Systems and methods for monitoring test data for autonomous operation of an autonomous vehicle

Abstract

Systems and methods of monitoring test data for autonomous operation of an autonomous vehicle are disclosed. A method for autonomous vehicle test data distribution and analysis is described. The method includes uploading drive session data from a computer of a drive site to a network-attached storage of the drive site. The method also includes uploading the drive session data from the network-attached storage of the drive site to a cloud-based storage location. The method further includes distributing the drive session data and work units from the cloud-based storage location to at least one research site separate from the drive site. The method further includes processing the drive session data by the at least one research site according to analysis / processing tasks associated with the work units.

Description

Technical Field

[0001] Certain aspects of this disclosure generally relate to machine learning, and more specifically, to systems and methods for monitoring test data of the autonomous operation of self-driving vehicles. Background Technology

[0002] Autonomous agents, such as self-driving cars and robots, are evolving rapidly. Self-driving cars rely on various methods of perceiving their environment. Unfortunately, the methods self-driving cars use to perceive their surroundings are not entirely reliable. Furthermore, because self-driving cars must interact with other vehicles, many critical issues arise. For example, a key issue is how to use machine learning to design vehicle control for autonomous vehicles.

[0003] Unfortunately, in situations involving complex interactions between vehicles (e.g., a controlled (autonomous) vehicle merging into a traffic lane), vehicle control via machine learning may be ineffective. Machine learning techniques for vehicle control are desired, enabling the selection of appropriate vehicle control actions for the autonomous vehicle. For example, the speed / acceleration / steering angle selected by the controlled (autonomous) vehicle can be applied as a vehicle control action. The autonomous test vehicle can then operate based on the selected vehicle control action. Unfortunately, test autonomous vehicles generate a massive amount of data (e.g., 100 gigabytes (GB)) during test runs. Systems and methods for accessing this test data at different research locations are desired. Summary of the Invention

[0004] A method for distributing and analyzing test data for autonomous vehicles is described. The method includes: uploading driving session data from a computer at a driving station to a network-attached storage device at the driving station. The method also includes: uploading the driving session data from the network-attached storage device at the driving station to a cloud-based storage location. The method further includes: distributing the driving session data and work units from the cloud-based storage location to at least one research station separate from the driving station. The method further includes: having said at least one research station process the driving session data according to analysis / processing tasks associated with the work units.

[0005] A non-transitory computer-readable medium is described, containing program code for the distribution and analysis of test data for autonomous vehicles. The program code is executed by a processor. The non-transitory computer-readable medium includes program code for uploading driving session data from a computer at a driving station to a network-attached memory at the driving station. The non-transitory computer-readable medium also includes program code for uploading driving session data from the network-attached memory at the driving station to a cloud-based storage location. The non-transitory computer-readable medium further includes program code for distributing driving session data and work units from the cloud-based storage location to at least one research station separate from the driving station. The non-transitory computer-readable medium also includes program code for processing the driving session data by said at least one research station according to analysis / processing tasks associated with work units.

[0006] A system for distributing and analyzing test data from autonomous vehicles is described. The system includes a driving station comprising a network-attached storage device and a computer. The computer is configured to upload driving session data from the computer at the driving station to the network-attached storage device in response to the insertion of a session data storage module of the test vehicle. The network-attached storage device is configured to upload the driving session data from the network-attached storage device at the driving station to a cloud-based storage location. The cloud-based storage location of the system is configured to distribute the driving session data and work units to research stations separate from the driving stations. At least two research stations are configured to process the driving session data according to analysis / processing tasks associated with work units received by the two research stations.

[0007] This has provided a fairly broad overview of the features and technical advantages of this disclosure in order to better understand the detailed description that follows. Other features and advantages of this disclosure will now be described. Those skilled in the art will understand that this disclosure can be readily used as a basis for modifying or designing other structures to achieve the same purpose as this disclosure. Those skilled in the art will also recognize that such equivalent structures do not depart from the teachings of this disclosure as set forth in the appended claims. When considered in conjunction with the accompanying drawings, the novel features considered to be features of this disclosure, the organization and methods of operation relating to the novel features, and other objects and advantages will be better understood from the following description. However, it will be clearly understood that each drawing is provided only for illustrative and descriptive purposes and is not intended to define any limitation of this disclosure. Attached Figure Description

[0008] When considered in conjunction with the accompanying drawings, similar reference numerals are used throughout the drawings, and the features, nature, and advantages of this disclosure will become clearer from the detailed description set forth below.

[0009] Figure 1The figure illustrates an example implementation of designing a neural network for a system-on-a-chip (SOC) using an autonomous vehicle test data distribution and analysis system, according to various aspects of this disclosure.

[0010] Figure 2 This is a block diagram illustrating an exemplary software architecture for modular artificial intelligence (AI) functions that can be used in an autonomous vehicle test data distribution and analysis system according to various aspects of this disclosure.

[0011] Figure 3 This is a diagram illustrating the hardware implementation of a system for distributing and analyzing test data for autonomous vehicles according to various aspects of this disclosure.

[0012] Figure 4 This is a diagram illustrating the ingestion process of a system and method for monitoring the status of test data at various memory storage locations and for running tasks on the test data, according to various aspects of this disclosure.

[0013] Figure 5 The figure illustrates sensor data images captured by a test autonomous vehicle operating based on a driving stack (e.g., a test vehicle application module) according to various aspects of this disclosure.

[0014] Figure 6 This is a flowchart illustrating a method for distributing and analyzing test data for autonomous vehicles according to various aspects of this disclosure.

[0015] Specific implementation method

[0016] The detailed descriptions that follow, taken in conjunction with the accompanying drawings, are intended as descriptions of various configurations, and not as representing the only configuration in which the concepts described herein can be practiced. The detailed descriptions include specific details intended to provide a comprehensive understanding of the concepts. However, it will be clear to those skilled in the art that these concepts can be practiced without these specific details. In some cases, well-known structures and components are shown in block diagram form to avoid obscuring these concepts.

[0017] Based on the teachings, those skilled in the art should understand that the scope of this disclosure is intended to cover any aspect of this disclosure, whether implemented independently of or in combination with any other aspect of this disclosure. For example, any number of the aspects set forth can be used to implement an apparatus or practice. Furthermore, the scope of this disclosure is intended to cover such apparatuses or methods practiced using structures, functions, or structures and functions other than those set forth in this disclosure. It should be understood that any aspect disclosed in this disclosure may be embodied by one or more elements of the claims.

[0018] While specific aspects are described herein, numerous variations and permutations of these aspects fall within the scope of this disclosure. Although some benefits and advantages of preferred aspects have been mentioned, the scope of this disclosure is not intended to be limited to specific benefits, uses, or objectives. Rather, the aspects of this disclosure are intended to be broadly applicable to different technologies, system configurations, networks, and protocols, some of which are illustrated by way of example in the figures and in the following description of preferred aspects. The detailed description and figures of this disclosure are illustrative only and are not intended to limit the scope of this disclosure as defined by the appended claims and their equivalents.

[0019] Autonomous agents, such as self-driving cars and robots, are evolving rapidly. Self-driving cars rely on various methods of perceiving their environment. Unfortunately, the methods self-driving cars use to perceive their surroundings are not entirely reliable. Furthermore, because self-driving cars must interact with other vehicles, many critical issues arise. For example, a key issue is how to use machine learning to design vehicle control for autonomous vehicles.

[0020] The automation of vehicle control on highways is rapidly advancing. These automated vehicles are expected to reduce traffic accidents and improve traffic efficiency. Specifically, machine learning techniques for vehicle control are expected to be used by selecting appropriate vehicle control actions for the autonomous vehicle. For example, the speed / acceleration / steering angle selected by the controlled (autonomous) vehicle can be applied as vehicle control actions. Unfortunately, in situations involving complex interactions between vehicles (e.g., when a controlled (autonomous) vehicle merges into a traffic lane), vehicle control via machine learning may be ineffective.

[0021] Specifically, safety is a critical issue when building autonomous agents that operate in human environments. This is especially true for autonomous driving, where high speeds, diverse environments, and complex dynamic interactions with numerous traffic participants, including vulnerable road users, present a significant challenge. Testing and validation of machine learning techniques for vehicle control are desired, particularly through the selection of appropriate vehicle control actions for the autonomous vehicle. For example, autonomous test vehicles can operate based on selected vehicle control actions. Unfortunately, testing autonomous vehicles generates a massive amount of data.

[0022] For example, testing autonomous vehicles generates enormous amounts of data (e.g., 100 gigabytes (GB)) during test runs. Transmitting such a large volume of data is difficult and expensive. Furthermore, storing this data in cloud-based storage for researchers' use may be impractical, as downloading and / or processing the data requires a significant amount of time depending on the researchers' needs. Therefore, ideally, the data should be located near the researchers. Systems and methods for accessing this test data from different research locations are desired.

[0023] This disclosure relates to systems and methods for monitoring the status of test data at various memory storage locations and for running tasks on the test data. Various aspects of this disclosure provide a test data pipeline that distributes relevant data from a test autonomous vehicle to a researcher's location for rapid use by the researcher. The test method begins when the test autonomous vehicle completes a test run and enters a garage. In this example, the sensor data storage module is removed from the test autonomous vehicle and inserted into a computer at the garage.

[0024] In this aspect of the disclosure, sensor data is provided to a cloud-based storage location and removed from the source site (e.g., a garage for testing autonomous vehicles). The raw sensor data can be stored at the cloud-based storage location. Furthermore, any irrelevant information can be removed from the sensor data; however, the complete, unfiltered version of the sensor data remains at the cloud-based storage location. The filtered information is then provided to a processing pipeline used as a distributed computing network. This aspect of the disclosure aims to enhance the ability to collect and rapidly distribute autonomous vehicle test data within a data pipeline, making it readily available to researchers.

[0025] According to this aspect of the disclosure, the processing pipeline enables each researcher (or research site) to receive sensor data and a processing unit. That is, researchers receive information and contribute processing power to process the ingested sensor data. Processing (e.g., work) units can correspond to specific analytical or processing tasks. For example, one research location might be responsible for indexing the sensor data to make it compatible with keyword searches. Thus, the assigned work unit could be processing the sensor data for indexing. Another research location might be responsible for performing machine learning tasks on the sensor data. Once a research location completes its work unit, the output is then pushed to a cloud-based storage location and to other research locations where the output can be used.

[0026] Figure 1The figure illustrates an example implementation of the system and method for distributing and analyzing autonomous vehicle test data using a system-on-a-chip (SOC) 100 for a vehicle vision system for an autonomous vehicle 140, as described above. According to certain aspects of this disclosure, the SOC 100 may include a single processor or a multi-core processor (e.g., a central processing unit (CPU) 102). Variables (e.g., neural signals and synaptic weights) associated with computing devices (e.g., weighted neural networks), system parameters, latency, frequency slot information, and task information may be stored in a memory block. The memory block may be associated with a neural processing unit (NPU) 108, CPU 102, graphics processing unit (GPU) 104, digital signal processor (DSP) 106, dedicated memory block 118, or may be distributed across multiple blocks. Instructions executed on the processor (e.g., CPU 102) may be loaded from the program memory associated with CPU 102 or from dedicated memory block 118.

[0027] The SOC 100 may also include additional processing blocks configured to perform specific functions, such as GPU 104, DSP 106, and connectivity block 110. Connectivity block 110 may include 4G LTE connectivity, unlicensed Wi-Fi connectivity, USB connectivity, etc. Connections, etc. Furthermore, the multimedia processor 112, in conjunction with the display 130, can select certified vehicle control actions, for example, based on the display 130 illustrating a view of the vehicle.

[0028] In some aspects, the NPU 108 may be implemented in the CPU 102, DSP 106, and / or GPU 104. The SOC 100 may also include a sensor processor 114, an image signal processor (ISP) 116, and / or navigation 120, which may include, for example, a Global Positioning System (GPS). The SOC 100 may be based on an Advanced Reduced Instruction Set Machine (ARM) instruction set, etc. In another aspect of this disclosure, the SOC 100 may be a server computer communicating with an autonomous vehicle 140. In this arrangement, the autonomous vehicle 140 may include the processor and other features of the SOC 100.

[0029] In this aspect of the disclosure, instructions loaded into the processor (e.g., CPU 102) or NPU 108 of the autonomous vehicle 140 may include code for uploading driving session data from a computer at the driving station to a network-attached memory of the driving station based on images captured by sensor processor 114. Instructions loaded into the processor (e.g., CPU 102) may also include code for uploading driving session data from the network-attached memory of the driving station to a cloud-based storage location in response to images captured by sensor processor 114. Instructions loaded into the processor (e.g., CPU 102) may also include code for distributing driving session data and work units from the cloud-based storage location to at least one research station separate from the driving station. Instructions loaded into the processor (e.g., CPU 102) may also include code for the at least one research station to process the driving session data according to analysis / processing tasks associated with the work units.

[0030] Figure 2 This is a block diagram illustrating a modular software architecture 200 for an autonomous vehicle test data distribution and analysis system, according to various aspects of this disclosure. Using this architecture, the planner / controller application 202 can be designed such that various processing blocks of the SOC 220 (e.g., CPU 222, DSP 224, GPU 226, and / or NPU 228) perform supporting computations during the run-time operation of the planner / controller application 202. Although Figure 2 A software architecture 200 for the distribution and analysis of test data from autonomous vehicles is described, but it should be recognized that the distribution and analysis of vehicle test data is not limited to autonomous agents. According to various aspects of this disclosure, the vehicle test data distribution and analysis capabilities are applicable to any vehicle type.

[0031] Planner / controller application 202 can be configured to invoke functions defined in user space 20, which may provide, for example, vehicle test data distribution and analysis services. Planner / controller application 202 can request compiled code associated with libraries defined in test data pipeline application programming interface (API) 206. Test data pipeline API 206 is configured to distribute test sensor data to cloud-based storage locations and remove test data provided to test data pipeline API 206 from source sites (e.g., garages testing autonomous vehicles). In response, compiled code for test data analysis API 207 enables each researcher (or research site) to receive sensor data and processing units. That is, researchers receive information and contribute processing power to process the ingested sensor data. According to test data analysis API 207, processing (e.g., work) units can correspond to specific analysis or processing tasks.

[0032] The runtime engine 208 can be compiled code of the runtime framework, and it can also be accessed by the planner / controller application 202. The planner / controller application 202 can cause the runtime engine 208 to take actions, for example, on the distribution and analysis of vehicle test data from sensor data of the test autonomous vehicle. When the autonomous vehicle encounters a safety situation, the runtime engine 208 can then send signals to the operating system 210, such as the Linux kernel 212, running on the SOC 220. Figure 2 The diagram illustrates the Linux kernel 212 as a software architecture for distributing and analyzing test data from autonomous vehicles. However, it should be understood that aspects of this disclosure are not limited to this exemplary software architecture. For example, other kernels can provide software architectures to support the distribution and analysis of test data from autonomous vehicles.

[0033] Operating system 210 then enables computation to be performed on CPU 222, DSP 224, GPU 226, NPU 228, or some combination thereof. CPU 222 is directly accessible by operating system 210, and other processing blocks are accessible via drivers, such as drivers 214-218 for DSP 224, GPU 226, or NPU 228. In the illustrated example, the deep neural network can be configured to run on a combination of processing blocks such as CPU 222 and GPU 226, or it can run on NPU 228 (if present).

[0034] The increasing complexity of software in autonomous vehicles makes ensuring their reliability more difficult. For example, despite improvements in overall safety measures, the risk of unexpected catastrophic failures remains. Specifically, safety is a critical issue when building autonomous agents that operate in human environments. Especially for autonomous driving, safety is a formidable challenge due to high speeds, diverse environments, and complex dynamic interactions with numerous traffic participants, including vulnerable road users. Testing and validation of machine learning techniques for vehicle control are desired, aiming to achieve this by selecting appropriate vehicle control actions for the autonomous vehicle. For example, autonomous test vehicles can operate based on selected vehicle control actions. Unfortunately, the autonomous vehicles being tested generate a massive amount of data.

[0035] For example, testing autonomous vehicles generates enormous amounts of data (e.g., 100 gigabytes (GB)) during test runs. Transmitting such a large volume of data is difficult and expensive. Furthermore, storing this data in cloud-based storage for researchers' use may be impractical, as downloading and / or processing the data requires a significant amount of time depending on the researchers' needs. Therefore, ideally, the data should be located near the researchers. Systems and methods for accessing this test data from different research locations are desired.

[0036] This disclosure relates to systems and methods for monitoring the status of test data at various memory storage locations and for running tasks on the test data. Various aspects of this disclosure provide a test data pipeline that distributes relevant data from a test autonomous vehicle to a researcher's location for rapid use by the researcher. The test method begins when the test autonomous vehicle completes a test run and enters a garage. In this example, the sensor data storage module is removed from the test autonomous vehicle and inserted into a computer at the garage.

[0037] According to this aspect of the disclosure, the processing pipeline enables each researcher (or research site) to receive sensor data and a processing unit. That is, researchers receive information and contribute processing power to process the ingested sensor data. Processing (e.g., work) units can correspond to specific analytical or processing tasks. For example, one research location might be responsible for indexing the sensor data to make it compatible with keyword searches. Thus, the assigned work unit could be processing the sensor data for indexing. Another research location might be responsible for performing machine learning tasks on the sensor data. Once a research location completes its work unit, the output is then pushed to a cloud-based storage location and to other research locations where the output can be used.

[0038] Figure 3 This diagram illustrates a hardware implementation of an autonomous vehicle test data distribution and analysis system 300 according to various aspects of this disclosure. The autonomous vehicle test data distribution and analysis system 300 can be configured to improve autonomous vehicle testing by using distributed analysis and processing of driving test session data from origin sites. The autonomous vehicle test data distribution and analysis system 300 includes a test agent control system 301, which can be a component of the vehicle, robotic equipment, or other non-autonomous equipment (e.g., a non-autonomous vehicle, a shared car, etc.). For example, as... Figure 3As shown, the test agent control system 301 is a component of the test autonomous vehicle 350. Aspects of this disclosure are not limited to the test agent control system 301 being a component of the test autonomous vehicle 350. Other devices, such as buses, motorcycles, or other similar non-autonomous vehicles, are also contemplated for implementing the test agent control system 301. In this example, the test autonomous vehicle 350 can be autonomous or semi-autonomous; however, other configurations of the test autonomous vehicle 350 are contemplated.

[0039] The test agent control system 301 can be implemented using an interconnect architecture generally represented by interconnect 346. Depending on the specific application and overall design constraints of the test agent control system 301, interconnect 346 may include any number of point-to-point interconnects, buses, and / or bridges. Interconnect 346 links together various circuits including one or more processors and / or hardware modules, represented by sensor module 302, vehicle perception module 310, processor 320, computer-readable medium 322, communication module 324, on-board unit 326, position module 328, motion module 329, planner module 330, and controller module 340. Interconnect 346 may also link various other circuits such as timing sources, peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further.

[0040] The test agent control system 301 includes a transceiver 342 coupled to a sensor module 302, a vehicle perception module 310, a processor 320, a computer-readable medium 322, a communication module 324, an on-board unit 326, a position module 328, a motion module 329, a planner module 330, and a controller module 340. The transceiver 342 is coupled to an antenna 344. The transceiver 342 communicates with various other devices via a transmission medium. For example, the transceiver 342 can receive commands via transmissions from a user or a connected vehicle. In this example, the transceiver 342 can receive / send information from / to connected vehicles near the test autonomous vehicle 350 from / to the vehicle perception module 310.

[0041] The test agent control system 301 includes a processor 320 coupled to a computer-readable medium 322. The processor 320 performs processing including executing software stored on the computer-readable medium 322 to provide functions according to this disclosure. When executed by the processor 320, the software causes the test agent control system 301 to perform various functions described for autonomous vehicle test data distribution and analysis for the test autonomous vehicle 350 or for any module (e.g., 302, 310, 324, 328, 329, 330 and / or 340). The computer-readable medium 322 can also be used to store data manipulated by the processor 320 when the software is executed.

[0042] Sensor module 302 can acquire measurement results via different sensors, such as first sensor 306 and second sensor 304. First sensor 306 can be a vision sensor (e.g., a stereo camera or red-green-blue (RGB) camera) for capturing 2D images. Second sensor 304 can be a ranging sensor, such as a light detection and ranging (LIDAR) sensor or a radio detection and ranging (RADAR) sensor. Of course, aspects of this disclosure are not limited to the aforementioned sensors; other types of sensors (e.g., thermal, sonar, and / or laser) are also contemplated for use with first sensor 306 or second sensor 304.

[0043] The measurement results of the first sensor 306 and the second sensor 304 can be processed by the processor 320, sensor module 302, vehicle perception module 310, communication module 324, on-board unit 326, position module 328, movement module 329, planner module 330, and / or controller module 340. In conjunction with the computer-readable medium 322, the measurement results of the first sensor 306 and the second sensor 304 are processed to achieve the functions described herein. In one configuration, data captured by the first sensor 306 and the second sensor 304 can be transmitted to a connected vehicle via transceiver 342. The first sensor 306 and the second sensor 304 can be coupled to or communicate with the test autonomous vehicle 350.

[0044] The location module 328 can determine the position of the test autonomous vehicle 350. For example, the location module 328 can use a Global Positioning System (GPS) to determine the position of the test autonomous vehicle 350. The location module 328 can implement a GPS unit compatible with Dedicated Short Range Communication (DSRC). DSRC-compliant GPS units include hardware and software that make the test autonomous vehicle 350 and / or location module 328 compatible with one or more of the following DSRC standards (including any derivatives or branches thereof): EN 12253:2004 Dedicated Short Range Communications—Physical Layer Using 5.8 GHz Microwaves (under review); EN 12795:2002 Dedicated Short Range Communications (DSRC)—DSRC Data Link Layer: Media Access and Logical Link Control (under review); EN 12834:2002 Dedicated Short Range Communications—Application Layer (under review); EN 13372:2004 Dedicated Short Range Communications (DSRC)—DSRC Profile for RTTT Applications (under review); and EN ISO14906:2004 Electronic Toll Collection—Application Interface.

[0045] Communication module 324 can facilitate communication via transceiver 342. For example, communication module 324 can be configured to provide communication capabilities via various wireless protocols, such as 5G New Radio (NR), Wi-Fi, LTE, 4G, 3G, etc. Communication module 324 can also communicate with other components of the test autonomous vehicle 350 that are not part of the test agent control system 301. Transceiver 342 can be a communication channel via network access point 360. This communication channel can include DSRC, LTE, LTE-D2D, mmWave, Wi-Fi (infrastructure mode), Wi-Fi (ad-hoc mode), visible light communication, television blank space communication, satellite communication, full-duplex wireless communication, or any other wireless communication protocol such as those mentioned herein.

[0046] The test agent control system 301 also includes a planner module 330 for route planning and a controller module 340 for controlling the movement of the test autonomous vehicle 350 via a motion module 329 for autonomous operation of the test autonomous vehicle 350. In one configuration, the controller module 340 can override user input when it is anticipated (e.g., predicted) that a collision will occur based on the autonomy level of the test autonomous vehicle 350. Modules can be software modules running in processor 320, software modules residing / stored in computer-readable medium 322 and / or hardware modules coupled to processor 320, or combinations thereof.

[0047] The National Highway Traffic Safety Administration (“NHTSA”) has defined different “levels” for autonomous vehicles (e.g., Level 0, Level 1, Level 2, Level 3, Level 4, and Level 5). For example, if an autonomous vehicle has a higher level number compared to another autonomous vehicle (e.g., Level 3 has a higher level number compared to Level 2 or Level 1), then the autonomous vehicle with the higher level number offers a greater combination and number of autonomous features relative to the vehicle with the lower level number. These different levels of autonomous vehicles are briefly described below.

[0048] Level 0: In Level 0 vehicles, the Advanced Driver Assistance Systems (ADAS) feature set installed in the vehicle does not provide any vehicle control, but can issue warnings to the driver. Level 0 vehicles are not autonomous or semi-autonomous vehicles.

[0049] Level 1: In a Level 1 vehicle, the driver is ready to take driving control of the autonomous vehicle at any time. The set of ADAS features installed in the autonomous vehicle can provide autonomous features in any combination, such as: adaptive cruise control (“ACC”); parking assist with automatic steering; and Lane Keeping Assist (“LKA”) Type II.

[0050] Level 2: In Level 2 vehicles, the driver is responsible for detecting objects and events in the road environment and responding if the ADAS feature set installed in the autonomous vehicle fails to respond correctly (based on the driver's subjective judgment). The ADAS feature set installed in the autonomous vehicle may include acceleration, braking, and steering. In Level 2 vehicles, the ADAS feature set installed in the autonomous vehicle can be deactivated immediately upon driver takeover.

[0051] Level 3: In a Level 3 ADAS vehicle, within a known, limited environment (such as a highway), the driver is able to safely divert his / her attention from driving tasks, but must still be prepared to take control of the autonomous vehicle when needed.

[0052] Level 4: In Level 4 vehicles, the ADAS feature set installed in the autonomous vehicle is capable of controlling the autonomous vehicle in all environments except for a few, such as severe weather. The driver of a Level 4 vehicle activates the automated system (which includes the ADAS feature set installed in the vehicle) only when it is safe to do so. When an automated Level 4 vehicle is activated, no driver attention is required for the autonomous vehicle to operate safely and consistently within accepted guidelines.

[0053] Level 5: In Level 5 vehicles, no human intervention is involved, except for setting the destination and starting the system. The automated system can drive to any location it is legally allowed to drive and make its own decisions (which may vary depending on the jurisdiction in which the vehicle is located).

[0054] Highly autonomous vehicles (“HAVs”) are Level 3 or higher autonomous vehicles. Accordingly, in some configurations, the test autonomous vehicle 350 is one of the following: Level 1 autonomous vehicle; Level 2 autonomous vehicle; Level 3 autonomous vehicle; Level 4 autonomous vehicle; Level 5 autonomous vehicle; and HAV.

[0055] The vehicle perception module 310 can communicate with the sensor module 302, processor 320, computer-readable medium 322, communication module 324, vehicle unit 326, position module 328, mobility module 329, planner module 330, controller module 340, and transceiver 342. In one configuration, the vehicle perception module 310 receives sensor data from the sensor module 302. The sensor module 302 can receive sensor data from a first sensor 306 and a second sensor 304. According to various aspects of this disclosure, the sensor module 302 can filter data to remove noise, encode data, decode data, merge data, extract frames, or perform other functions. In an alternative configuration, the vehicle perception module 310 can receive sensor data directly from the first sensor 306 and the second sensor 304.

[0056] like Figure 3As shown, the vehicle perception module 310 includes a test vehicle application module 312, a driving log telemetry module 314, and a session data storage module 316. The test vehicle application module 312 and the driving log telemetry module 314 can be components of the same or different artificial neural networks (such as deep convolutional neural networks (CNN)). The vehicle perception module 310 is not limited to CNN. The vehicle perception module 310 receives data streams from a first sensor 306 and / or a second sensor 304. The data streams may include 2D RGB images from the first sensor 306 and LIDAR data points from the second sensor 304. The data streams may include multiple frames, such as image frames of a scene.

[0057] This configuration of the vehicle perception module 310 includes a test vehicle application module 312 (e.g., a driving stack) for operating the test autonomous vehicle 350 during a driving session. The driving session is cataloged along with telemetry information stored by the driving log telemetry module 314 and sensor data from the sensor module 302. The telemetry information and sensor data are stored in a removable session data storage module 316. In this example, the test agent control system 301 and the test autonomous vehicle 350 are associated with a site that includes a garage machine 370, a network-attached storage (NAS) 380, and a site virtual machine 390. Figure 4 The ingestion process is further illustrated.

[0058] Figure 4 This diagram illustrates an ingestion process 400 according to various aspects of the present disclosure, which implements a system and method for monitoring the status of test data at various memory storage locations and running tasks on the test data. In this example, the ingestion process 400 begins when the test autonomous vehicle 350 completes a test run (“session”) and enters a garage. In this example, the session data storage module 316 is removed from the test autonomous vehicle 350 and inserted into a computer (e.g., garage machine 370).

[0059] In this example, the session data storage module 316 is inserted into the garage machine 370 to trigger the upload of session data from the garage machine 370 to the site NAS 380 at origin site 402. In one configuration, the upload of session data to the site NAS 380 triggers a message on a session data queue monitored by a site virtual machine 390. In response, the site virtual machine 390 uploads the session data from origin site 402 (e.g., a driving site) to a cloud-based storage location 430. Furthermore, session data is removed from origin site 402 (e.g., garage machine 370 and / or site NAS 380).

[0060] Raw session data can be stored at cloud-based storage location 430. Furthermore, any irrelevant information can be removed from the session data; however, the complete, unfiltered version of the session data remains at cloud-based storage location 430. The filtered information can be provided to data pipeline 440. Ingestion processing 400 serves as a distributed computing network for distributing the filtered information to research sites (e.g., site 1, site 2, etc.). This aspect of the disclosure focuses on the ability to collect and rapidly distribute autonomous vehicle test data via a distributed computing network, making it easily accessible to researchers at other sites (e.g., site 1, site 2, etc.).

[0061] Ingestion processing 400 provides test data distribution to researchers at research sites for rapid use, distributing relevant data from the test autonomous vehicle to them. For example, test data distribution to a first research site 410 enables researchers (or research sites) to receive session data and processing units. Similarly, in response to ingestion processing 400, test data and processing unit distribution to a second research site 420 provides researchers with test information and processing units that contribute processing power to the session data processing.

[0062] Processing (e.g., work) units can correspond to specific analysis or processing tasks. For example, a first research site 410 could be responsible for indexing session data to make it compatible with keyword searches in a network-attached memory of a second research site 420. Thus, the assigned work unit could be processing session data for indexing. The second research site 420 could be responsible for performing machine learning tasks on the session data. Once a research site completes its work unit, the output is then pushed to a cloud-based storage location 430 and other research locations, such as a data pipeline 440, where the output can be used.

[0063] Figure 5 The figure illustrates sensor data images 500 captured by a test autonomous vehicle 350 operating based on a driving stack (e.g., test vehicle application module 312) according to various aspects of this disclosure. To illustrate the operation of the test autonomous vehicle 350, a simple scenario is considered, illustrated by sensor data images 500 of the test autonomous vehicle 350 driving on a straight section 502 of a highway.

[0064] In this example, the vehicle perception module 310 (e.g., using LIDAR) determines that the only obstacle in front of the test autonomous vehicle 350 is the leading vehicle 510 traveling in the same direction at a distance of 100 meters. The test autonomous vehicle 350 is approaching the leading vehicle 510 at a speed of 10 meters per second (m / s) and is capable of traveling at 5 meters per second squared (m / s). 2The vehicle decelerates at a rate of [missing information]. This means that if the stopping distance is 10 meters, and the speed is doubled, the stopping distance will increase to 40 meters. The controller module 340 can propose a vehicle control action to increase the vehicle's speed to 20 meters per second. [The text abruptly ends here, likely due to an incomplete sentence or missing information.] Figure 4 The acquisition processing shown involves storing the sensor data captured during this test driving session, along with telemetry information captured by the driving log telemetry module 314 (e.g., using an inertial measurement unit (IMU)), in the session data module for subsequent processing.

[0065] Figure 6 This is a flowchart illustrating a method for distributing and analyzing test data for autonomous vehicles according to various aspects of this disclosure. Figure 6 Method 600 begins at block 602, where driving session data is uploaded from the driving station's computer to a network-attached storage device at the driving station. For example, as... Figure 4 As described, inserting the session data storage module 316 into the garage machine 370 triggers the upload of session data from the garage machine 370 to the site NAS 380 at the origin site 402. In block 604, driving session data is uploaded from the driving site's network-attached storage to a cloud-based storage location. For example, as... Figure 4 As shown, uploading session data to site NAS 380 triggers a message on the session data queue, which is monitored by site virtual machine 390. In response, site virtual machine 390 uploads session data from origin site 402 (e.g., driving site) to cloud-based storage location 430. Furthermore, the session data is removed from origin site 402 (e.g., garage machine 370 and / or site NAS 380).

[0066] Refer again Figure 6 In block 606, driving session data and work units from cloud-based storage locations are distributed to at least one research site separate from the driving station. For example, as Figure 4 As shown, the ingestion process 400 provides test data distribution, distributing relevant data from the test autonomous vehicle to researchers at their locations for rapid use. For example, test data distribution to a first research site 410 enables researchers (or the research site) to receive session data and processing units. Similarly, in response to the ingestion process 400, test data and processing unit distribution to a second research site 420 provides researchers with test information and processing units that contribute processing capabilities to the processing of session data.

[0067] In block 608, driving session data is processed by at least one research station according to analysis / processing tasks associated with the work unit. For example, such as... Figure 4As shown, processing (e.g., work) units can correspond to specific analysis or processing tasks. For example, a first research station 410 could be responsible for indexing session data to make it compatible with keyword searches in a network-attached memory of a second research station 420. Thus, the assigned work unit could be processing session data to index it. The second research station 420 could be responsible for performing machine learning tasks on the session data. Once a research station completes its work unit, the output is then pushed to a cloud-based storage location 430 and other research locations, such as a data pipeline 440, where the output can be used.

[0068] Method 600 further includes uploading the processed session data from at least one research site to a cloud-based storage location. Method 600 also includes distributing the processed session data from the cloud-based storage location to at least another research site. Method 600 further includes completing a driving session for the test autonomous vehicle. Method 600 further includes removing the session data storage module from the test autonomous vehicle. Method 600 further includes storing the driving session data from the session data storage module to a computer at the driving site.

[0069] The method also includes monitoring message queues associated with network-attached storage at the driving station by a virtual machine at the driving station. Method 600 further includes uploading driving session data from network-attached storage at the driving station to a cloud-based storage location by the virtual machine at the driving station in response to the message queues. In method 600, a work unit may include performing machine learning on sensor data within the driving session data. Additionally, a work unit may include performing filtering operations on the driving session data.

[0070] Various aspects of this disclosure relate to systems and methods for monitoring the status of test data at various memory storage locations and for performing tasks on the test data. These aspects also provide test data pipelines for distributing relevant data from test autonomous vehicles to researcher locations for rapid use by researchers.

[0071] The various operations described above can be performed by any suitable means capable of performing the corresponding functions. Means may include various hardware and / or software components and / or modules, including but not limited to circuits, application-specific integrated circuits (ASICs), or processors. Generally, where operations are illustrated in the figures, those operations may have equivalent means with similar numbering plus functional components.

[0072] As used herein, the term "determine" encompasses a wide variety of actions. For example, "determine" can include calculation, operation, processing, deduction, investigation, search (e.g., searching in a table, database, or other data structure), ascertainment, etc. Furthermore, "determine" can include receiving (e.g., receiving information), accessing (e.g., accessing data in memory), etc. Moreover, "determine" can include resolving, selecting, choosing, establishing, etc.

[0073] As used in this article, the phrase “at least one” in the list of items refers to any combination of those items, including a single member. As an example, “at least one of a, b, or c” is intended to cover: a, b, c, ab, ac, bc, and abc.

[0074] The various illustrative logic blocks, modules, and circuits described in conjunction with this disclosure may be implemented or performed using a processor, digital signal processor (DSP), ASIC, field-programmable gate array (FPGA) or other programmable logic device (PLD), discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. The processor may be a microprocessor, but alternatively, it may be any commercially available processor, controller, microcontroller, or state machine specifically configured as described herein. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP core, or any other such configuration.

[0075] The steps of the methods or algorithms described in this disclosure can be implemented directly in hardware, as a software module executed by a processor, or a combination of both. The software module can reside in any form of storage medium known in the art. Some examples of usable storage media include random access memory (RAM), read-only memory (ROM), flash memory, erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), registers, hard disks, removable disks, CD-ROMs, etc. The software module may include one or more instructions and may be distributed across several different code segments, across different programs, and across multiple storage media. The storage medium can be coupled to the processor so that the processor can read information from and write information to the storage medium. Alternatively, the storage medium can be integrated with the processor.

[0076] The methods disclosed herein include one or more steps or actions for implementing the described methods. The method steps and / or actions may be interchanged with each other without departing from the scope of the claims. In other words, unless a specific order of steps or actions is specified, the order and / or use of particular steps and / or actions may be modified without departing from the scope of the claims.

[0077] The described functionality can be implemented in hardware, software, firmware, or any combination thereof. If implemented in hardware, an example hardware configuration may include a processing system within the device. The processing system may be implemented using a bus architecture. Depending on the specific application and overall design constraints of the processing system, the bus may include any number of interconnect buses and bridges. The bus can link various circuits, including processors, machine-readable media, and bus interfaces. The bus interface can connect, for example, a network adapter to the processing system via the bus. The network adapter can implement signal processing functions. In some respects, user interfaces (e.g., keyboards, displays, mice, joysticks, etc.) may also be connected to the bus. The bus can also connect various other circuits, such as timing sources, peripheral devices, voltage regulators, power management circuits, etc., which are well known in the art and will not be described further.

[0078] The processor can be responsible for managing the bus and processing, including executing software stored on a machine-readable medium. Examples of processors particularly configurable according to this disclosure include microprocessors, microcontrollers, DSP processors, and other circuits that can execute software. Software should be interpreted broadly as referring to instructions, data, or any combination thereof, whether referred to as software, firmware, middleware, microcode, hardware description language, or others. For example, a machine-readable medium may include RAM, flash memory, ROM, programmable read-only memory (PROM), EPROM, EEPROM, registers, disk, optical disk, hard disk drive, or any other suitable storage medium or any combination thereof. The machine-readable medium may be implemented in a computer program product. The computer program product may include packaging materials.

[0079] In hardware implementations, machine-readable media can be part of a processing system separate from the processor. However, as those skilled in the art will readily understand, machine-readable media, or any portion thereof, can be external to the processing system. For example, machine-readable media may include transmission lines, carrier waves modulated by data, and / or device-separate computer products, all of which can be accessed by the processor via a bus interface. Alternatively, or additionally, machine-readable media, or any portion thereof, may be integrated into the processor, such as in the case of a cache and / or a file of special-purpose registers. Although the various components discussed can be described as having a specific location, such as local components, they can also be configured in various ways, such as certain components being configured as part of a distributed computing system.

[0080] The processing system may be configured with one or more microprocessors providing processor functionality and external memory providing at least a portion of machine-readable medium, all linked together with other supporting circuitry via an external bus architecture. Alternatively, the processing system may include one or more neuromorphic processors for implementing the neuron and nervous system models described herein. As another alternative, the processing system may be implemented using an ASIC that integrates the processor, bus interface, user interface, supporting circuitry, and at least a portion of the machine-readable medium onto a single chip, or using one or more FPGAs, PLDs, controllers, state machines, gating logic, discrete hardware components, or any other suitable circuitry, or any combination of circuitry capable of performing the various functions described in this disclosure. Those skilled in the art will recognize that the best implementation of the described functions of the processing system depends on the specific application and the overall design constraints imposed on the system as a whole.

[0081] Machine-readable media may include several software modules. Software modules include instructions that, when executed by a processor, cause the processing system to perform various functions. Software modules may include transfer modules and receive modules. Each software module may reside in a single storage device or be distributed across multiple storage devices. For example, a software module may be loaded from a hard disk drive into RAM when a triggering event occurs. During the execution of a software module, the processor may load some instructions into a cache to improve access speed. Then, one or more cache lines may be loaded into a special-purpose register file for execution by the processor. When referring to the functionality of a software module below, it is understood that this functionality is implemented by the processor when executing the instructions of that software module. Furthermore, it should be understood that aspects of this disclosure improve the functionality of processors, computers, machines, or other systems that implement these aspects.

[0082] If implemented in software, functionality can be stored or transmitted as one or more instructions or code on a non-transitory computer-readable medium. Computer-readable media include computer storage media and communication media, both of which include any medium that facilitates the transfer of a computer program from one place to another. Storage media can be any available medium that can be accessed by a computer. For example, and not limitingly, such computer-readable media can include RAM, ROM, EEPROM, CD-ROM or other optical disc storage, disk storage or other magnetic storage devices, or any other medium that can be used to carry or store the required program code in the form of instructions or data structures and that can be accessed by a computer. Furthermore, any connection is appropriately referred to as a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared (IR), radio, and microwave, then coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of medium. The terms "disk" and "optical disk" as used herein include optical discs (CDs), laser discs, optical discs, digital versatile optical discs (DVDs), floppy disks, and Blu-ray discs, wherein disks typically reproduce data magnetically, while optical discs reproduce data optically using lasers. Therefore, in some aspects, computer-readable media may include non-transitory computer-readable media (e.g., tangible media). Furthermore, in other aspects, computer-readable media may include transient computer-readable media (e.g., signals). Combinations of the above should also be included within the scope of computer-readable media.

[0083] Therefore, certain aspects may include a computer program product for performing the operations described herein. For example, such a computer program product may include a computer-readable medium having instructions stored thereon (and / or encoded thereon) that can be executed by one or more processors to perform the operations described herein. In some aspects, the computer program product may include packaging material.

[0084] Furthermore, it should be understood that modules and / or other suitable components for performing the methods and techniques described herein can be downloaded and / or otherwise obtained by the user terminal and / or base station where applicable. For example, such a device can be coupled to a server to facilitate the delivery of components for performing the methods described herein. Alternatively, the various methods described herein can be provided via storage components (e.g., RAM, ROM, physical storage media such as CDs or floppy disks, etc.), so that the user terminal and / or base station can obtain the various methods when the storage components are coupled to or provided to the device. Furthermore, any other suitable techniques can be used to provide the methods and techniques described herein to the device.

[0085] It should be understood that the claims are not limited to the precise configuration and components described above. Various modifications, alterations, and variations may be made to the arrangement, operation, and details of the methods and apparatus described above without departing from the scope of the claims.

Claims

1. A method for distributing and analyzing test data for autonomous vehicles, comprising: Upload driving session data from the computer at the driving station to the network-attached storage device at the driving station; Upload driving session data from the network-attached storage at the driving station to a cloud-based storage location; Distribute driving session data and work units from cloud-based storage locations to at least one research site that is separate from the driving site; as well as The driving session data is processed by the at least one research station according to the analysis / processing tasks associated with the work unit. The driving session data includes: telemetry information measured during the test run of the autonomous vehicle and sensor data captured during the test run of the autonomous vehicle.

2. The method according to claim 1, further comprising: The processed session data is uploaded from the at least one research site to a cloud-based storage location; as well as The processed session data will be distributed from the cloud-based storage location to at least another research site.

3. The method according to claim 1, further comprising: Complete the driving session for the test autonomous vehicle; Remove the session data storage module from the test autonomous vehicle; as well as The driving session data from the session data storage module is stored in the computer at the driving station.

4. The method of claim 1, wherein, Driving session data uploaded from a storage device attached to the network includes: The message queues associated with the network-attached memory of the driving station are monitored by the virtual machine at the driving station; and The virtual machine at the driving station responds to the message queue to upload driving session data from the driving station's network-attached storage to a cloud-based storage location.

5. The method of claim 4, further comprising: The virtual machine at the driving station deletes driving session data from the computer at the driving station and / or the storage attached to the network.

6. The method according to claim 1, wherein, The work unit includes performing machine learning on sensor data from driving session data.

7. The method according to claim 1, wherein, The work unit includes filtering operations on driving session data.

8. The method according to claim 1, wherein, The raw sensor data from driving sensor data in the driving session data is stored in a cloud-based storage location, and any irrelevant information is removed from the raw sensor data at the at least one research site.

9. A non-transitory computer-readable medium storing program code for the distribution and analysis of autonomous vehicle test data, the program code being executed by a processor and comprising: Program code used to upload driving session data from the computer at the driving station to the network-attached memory at the driving station; Program code used to upload driving session data from a network-attached storage device at a driving station to a cloud-based storage location; Program code for distributing driving session data and work units from cloud-based storage locations to at least one research site separate from the driving site; as well as Program code for processing driving session data by the at least one research station according to analysis / processing tasks associated with the work unit. The driving session data includes: telemetry information measured during the test run of the autonomous vehicle and sensor data captured during the test run of the autonomous vehicle.

10. The non-transitory computer-readable medium of claim 9, further comprising: Program code for uploading processed session data from the at least one research site to a cloud-based storage location; as well as Program code used to distribute processed session data from a cloud-based storage location to at least another research site.

11. The non-transitory computer-readable medium of claim 9, further comprising: Program code used to complete the driving session of the test autonomous vehicle; Program code used to remove the session data storage module from the test autonomous vehicle; as well as Program code used to store driving session data from the session data storage module to the computer at the driving station.

12. The non-transitory computer-readable medium according to claim 9, wherein, The program code used to upload driving session data from a storage device attached to the network includes: Program code for monitoring message queues associated with network-attached memory at the driving station by a virtual machine; and Program code used by a virtual machine at a driving station to upload driving session data from a network-attached storage device at the driving station to a cloud-based storage location in response to a message queue.

13. The non-transitory computer-readable medium of claim 12, further comprising: Program code for deleting driving session data from the driving station's computer and / or storage attached to the network by the virtual machine at the driving station.

14. A system for distributing and analyzing test data of autonomous vehicles, the system comprising: A driving station includes a network-attached storage device and a computer configured to upload driving session data from the computer at the driving station to the network-attached storage device in response to the insertion of a session data storage module of a test vehicle. The network-attached storage device is configured to upload driving session data from the network-attached storage device at the driving station to a cloud-based storage location. Cloud-based storage locations are configured to distribute driving session data and work units to multiple research sites that are separate from driving stations. as well as At least two of the plurality of research sites are configured to process driving session data according to analysis / processing tasks associated with work units received by the two research sites. The work unit includes performing machine learning on sensor data in driving session data.

15. The system according to claim 14, wherein, The work unit includes filtering operations on driving session data.

16. The system according to claim 14, wherein, The raw sensor data from driving sensor data in the driving session data is stored in a cloud-based storage location, and any irrelevant information is removed from the raw sensor data at the at least one research site.

Images