around the vehicle 900. In addition, 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 ... Fig.9A and Fig. 9B Described in more detail.
[0186] The vehicle 900 may further include a vibration sensor 942. The vibration sensor 942 may measure vibrations of components of the vehicle, such as axles. For example, changes in vibration may indicate changes in the road surface. In another example, when two or more vibration sensors 942 are used, the difference between the vibrations may be used to determine friction or slip of the road surface (e.g., when there is a vibration difference between a powered drive shaft and a freely rotating shaft).
[0187] The vehicle 900 may include an ADAS system 938. In some examples, the ADAS system 938 may include a SoC. The ADAS system 938 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.
[0188] The ACC system may use a RADAR sensor 960, a LiDAR sensor 964, and / or a camera. The ACC system may include a longitudinal ACC and / or a lateral ACC. The longitudinal ACC monitors and controls the distance to the vehicle immediately in front of the vehicle 900, and automatically adjusts the vehicle speed to maintain a safe distance from the vehicle in front. The lateral ACC performs distance keeping and suggests that the vehicle 900 change lanes when necessary. The lateral ACC is related to other ADAS applications such as LCA and CWS.
[0189] CACC uses information from other vehicles, which can be received indirectly from other vehicles via a wireless link or through a network connection (e.g., through the Internet) via a network interface 924 and / or a wireless antenna 926. 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 900 and in the same lane as it), while the I2V communication concept provides information about traffic farther ahead. The CACC system may include either or both of the I2V and V2V information sources. Given information about the vehicle ahead of the vehicle 900, CACC can be more reliable, and it is possible to improve the smoothness of traffic flow and reduce road congestion.
[0190] The FCW system is designed to alert the driver to hazards so that the driver can take corrective action. The FCW system uses a front camera and / or RADAR sensor 960 coupled to a dedicated processor, DSP, FPGA and / or ASIC, which is electrically coupled to driver feedback such as a display, speaker and / or vibrating components. The FCW system can provide warnings in the form of, for example, sound, visual warnings, vibrations and / or rapid brake pulses.
[0191] The AEB system detects an impending front collision with another vehicle or other object and can automatically apply the brakes if the driver does not take corrective action within specified time or distance parameters. The AEB system can use a front camera and / or RADAR sensor 960 coupled to a dedicated processor, DSP, FPGA and / or ASIC. When the AEB system detects a hazard, it typically first alerts the driver to take corrective action to avoid the collision, and if the driver does not take corrective action, the AEB system can automatically apply the brakes in an effort to prevent or at least mitigate the effects of the predicted collision. The AEB system may include technologies such as dynamic brake support and / or collision approach braking.
[0192] The LDW system provides visual, audible, and / or tactile warnings, such as steering wheel or seat vibrations, to alert the driver when the vehicle 900 crosses a lane marking. When the driver indicates an intention to leave the lane, by activating a turn signal, the LDW system is not activated. The LDW system may use a front-facing camera coupled to a dedicated processor, DSP, FPGA, and / or ASIC that is electrically coupled to driver feedback such as a display, speaker, and / or vibrating component.
[0193] The LKA system is a variation of the LDW system. If the vehicle 900 begins to leave the lane, the LKA system provides steering input or braking to correct the vehicle 900.
[0194] The BSW system detects and warns the driver of vehicles in the car's blind spot. The BSW system can provide visual, auditory 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 960 coupled to a dedicated processor, DSP, FPGA and / or ASIC, which is electrically coupled to driver feedback such as a display, speaker and / or vibration components.
[0195] The RCTW system can provide visual, audible, and / or tactile notifications when an object is detected outside the range of the rear camera while the vehicle 900 is in reverse. Some RCTW systems include AEB to ensure that the vehicle brakes are applied to avoid a crash. The RCTW system can use one or more rear RADAR sensors 960 coupled to a dedicated processor, DSP, FPGA, and / or ASIC that is electrically coupled to driver feedback such as a display, speaker, and / or vibrating component.
[0196] Conventional ADAS systems may be prone to false positive results, which may annoy and distract the driver, but are typically not catastrophic because the ADAS system alerts the driver and allows the driver to decide whether the safety condition really exists and take action accordingly. However, in the autonomous vehicle 900, in the case of conflicting results, the vehicle 900 itself must decide whether to pay attention to the results from the main computer or the auxiliary computer (e.g., the first controller 936 or the second controller 936). For example, in some embodiments, the ADAS system 938 can be a backup and / or auxiliary computer for providing perception information to the backup computer rationality module. The backup computer rationality monitor can run redundant and diverse software on hardware components to detect faults in perception and dynamic driving tasks. The output from the ADAS system 938 can be provided to the supervisory MCU. If the outputs from the main computer and the auxiliary computer conflict, the supervisory MCU must determine how to coordinate the conflict to ensure safe operation.
[0197] In some examples, the master computer can be configured to provide a confidence score to the supervisory MCU, indicating the master computer's confidence in the selected result. If the confidence score exceeds a threshold, the supervisory MCU can follow the direction of the master computer, regardless of whether the auxiliary computer provides conflicting or inconsistent results. In the case where the confidence score does not meet the threshold and the master computer and the auxiliary computer indicate different results (e.g., conflicts), the supervisory MCU can arbitrate between these computers to determine the appropriate result.
[0198] The supervisory MCU can be configured to run a neural network that is trained and configured to determine conditions under which the auxiliary computer provides a false alarm based on outputs from the primary computer and the auxiliary computer. Thus, the neural network in the supervisory MCU can learn when the output of the auxiliary computer can be trusted and when it cannot. For example, when the auxiliary 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 dangerous, such as a drainage grid or manhole cover that triggers an alarm. Similarly, when the auxiliary computer is a camera-based LDW system, the neural network in the supervisory MCU can learn to ignore the LDW when a cyclist or pedestrian is present and lane departure is actually the safest strategy. In an embodiment including a neural network running on the supervisory MCU, the supervisory MCU may include at least one of a DLA or GPU suitable for running the neural network with associated memory. In a preferred embodiment, the supervisory MCU may include a component of SoC 904 and / or be included as a component of SoC 904.
[0199] In other examples, the ADAS system 938 may include an auxiliary computer that uses traditional computer vision rules to perform ADAS functions. In this way, the auxiliary computer can use classic computer vision rules (if-then), and the presence of a neural network in the supervisory MCU can improve reliability, safety, and performance. For example, diverse implementations and intentional non-identity make the entire system more fault-tolerant, especially for failures caused by software (or software-hardware interface) functions. For example, if there is a software vulnerability or error in the software running on the main computer and the non-identical software code running on the auxiliary computer provides the same overall result, the supervisory MCU can be more confident that the overall result is correct and that the vulnerability in the software or hardware on the main computer does not cause a substantial error.
[0200] In some examples, the output of the ADAS system 938 can be fed to the perception block of the main computer and / or the dynamic driving task block of the main computer. For example, if the ADAS system 938 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 auxiliary computer can have its own neural network that is trained and thus reduces the risk of false positives as described herein.
[0201] The vehicle 900 may further include an infotainment SoC 930 (e.g., an in-vehicle infotainment system (IVI)). Although illustrated and described as an SoC, the infotainment system may not be an SoC and may include two or more discrete components. The infotainment SoC 930 may include a combination of hardware and software that may be used to provide audio (e.g., music, personal digital assistant, navigation instructions, news, radio, etc.), video (e.g., TV, movies, streaming media, etc.), phone (e.g., hands-free calling), network connectivity (e.g., LTE, Wi-Fi, etc.), and / or information services (e.g., navigation system, rear parking assistance, radio data system, vehicle-related information such as fuel level, total distance covered, brake fuel level, oil level, door open / closed, air filter information, etc.) to the vehicle 900. For example, the infotainment SoC 930 may include a radio, a disc player, a navigation system, a video player, USB and Bluetooth connections, an onboard computer, onboard entertainment, Wi-Fi, steering wheel audio controls, hands-free voice controls, a head-up display (HUD), an HMI display 934, 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 930 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 938, 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.
[0202] The infotainment SoC 930 may include GPU functionality. The infotainment SoC 930 may communicate with other devices, systems, and / or components of the vehicle 900 via a bus 902 (e.g., a CAN bus, Ethernet, etc.). In some examples, the infotainment SoC 930 may be coupled to a supervisory MCU so that in the event of a failure of a main controller 936 (e.g., a main and / or backup computer of the vehicle 900), the GPU of the infotainment system may perform some self-driving functions. In such an example, the infotainment SoC 930 may place the vehicle 900 in a driver safety parking mode as described herein.
[0203] The vehicle 900 may further include an instrument cluster 932 (e.g., a digital instrument panel, an electronic instrument cluster, a digital instrument panel, etc.). The instrument cluster 932 may include a controller and / or a supercomputer (e.g., a separate controller or a supercomputer). The instrument cluster 932 may include a set of instruments, such as a speedometer, a fuel level, an oil pressure, a tachometer, an odometer, a turn indicator, a shift position indicator, a seat belt warning light, a parking brake warning light, an engine fault light, an airbag (SRS) system information, a lighting control, a safety system control, navigation information, and the like. In some examples, information may be displayed and / or shared between the infotainment SoC 930 and the instrument cluster 932. In other words, the instrument cluster 932 may be included as part of the infotainment SoC 930, or vice versa.
[0204] Fig.9D For a cloud-based server and Fig.9A 900. System 976 may include a server 978, a network 990, and a vehicle including the vehicle 900. The server 978 may include a plurality of GPUs 984 (A) -984 (H) (collectively referred to herein as GPUs 984), PCIe switches 982 (A) -982 (H) (collectively referred to herein as PCIe switches 982), and / or CPUs 980 (A) -980 (B) (collectively referred to herein as CPUs 980). The GPUs 984, CPUs 980, and PCIe switches may be interconnected with high-speed interconnects and / or PCIe connections 986 such as, for example and without limitation, the NVLink interface 988 developed by NVIDIA. In some examples, the GPUs 984 are connected via NVLink and / or NVSwitch SoCs, and the GPUs 984 and PCIe switches 982 are connected via PCIe interconnects. Although eight GPUs 984, two CPUs 980, and two PCIe switches are illustrated, this is not intended to be limiting. Depending on the embodiment, each of the servers 978 may include any number of GPUs 984, CPUs 980, and / or PCIe switches. For example, each of the servers 978 may include eight, sixteen, thirty-two, and / or more GPUs 984.
[0205] Server 978 may receive image data over network 990 and from a vehicle, the image data representing images showing unexpected or changed road conditions, such as recently begun road work. Server 978 may transmit neural network 992, updated neural network 992, and / or map information 994, including information about traffic and road conditions, over network 990 and to the vehicle. Updates to map information 994 may include updates to HD map 922, such as information about construction sites, potholes, curves, flooding, or other obstacles. In some examples, neural network 992, updated neural network 992, and / or map information 994 may have been generated from new training and / or data received from any number of vehicles in the environment and / or based on experience from training performed at a data center (e.g., using server 978 and / or other servers).
[0206] Server 978 can be used to train a machine learning model (e.g., a neural network) based on training data. The training data can be generated by the vehicle, and / or can be generated in a simulation (e.g., using a game engine). In some examples, the training data is labeled (e.g., when the neural network benefits from supervised learning) and / or undergoes other preprocessing, while in other examples, the training data is not labeled and / or preprocessed (e.g., when the neural network does not require supervised learning). Training can be performed according to any one or more categories of machine learning techniques, including but not limited to: categories such as supervised training, semi-supervised training, unsupervised training, self-learning, reinforcement learning, joint learning, transfer learning, feature learning (including main components and cluster analysis), multilinear subspace learning, manifold learning, representation learning (including alternate dictionary learning), rule-based machine learning, anomaly detection, and any variants or combinations thereof. Once the machine learning model is trained, the machine learning model can be used by the vehicle (e.g., transmitted to the vehicle via network 990), and / or the machine learning model can be used by server 978 to remotely monitor the vehicle.
[0207] In some examples, server 978 can receive data from the vehicle and apply the data to the latest real-time neural network for real-time intelligent reasoning. Server 978 may include a deep learning supercomputer and / or a dedicated AI computer powered by GPU 984, such as DGX and DGX Station machines developed by NVIDIA. However, in some examples, server 978 may include a deep learning infrastructure of a data center powered only by CPUs.
[0208] The deep learning infrastructure of server 978 may be capable of rapid real-time inference, and may use this capability to assess and verify the health of the processors, software, and / or associated hardware in vehicle 900. For example, the deep learning infrastructure may receive periodic updates from vehicle 900, such as a sequence of images and / or objects located in the sequence of images that vehicle 900 has located (e.g., via computer vision and / or other machine learning object classification techniques). The deep learning infrastructure may run its own neural network to identify objects and compare them to the objects identified by vehicle 900, and if the results do not match and the infrastructure concludes that the AI in vehicle 900 has failed, then server 978 may transmit a signal to vehicle 900 instructing the fail-safe computer of vehicle 900 to take control, notify passengers, and complete a safe parking maneuver.
[0209] For reasoning, the server 978 may include a GPU 984 and one or more programmable reasoning accelerators (e.g., NVIDIA's TensorRT). The combination of GPU-powered servers and reasoning acceleration can make real-time responses possible. In other examples, such as when performance is not so important, CPU, FPGA, and other processor-powered servers can be used for reasoning.
[0210] Example computing device
[0211] Fig.10 1 is a block diagram of an example computing device 1000 suitable for implementing some embodiments of the present disclosure. The computing device 1000 may include an interconnect system 1002 that directly or indirectly couples the following devices: a memory 1004, one or more central processing units (CPUs) 1006, one or more graphics processing units (GPUs) 1008, a communication interface 1010, an input / output (I / O) port 1012, an input / output component 1014, a power supply 1016, one or more presentation components 1018 (e.g., a display), and one or more logic units 1020. In at least one embodiment, one or more computing devices 1000 may include one or more virtual machines (VMs), and / or any of their components may include virtual components (e.g., virtual hardware components). For non-limiting examples, one or more of the GPUs 1008 may include one or more vGPUs, one or more of the CPUs 1006 may include one or more vCPUs, and / or one or more of the logic units 1020 may include one or more virtual logic units. As such, one or more computing devices 1000 may include discrete components (eg, a full GPU dedicated to computing device 1000 ), virtual components (eg, a portion of a GPU dedicated to computing device 1000 ), or a combination thereof.
[0212] although Fig.10 The various blocks of are shown as being connected via an interconnect system 1002 having lines, but this is not intended to be limiting and is merely for clarity. For example, in some embodiments, a presentation component 1018 such as a display device may be considered an I / O component 1014 (e.g., if the display is a touch screen). As another example, CPU 1006 and / or GPU 1008 may include memory (e.g., memory 1004 may represent a storage device in addition to the memory of GPU 1008, CPU 1006, and / or other components). In other words, Fig.10 The computing devices described herein are illustrative only. No distinction is made between categories such as "workstations," "servers," "laptops," "desktops," "tablets," "client devices," "mobile devices," "handheld devices," "game consoles," "electronic control units (ECUs)," "virtual reality systems," and / or other device or system types, as all are considered Fig.10 within the range of computing devices.
[0213] The interconnection system 1002 may represent one or more links or buses, such as an address bus, a data bus, a control bus, or a combination thereof. The interconnection system 1002 may include one or more bus or link 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. For example, the CPU 1006 may be directly connected to the memory 1004. In addition, the CPU 1006 may be directly connected to the GPU 1008. In the case where there is a direct or point-to-point connection between components, the interconnection system 1002 may include a PCIe link to perform the connection. In these examples, the computing device 1000 does not need to include a PCI bus.
[0214] Memory 1004 may include any of a variety of computer-readable media. Computer-readable media may be any available media that can be accessed by computing device 1000. Computer-readable media may include volatile and nonvolatile media and removable and non-removable media. By way of example and not limitation, computer-readable media may include computer storage media and communication media.
[0215] Computer storage media may include volatile and nonvolatile media and / or removable and non-removable media implemented in any method or technology for storing information such as computer-readable instructions, data structures, program modules, and / or other data types. For example, memory 1004 may store computer-readable instructions (e.g., representing programs and / or program elements, such as an operating system). Computer storage media may include, but are not limited to, RAM, ROM, EEPROM, flash memory or other storage technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by computing device 1000. As used herein, computer storage media does not include the signals themselves.
[0216] 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 delivery media. The term "modulated data signal" may refer to a signal that has one or more of its characteristics set or changed in such a way that information is encoded into the signal. By way of example and not limitation, computer storage media may include wired media such as a wired network or a direct wired connection, and wireless media such as sound, RF, infrared, and other wireless media. Any combination of the above should also be included within the scope of computer readable media.
[0217] The CPU 1006 may be configured to execute at least some of the computer-readable instructions so as to control one or more components of the computing device 1000 to perform one or more of the methods and / or processes described herein. Each of the CPUs 1006 may include one or more cores (e.g., one, two, four, eight, twenty-eight, seventy-two, etc.) capable of simultaneously processing a large number of software threads. The CPU 1006 may include any type of processor, and may include different types of processors, depending on the type of computing device 1000 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 1000, the processor may be an Advanced RISC Machine (ARM) processor implemented using Reduced Instruction Set Computing (RISC) or an x86 processor implemented using Complex Instruction Set Computing (CISC). The computing device 1000 may also include one or more CPUs 1006 in addition to one or more microprocessors or supplementary coprocessors such as math coprocessors.
[0218] In addition to or as an alternative to CPU 1006, one or more GPUs 1008 may be configured to execute at least some of the computer-readable instructions to control one or more components of computing device 1000 to perform one or more of the methods and / or processes described herein. One or more GPUs 1008 may be integrated GPUs (e.g., with one or more CPUs 1006) and / or one or more GPUs 1008 may be discrete GPUs. In an embodiment, one or more GPUs 1008 may be coprocessors of one or more CPUs 1006. Computing device 1000 may use GPU 1008 to render graphics (e.g., 3D graphics) or perform general purpose computing. For example, one or more GPUs 1008 may be used for general purpose computing (GPGPU) on GPUs. One or more GPUs 1008 may include hundreds of thousands of cores capable of processing hundreds of thousands of software threads simultaneously. GPU 1008 may generate pixel data for outputting an image in response to a rendering command (e.g., a rendering command from CPU 1006 received via a host interface). GPU 1008 may include graphics memory, such as display memory, for storing pixel data or any other suitable data, such as GPGPU data. Display memory may be included as part of memory 1004. One or more GPUs 1008 may include two or more GPUs operating in parallel (e.g., via a link). The link may connect the GPUs directly (e.g., using NVLINK) or may connect the GPUs through a switch (e.g., using NVSwitch). When combined together, each GPU 1008 may generate pixel data or GPGPU data for different parts of the output or different outputs (e.g., a first GPU for a first image and a second GPU for a second image). Each GPU may include its own memory or may share memory with other GPUs.
[0219] In addition to or as an alternative to the CPU 1006 and / or the GPU 1008, the logic unit 1020 may be configured to execute at least some of the computer-readable instructions to control one or more components of the computing device 1000 to perform one or more of the methods and / or processes described herein. In an embodiment, the CPU 1006, the GPU 1008, and / or the logic unit 1020 may perform any combination of methods, processes, and / or portions thereof separately or jointly. The one or more logic units 1020 may be a part of and / or integrated into one or more of the CPU 1006 and / or the GPU 1008 and / or the one or more logic units 1020 may be a discrete component or otherwise external to the CPU 1006 and / or the GPU 1008. In an embodiment, the one or more logic units 1020 may be a coprocessor of the one or more CPUs 1006 and / or the one or more GPUs 1008.
[0220] Examples of logic unit 1020 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.
[0221] The communication interface 1010 may include one or more receivers, transmitters, and / or transceivers that enable the computing device 1000 to communicate with other computing devices via an electronic communication network, including wired and / or wireless communications. The communication interface 1010 may include components and functions that enable communication over any of a number of different networks, such as wireless networks (e.g., Wi-Fi, Z-wave, Bluetooth, Bluetooth LE, ZigBee, etc.), wired networks (e.g., communicating over Ethernet or InfiniBand), low power wide area networks (e.g., LoRaWAN, SigFox, etc.), and / or the Internet. In one or more embodiments, the one or more logic units 1020 and / or the communication interface 1010 may include one or more data processing units (DPUs) to transmit data received over a network and / or over the interconnect system 1002 directly to (e.g., memory) one or more GPUs 1008.
[0222] I / O ports 1012 can enable computing device 1000 to be logically coupled to other devices including I / O components 1014, presentation components 1018, and / or other components, some of which can be built into (e.g., integrated into) computing device 1000. Illustrative I / O components 1014 include microphones, mice, keyboards, joysticks, game pads, game controllers, satellite dishes, scanners, printers, wireless devices, and the like. I / O components 1014 can provide a natural user interface (NUI) that processes air gestures, voice, or other physiological inputs generated by a user. In some examples, the input can be transmitted to an appropriate network element for further processing. 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 computing device 1000 (as described in more detail below). Computing device 1000 may include a depth camera such as a stereo camera system, an infrared camera system, an RGB camera system, touch screen technology, and combinations of these for gesture detection and recognition. Additionally, computing device 1000 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 1000 to render immersive augmented reality or virtual reality.
[0223] The power supply 1016 may include a hardwired power supply, a battery power supply, or a combination thereof. The power supply 1016 may provide power to the computing device 1000 to enable the components of the computing device 1000 to operate.
[0224] The presentation component 1018 may include a display (e.g., a monitor, a touch screen, a television screen, a head-up display (HUD), other display types, or a combination thereof), a speaker, and / or other presentation components. The presentation component 1018 may receive data from other components (e.g., GPU 1008, CPU 1006, DPU, etc.) and output the data (e.g., as an image, video, sound, etc.).
[0225] Sample Data Center
[0226] Fig.11 An example data center 1100 that can be used in at least one embodiment of the present disclosure is shown. The data center 1100 can include a data center infrastructure layer 1110, a framework layer 1120, a software layer 1130, and / or an application layer 1140.
[0227] like Fig.11As shown, the data center infrastructure layer 1110 may include a resource coordinator 1112, grouped computing resources 1114, and node computing resources ("node CRs") 1116(1)-1116(N), where "N" represents any complete positive integer. In at least one embodiment, the node CRs 1116(1)-1116(N) may include, but are not limited to, any number of central processing units ("CPUs") or other processors (including 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 or disk drives), network input / output ("NW I / O") devices, network switches, virtual machines ("VMs"), power modules and / or cooling modules, etc. In some embodiments, one or more of the node CRs 1116(1)-1116(N) may correspond to a server having one or more of the above-mentioned computing resources. Furthermore, in some embodiments, nodes CR1116(1)-11161(N) may include one or more virtual components, such as vGPU, vCPU, etc., and / or one or more of nodes CR1116(1)-1116(N) may correspond to a virtual machine (VM).
[0228] In at least one embodiment, the grouped computing resources 1114 may include a separate grouping of node CR1116 housed in one or more racks (not shown), or many racks housed in a data center at different geographical locations (also not shown). The separate grouping of node CR1116 in the grouped computing resources 1114 may include grouped computing, network, memory or storage resources that can be configured or allocated to support one or more workloads. In at least one embodiment, several node CR1116 including 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 may also include any number of power modules, cooling modules and / or network switches in any combination.
[0229] Resource coordinator 1122 may configure or otherwise control one or more nodes CR 1116(1)-1116(N) and / or grouped computing resources 1114. In at least one embodiment, resource coordinator 1122 may include a software design infrastructure ("SDI") management entity for data center 1100. Resource coordinator 1122 may include hardware, software, or some combination thereof.
[0230] In at least one embodiment, Fig.11As shown, the framework layer 1120 may include a job scheduler 1133, a configuration manager 1134, a resource manager 1136, and / or a distributed file system 1138. The framework layer 1120 may include a framework that supports software 1132 of the software layer 1130 and / or one or more applications 1142 of the application layer 1140. The software 1132 or the application 1142 may include network-based service software or applications, such as those provided by Amazon Web Services, Google Cloud, and Microsoft Azure. The framework layer 1120 may be, but is not limited to, a free and open source software network application framework (such as Apache Spark) that can utilize the distributed file system 1138 for large-scale data processing (e.g., "big data"). TM (hereinafter referred to as "Spark")). In at least one embodiment, the job scheduler 1133 may include a Spark driver to facilitate scheduling workloads supported by different layers of the data center 1100. The configuration manager 1134 may be able to configure different layers, such as the software layer 1130 and the framework layer 1120 (which includes Spark and a distributed file system 1138 for supporting large-scale data processing). The resource manager 1136 may be able to manage clustered or grouped computing resources mapped to the distributed file system 1138 and the job scheduler 1133 or allocated to support the distributed file system 1138 and the job scheduler 1133. In at least one embodiment, the clustered or grouped computing resources may include grouped computing resources 1114 at the data center infrastructure layer 1110. The resource manager 1136 may coordinate with the resource coordinator 1112 to manage these mapped or allocated computing resources.
[0231] In at least one embodiment, the software 1132 included in the software layer 1130 may include software used by at least a portion of the node CRs 1116(1)-1116(N), the grouped computing resources 1114, and / or the distributed file system 1138 of the framework layer 1120. 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.
[0232] In at least one embodiment, the applications 1142 included in the application layer 1140 may include one or more types of applications used by at least a portion of the node CRs 1116(1)-1116(N), the grouped computing resources 1114, and / or the distributed file system 1138 of the framework layer 1120. The one or more types of applications may include, but are not limited to, any number of genomic 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.
[0233] In at least one embodiment, any of configuration manager 1134, resource manager 1136, and resource coordinator 1112 can implement any number and type of self-modification actions based on any amount and type of data acquired in any technically feasible manner. The self-modification actions can save a data center operator of data center 1100 from making potentially poor configuration decisions and potentially avoiding underutilized and / or poorly performing portions of the data center.
[0234] According to one or more embodiments described herein, data center 1100 may include tools, services, software, or other resources to train one or more machine learning models or use one or more machine learning models to predict or infer information. For example, one or more machine learning models may be trained by calculating weight parameters according to a neural network architecture using the software and / or computing resources described above with respect to data center 1100. In at least one embodiment, a trained or deployed machine learning model corresponding to one or more neural networks may be used to infer or predict information using the resources described above with respect to data center 1100 using weight parameters calculated by one or more training techniques, such as, but not limited to, those described herein.
[0235] In at least one embodiment, the data center 1100 may use a CPU, an application-specific integrated circuit (ASIC), a GPU, an FPGA, and / or other hardware (or virtual computing resources corresponding thereto) to perform training and / or inference using the above resources. In addition, one or more software and / or hardware resources described above may be configured to allow a user to train or perform services that infer information, such as image recognition, speech recognition, or other artificial intelligence services.
[0236] Example network environment
[0237] A network environment suitable for implementing embodiments of the present disclosure may include one or more client devices, servers, network attached storage (NAS), other backend devices, and / or other device types. The client devices, servers, and / or other device types (e.g., each device) may be configured to communicate with the client device, server, or other device types. Fig.10 The computing device 1000 of the embodiment of the present invention may be implemented on one or more instances of the computing device 1000 - for example, each device may include similar components, features and / or functions of the computing device 1000. 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 1100, and the example of the data center 1100 is described in this article. Fig.11 Describe in more detail.
[0238] The components of the network environment can communicate with each other through a network, which can be wired, wireless, or both. The network can include multiple networks, or one of multiple networks. For example, the network can include one or more wide area networks (WANs), one or more local area networks (LANs), one or more public networks, such as the Internet and / or a 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.
[0239] 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.
[0240] In at least one embodiment, the network environment may include one or more cloud-based network environments, distributed computing environments, combinations thereof, and the like. A cloud-based network environment may include a framework layer, a job scheduler, a resource manager, and a distributed file system implemented on one or more servers, which may include one or more core network servers and / or edge servers. The framework layer may include a framework for supporting software at the software layer and / or one or more applications at the application layer. The software or application may include network-based service software or application programs, respectively. In an embodiment, one or more client devices may use network-based service software or application programs (e.g., by accessing the service software and / or application programs via one or more application programming interfaces (APIs)). The framework layer may be, but is not limited to, a type of free and open source software network application framework, such as a distributed file system that may be used for large-scale data processing (e.g., "big data").
[0241] A cloud-based network environment can provide cloud computing and / or cloud storage that performs any combination of the computing and / or data storage functions (or one or more portions thereof) described herein. Any of these various functions can be distributed across multiple locations of 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 a user (e.g., a 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).
[0242] The client device may include Fig.10 At least some of the components, features, and functionality of the described example computing device 1000. By way of example and not limitation, the client device may be embodied as a personal computer (PC), a laptop computer, a mobile device, a smart phone, a tablet computer, a smart watch, a wearable computer, a personal digital assistant (PDA), an MP3 player, a virtual reality headset, a global positioning system (GPS) or device, a video player, a camera, a surveillance device or system, a vehicle, a vessel, an aircraft, a virtual machine, a drone, a robot, a handheld communication device, a hospital device, a gaming device or system, an entertainment system, an in-vehicle computer system, an embedded system controller, a remote control, an appliance, a consumer electronic device, a workstation, an edge device, any combination of the devices described, or any other suitable device.
[0243] The present disclosure may be described in the general context of machine-usable instructions or computer codes executed by a computer or other machine such as a personal digital assistant or other handheld device, including computer-executable instructions such as program modules. Typically, program modules including routines, programs, objects, components, data structures, etc. refer to codes that perform specific tasks or implement specific abstract data types. The present disclosure may be practiced in a variety of system configurations, including handheld devices, consumer electronics, general-purpose computers, more specialized computing devices, etc. The present disclosure may also be practiced in a distributed computing environment in which tasks are performed by remote processing devices linked through a communication network.
[0244] As used herein, the description of "and / or" about two or more elements should be interpreted as referring to only one element or a combination of elements. For example, "element A, element B and / or element C" may include only element A, only element B, only element C, element A and element B, element A and element C, element B and element C, or elements A, B and C. In addition, "at least one of element A or element B" may include at least one of element A, at least one of element B, or at least one of element A and at least one of element B. Further, "at least one of element A and element B" may include at least one of element A, at least one of element B, or at least one of element A and at least one of element B.
[0245] The subject matter of the present disclosure is described in detail herein to meet statutory requirements. However, the description itself is not intended to limit the scope of the present disclosure. On the contrary, the inventors have contemplated that the claimed subject matter may also be embodied in other ways to include steps different from the steps described herein in conjunction with other current or future technologies or combinations of similar steps. Moreover, although the terms "step" and / or "block" may be used herein to imply different elements of the method employed, these terms should not be interpreted as implying any particular order among or between the various steps disclosed herein, unless the order of the steps is explicitly described.
Claims
1. A method comprising: determining one or more first values of one or more parameters associated with one or more objects based at least on first sensor data generated using a sensor of a first type; determining one or more second values of the one or more parameters associated with the one or more objects based at least on second sensor data generated using a sensor of a second type; generating input data representing the one or more first values and the one or more second values; determining, using one or more neural networks and based at least on the input data, one or more third values of the one or more parameters associated with the one or more objects; as well as One or more operations are performed, by the machine, based at least on the one or more third values of the one or more parameters.
2. The method of claim 1, wherein generating the input data comprises: Generating said input data representing at least: one or more first inputs, the one or more first inputs indicating the one or more first values of the one or more parameters; as well as One or more second inputs, the one or more second inputs indicating the one or more second values of the one or more parameters.
3. The method of claim 1 , wherein the one or more parameters comprises a plurality of parameters, the one or more first values comprises a plurality of first values, and the one or more second values comprises a plurality of second values, and wherein generating the input data comprises: Generating said input data representing at least: a plurality of first grids, individual first grids of the plurality of first grids indicating at least a first value of the plurality of first values of a parameter of the plurality of parameters; as well as A plurality of second grids, individual second grids of the plurality of second grids indicating at least a second value of the plurality of second values of the parameter of the plurality of parameters.
4. The method of claim 3, wherein: The plurality of first grids includes a number of grids corresponding to the number of parameters in the plurality of parameters; and The plurality of second grids include a number of grids corresponding to the number of the parameters among the plurality of parameters.
5. The method of claim 1, generating the input data comprises: Generating said input data representing at least: a first set of inputs indicating at least a first portion of the one or more first values associated with a first object of the one or more objects; a second set of inputs indicating at least a second portion of the one or more first values associated with a second object of the one or more objects, the second object being within a threshold distance from the first object; a third set of inputs indicating at least a first portion of the one or more second values associated with the first object; as well as A fourth set of inputs indicates at least a second portion of the one or more second values associated with the second object.
6. The method of claim 1, wherein the input data represents one or more grids indicating the one or more first values and the one or more second values, and wherein an individual grid of the one or more grids includes a number of portions corresponding to a number of areas within an environment.
7. The method of claim 1 , wherein determining the one or more third values of the one or more parameters associated with the one or more objects comprises: generating, using the one or more neural networks and based at least on the input data, output data representing one or more grids, individual grids of the one or more grids indicating a third value of the one or more third values, the third value being associated with a parameter of the one or more parameters; as well as The one or more third values of the one or more parameters associated with the one or more objects are determined based at least on the one or more grids.
8. The method of claim 1, wherein: determining the one or more first values of the one or more parameters associated with the one or more objects using a first processing pipeline associated with a sensor of the first type; as well as Determining the one or more second values of the one or more parameters associated with the one or more objects uses a second processing pipeline associated with the second type of sensor.
9. The method of claim 1, wherein the one or more parameters include one or more of the following: a first position, the first position being associated with a first direction; a second position, the second position being associated with a second direction; a third position associated with a third direction; high; width; length; a first speed in the first direction; a second speed in the second direction; orientation; posture; or Classification.
10. A system comprising: One or more processing units, the one or more processing units being configured to: receiving first data representing one or more first values of one or more parameters associated with one or more objects, the first data generated using first sensor data associated with a sensor of a first type; receiving second data representing one or more second values of the one or more parameters associated with the one or more objects, the second data generated using second sensor data associated with a second type of sensor; generating input data representing the one or more first values and the one or more second values; as well as determining, using one or more neural networks and based at least on the input data, one or more third values of the one or more parameters associated with the one or more objects; One or more operations are performed, by the machine, based at least on the one or more third values.
11. The system of claim 10, wherein the generating of the input data comprises: Generating said input data representing at least: one or more first inputs, the one or more first inputs indicating the one or more first values of the one or more parameters; as well as One or more second inputs, the one or more second inputs indicating the one or more second values of the one or more parameters.
12. The system of claim 10, wherein the one or more parameters comprises a plurality of parameters, the one or more first values comprises a plurality of first values, and the one or more second values comprises a plurality of second values, and wherein the generating of the input data comprises: Generating said input data representing at least: a plurality of first inputs, individual first inputs of the plurality of first inputs indicating at least a first value of the plurality of first values of a parameter of the plurality of parameters; as well as A plurality of second inputs, individual second inputs of the plurality of second inputs indicating at least a second value of the plurality of second values of the parameter of the plurality of parameters.
13. The system of claim 12, wherein: The plurality of first inputs comprises a number of inputs corresponding to the number of parameters in the plurality of parameters; and The plurality of second inputs includes a number of inputs corresponding to a number of the parameters in the plurality of parameters.
14. The system of claim 12, wherein the plurality of first inputs and the plurality of second inputs are associated with a first object of the one or more objects, and wherein the input data further represents: a plurality of third inputs, individual third inputs of the plurality of third inputs indicating at least a third value of the plurality of first values of the parameter of the plurality of parameters; and a plurality of fourth inputs, individual fourth inputs of the plurality of fourth inputs indicating at least a fourth value of the plurality of second values of the parameter of the plurality of parameters, The plurality of third inputs and the plurality of fourth inputs are associated with second objects of the one or more second objects that are located within a threshold distance from the first object.
15. The system of claim 10, wherein the input data represents one or more inputs indicating the one or more first values and the one or more second values, wherein an individual one of the one or more inputs includes a number of portions corresponding to a number of areas within an environment.
16. The system of claim 10, wherein determining the one or more third values of the one or more parameters associated with the one or more objects comprises: generating, using the one or more neural networks and based at least on the input data, output data representing one or more outputs, an individual one of the one or more outputs indicating a third one of the one or more third values, the third value being associated with a parameter of the one or more parameters; as well as The one or more third values of the one or more parameters associated with the one or more objects are determined based at least on the one or more outputs.
17. The system of claim 10, wherein: generating the one or more first values of the one or more parameters associated with the one or more objects using a first processing pipeline associated with the first type of sensor; and The one or more second values of the one or more parameters associated with the one or more objects are generated using a second processing pipeline associated with the second type of sensor.
18. The system of claim 10, wherein the system is 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 implemented using edge devices; Systems implemented using robots; Systems for performing conversational AI operations; A system implementing one or more large language models (LLMs); Systems for generating synthetic data; A system combining 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 processor comprising: One or more processing units, the one or more processing units being used to perform one or more control operations by the machine based at least on a fused output of a neural network, the fused output being determined based at least on the neural network processing one or more first inputs associated with one or more first values of one or more parameters and one or more second inputs associated with one or more second values of the one or more parameters, the one or more first inputs being generated using first input data associated with a first type of sensor, and the one or more second inputs being generated using second sensor data associated with a second type of sensor.
20. The processor of claim 19, wherein the processor is 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 implemented using edge devices; Systems implemented using robots; Systems for performing conversational AI operations; A system implementing one or more large language models (LLMs); Systems for generating synthetic data; A system combining 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.
