Systems and devices for on-demand feature set data collection in autonomous vehicles and autonomous driving systems

Abstract

A data collection system in an on-demand feature set of a motor vehicle includes at least one data collection device equipped with the motor vehicle. In response to a data collection request submitted by a user, a data collection unit receives data from the at least one data collection device. A data scenario group combines portions of the data received by the data collection unit and stored in the data collection unit of the motor vehicle or at a remote server. A control logic device receives the data collection request and activates the collection of the data. The control logic device includes a data collection strategy, if different types of data are available, a reference algorithm loaded or deployed, a storage capacity allocated for the data, and a storage strategy for saving the data.

Description

Technical Field

[0001] This disclosure relates to the collection of real-world data in autonomous vehicles. Background Technology

[0002] Motor vehicles, including autonomous vehicles, and systems including autonomous driving systems require a large amount of data to support normal operation. Data is essential for various scenarios that are difficult or impossible to create or are known during the vehicle design phase. These scenarios include operation under various environmental conditions, such as rainy weather, snowy weather, icy conditions, strong winds, night driving, and urban or suburban driving. Therefore, such vehicles and systems need to collect real-world data for operation.

[0003] Vehicles can be equipped with devices to collect real-world data, such as radar sensors, laser imaging detection and imaging (lidar) sensors, and camera systems. These devices and systems, collectively referred to below as sensors, continuously receive and generate large amounts of data. During a specific operation of a motor vehicle, only a portion of the collected sensor data may be needed. For example, when operating during a rural driving event, fewer other motor vehicles and fewer targets, such as traffic lights, road signs, and road intersections, are expected. Therefore, the data available for and related to the operation of the motor vehicle can be more closely correlated with real-time conditions, such as road conditions, available lights, temperature, etc. When accessing a large database of collected data to assist driving during these conditions, and when all sensor data collected in real time must be sorted, much of the data may be irrelevant, and some of the currently collected data may be redundant because sensors continuously receive inherently repetitive data. Therefore, continuously evaluating all collected sensor data (including all sensor data collected in real time) is an unnecessary computational burden.

[0004] Therefore, while current autonomous vehicle data collection systems have achieved their intended purpose, a new and improved system is still needed to collect, prioritize, and use real-time data for autonomous vehicle operation. Summary of the Invention

[0005] According to several aspects, a data collection system for on-demand feature sets of motor vehicles includes at least one data collection device equipped with the motor vehicle. In response to a data collection request submitted by a user, a data collection unit receives data from at least one data collection device. A data scenario group combines portions of the data received by the data collection unit and the data stored in the data collection unit of the motor vehicle or at a remote server. A control logic device receives the data collection request and activates data collection. The control logic device includes: a data collection strategy, which may differ if different types of data are available; a reference algorithm that is loaded or deployed; storage capacity allocated for the data; and a storage strategy for storing the data.

[0006] In another aspect of this disclosure, the storage capacity, including the size, is designed to store the data with lock-free and wait-free storage management features.

[0007] In another aspect of the invention, the lock-free, wait-free storage management feature is configured as a modified ping-pong buffer mechanism comprising two caches defining a first cache and a second cache, the first cache and the second cache having the same size, wherein at any time, one of the first cache or the second cache is marked for read functionality, and the other of the first cache or the second cache is marked for write functionality.

[0008] In another aspect of this disclosure, when predetermined data conditions are met, the control logic device triggers the recorder device.

[0009] In another aspect of this disclosure, the storage capacity includes at least one of a local storage device and a remote storage device located in a remote server; wherein the recorder device is responsible for writing data to the local storage device or the remote storage device and for managing the storage of the data.

[0010] In another aspect of this disclosure, the data collection request definition indicates what data to collect and the predetermined format of the collection level.

[0011] In another aspect of this disclosure, data from at least one data collection device is collected to the target feature "on demand" or according to a request from a system user, including occupants of a motor vehicle, using a strategy that includes activation conditions, data selection, collection duration, and data storage.

[0012] In another aspect of this disclosure, the data scenario group includes multiple data scenarios, some or all of which are retrieved from a remote server on demand and returned to the data collection unit via a first remote request. The first remote request is forwarded to the remote server via a wireless device, and the wireless device transmits a second remote request to the remote server.

[0013] In another aspect of this disclosure, the data defines at least one of the following scenarios: snowfall event scenario, nighttime driving event scenario, rural driving event scenario, urban driving event scenario, vehicle accident event scenario, residential street scenario, and highway scenario.

[0014] In another aspect of this disclosure, the at least one data collection device includes at least one of a forward-facing camera, a rear-facing camera, and a radar or laser imaging detection and imaging (lidar) device.

[0015] According to several aspects, a data collection system for on-demand feature sets of motor vehicles includes at least one data collection device equipped with an autonomous motor vehicle. In response to a data collection request submitted by a user, a data collection unit receives data from at least one data collection device. A control logic device receives the data collection request. An actuator of the control logic device executes a reference algorithm to generate a reference output. A comparator of the control logic device compares the reference output with data from at least one data collection device. A recorder device of the control logic device is triggered when predetermined data conditions are met, and data collection is activated. The size of a lock-free, wait-free storage management feature is designed to store the data.

[0016] In another aspect of this disclosure, the control logic device includes a data collection strategy that differs depending on the type of data available.

[0017] In another aspect of this disclosure, a signal generated by the control logic device activates data collection according to a data collection strategy.

[0018] In another aspect of this disclosure, a storage strategy employing storage management features is used to preserve data.

[0019] In another aspect of this disclosure, the reference algorithm includes object detection features for real-time detection and recognition of objects in a scene.

[0020] In another aspect of this disclosure, the reference algorithm defines an application that can be deployed independently.

[0021] In another aspect of this disclosure, the reference algorithm has a unique identifier and a structure including the inputs required for the unique identifier, the type and size of the data, the output type and output size, and an execution environment that defines the type and frequency of the processor for the service, available memory, and control logic devices. When the inputs and execution environment are loaded, the executor executes the reference algorithm.

[0022] According to several aspects, a method for performing on-demand feature set data collection for a motor vehicle includes: providing at least one data collection device equipped with the motor vehicle; receiving data from the at least one data collection device in a data collection unit in response to a data collection request submitted by a user; combining portions of the data received by the data collection unit into a data scene group, and storing the data scene group in one of the data collection units of the motor vehicle or at a remote server; transmitting the data collection request to a control logic device and activating the data collection.

[0023] In another aspect of this disclosure, the method further includes: employing a data collection strategy, wherein the data collection strategy differs if different types of data are available in the control logic device; and applying a reference algorithm to the data after employing the data collection strategy.

[0024] In another aspect of this disclosure, the method further includes: allocating storage space for the data; employing a storage strategy to store the data in the storage space; and retrieving data from the storage space on demand when a data collection request is received from a user.

[0025] The present invention may also include the following solutions.

[0026] 1. A data collection system for on-demand feature set of motor vehicles, comprising: Equipped with at least one data collection device for motor vehicles; A data collection unit that receives data from the at least one data collection device in response to a data collection request submitted by a user; A data scene group, which is a combination of portions of the data received by the data collection unit and stored in the data collection unit of the motor vehicle or at a remote server; A control logic device that receives the data collection request and activates the data collection, and has: Data collection strategies can vary depending on the type of data available. A reference algorithm that generates a reference output, which is compared with data from the at least one data collection device, and activates the collection of the data. Storage capacity allocated for data; and Storage strategy for storing the data.

[0027] 2. The system according to Scheme 1, wherein the storage capacity includes a size designed to store the data with lock-free and wait-free storage management features.

[0028] 3. The system according to Scheme 2, wherein the lock-free, wait-free storage management feature is configured as a modified ping-pong buffer mechanism, the ping-pong buffer mechanism comprising two caches, the two caches defining a first cache having a first cache size and a second cache having a second cache size having the same size as the first cache size, wherein at any time, one of the first cache or the second cache is marked for read functionality, and the other of the first cache or the second cache is marked for write functionality.

[0029] 4. The system according to Scheme 1 further includes a recorder device, which is triggered by the control logic device when predetermined data conditions are met.

[0030] 5. The system according to Scheme 4, wherein the storage capacity includes at least one of a local storage device and a remote storage device located in the remote server; wherein the recorder device is responsible for writing data to the local storage device or the remote storage device; and is responsible for managing the storage of the data.

[0031] 6. The system according to Scheme 1, wherein the data collection request is defined in a predetermined format, the predetermined format indicating which data in the data to be collected and the collection level.

[0032] 7. The system according to claim 1, wherein data received from the at least one data collection device is collected to target features on an "on-demand" basis or at the request of a user, the user including occupants of a motor vehicle, and the data collection is collected using one or more data collection strategies, the data collection strategies defining at least one of activation conditions, data selection, collection duration, and data storage.

[0033] 8. The system according to Scheme 1, wherein the data scenario group includes multiple data scenarios, some or all of which are retrieved from the remote server on an "on-demand" basis and returned to the data collection unit via a first remote request, the first remote request being forwarded to the remote server via a wireless device, the wireless device also transmitting a second remote request to the remote server.

[0034] 9. The system according to Scheme 1, wherein the data defines at least one of a snow event scenario, a night driving event scenario, a rural driving event scenario, an urban driving event scenario, a vehicle accident event scenario, a residential street scenario, and a highway scenario.

[0035] 10. The system according to claim 1, wherein the at least one data collection device includes at least one of a forward-facing camera, a rear-facing camera, and a radar or laser imaging detection and imaging (lidar) device.

[0036] 11. A data collection system for on-demand feature sets of motor vehicles, comprising: Equipped with at least one data collection device for autonomous motor vehicles; A data collection unit that receives data collected by the at least one data collection device in response to a data collection request submitted by a user; A control logic device that receives the data collection request; The actuator of the control logic device executes a reference algorithm to generate a reference output; The comparator of the control logic device compares a reference output with data collected by the at least one data collection device. The control logic device has a recording device that is triggered and activates data collection when predetermined data conditions are met; and Lock-free and wait-free storage management features, the size of which is designed to store the data.

[0037] 12. The system according to claim 11, wherein the control logic device includes a data collection strategy, wherein the data collection strategy is different if different types of data collected by the at least one data collection device are available.

[0038] 13. The system according to claim 12 further includes a signal generated by the control logic device to activate data collection according to the data collection strategy.

[0039] 14. The system according to claim 11 further includes a storage strategy, which is used by the lock-free, wait-free storage management feature to store data collected by the at least one data collection device.

[0040] 15. The system according to Scheme 11, wherein the reference algorithm includes real-time detection and identification of object detection features in the scene.

[0041] 16. The system according to Scheme 11, wherein the reference algorithm defines an application that can be deployed independently.

[0042] 17. The system according to claim 11, wherein the reference algorithm has a unique identifier and a structure, the structure including inputs required for the unique identifier, the type and size of data collected by at least one data collection device, the output type and output size, and an execution environment defining the type and frequency of processors for services, available memory, and control logic devices, wherein the executor executes the reference algorithm when the inputs and execution environment are loaded.

[0043] 18. A method for performing on-demand data collection from a feature set, comprising: Provide at least one data collection device equipped with a motor vehicle; In response to a user-submitted data collection request, data is received from at least one data collection device in the data collection unit; A portion of the data received by the data collection unit is combined as a data scenario group, and the data scenario group is stored in either the data collection unit of the motor vehicle or at a remote server. The data collection request is transmitted to the control logic device and the data collection is activated.

[0044] 19. The method according to Scheme 18 further includes: The data collection strategy varies depending on the type of data available in the control logic device; and The reference algorithm is applied to the data after the data collection strategy is deployed.

[0045] 20. The method according to Scheme 19 further includes: Allocate storage devices for the data; A storage strategy is employed to store data in the storage device; and Data is retrieved from the storage device as needed after a data collection request is received from the user.

[0046] Further areas of application will become apparent from the description provided herein. It should be understood that the descriptions and specific examples are intended for illustrative purposes only and are not intended to limit the scope of this disclosure. Attached Figure Description

[0047] The accompanying drawings described herein are for illustrative purposes only and are not intended to limit the scope of this disclosure in any way.

[0048] Figure 1 This is a system diagram of a data collection system based on an on-demand feature set of exemplary aspects; Figure 2 It is for identification and operation targeting Figure 1 A flowchart showing the controller's associated features and data collected by the system; Figure 3 It is used for activation Figure 1 The flowchart of the control logic system part of the system's data request; and Figure 4 yes Figure 1 A diagram illustrating the lock-free and wait-free memory management features of the system. Detailed Implementation

[0049] The following description is exemplary in nature and is not intended to limit this disclosure, application, or use.

[0050] refer to Figure 1 A data collection system 10 provides a feature set for motor vehicle 12 on demand, which may define an autonomous vehicle or an autonomous driving system for a vehicle. Motor vehicle 12 is equipped with at least one data collection device and, depending on various aspects, multiple data collection devices, which may include, but are not limited to, a forward-facing camera 14, a rear-facing camera 16, and a radar or laser imaging detection and imaging (lidar) device 18. Multiple data collection devices collect data from multiple sources. Data collection unit 20 receives data collected by the multiple data collection devices. For example, data collection unit 20 may receive data when an occupant 24 of motor vehicle 12, such as an operator or passenger, submits a data request 22.

[0051] The collected data can define one or more data scenarios, and therefore can differ between one or more data scenarios, including snow event scenario 26, night driving event scenario 28, rural driving event scenario 30, urban driving event scenario 32, and vehicle accident event scenario 34. Further examples of data scenarios can include non-motion scenarios, including but not limited to residential street scenario 36 and highway scenario 38. The collected data can be combined into a data scenario group 40, either entirely or as part of the collected data. The data scenario group can be stored in the data collection unit 20 of the motor vehicle 12, or it can be stored at a remote server 42 via a first remote request 44, which is forwarded to the remote server 42 via a wireless device 46 that transmits a second remote request 48 to the remote server 42. Some or all of the data scenarios in the data scenario group 40 can be retrieved from the remote server 42 on demand and returned to the data collection unit 20 via wireless transmission by reversing the second remote request 48 and the first remote request 44.

[0052] refer to Figure 2 And refer again Figure 1The on-demand feature set data collection system 10 can be implemented as a separate device such as a software component or a controller 50 (collectively referred to below as controller 50). Controller 50 can be deployed in an autonomous vehicle or a perception-enabled motor vehicle 12 equipped with an autonomous driving system. As described above, perception capabilities can include, but are not limited to, a forward-facing camera 14, a rear-facing camera 16, and a radar or lidar device 18 to collect real-world data on various scenarios that are difficult or impossible to create during the vehicle design phase. The on-demand feature set data collection system 10 allows for the collection of data related to target features only "on demand," defined as any time the system user wants or needs something, and whenever the system user requests it, the system user can include the occupants 24 of the motor vehicle 12. Requests can include various strategies, including activation conditions, data selection, collection duration, and data storage.

[0053] The architecture of controller 50 for on-demand, feature-oriented data collection acquires: sensor data from necessary data collection devices and other vehicle sensors in data scene group 40; output from the feature of interest 52 to be collected; and data collection requests from local or remote users. Data collection requests, such as data request 22, first remote request 44, or second remote request 48, have a predetermined format indicating what data to collect and at what level. The collection level can be defined, for example, collection at a specific scene, or collection when the feature of interest 52 produces an output different from a predetermined or reference algorithm 54.

[0054] The control logic unit 56 of the controller 50 receives data request 22, identifies potentially different data collection strategies 58 among different types of available data, or deploys the reference algorithm 54 based on whether it already exists locally or needs to be retrieved from a remote location, such as from a remote server 42, allocates sufficient storage capacity for data collection employing storage strategy 60, and initiates data collection. The control logic unit 56 continuously monitors collection conditions and triggers data recording using the recorder unit 62 when predetermined data conditions are met. The recorder unit 62 is responsible for writing data to local storage device 64 or to remote storage device 66, which may be located on the remote server 42, and for managing data storage. The invocation and termination of the data collection service are controlled by requests from higher system-level devices 68.

[0055] In the data collection system 10 with an on-demand feature set, multiple data collection services may exist to collect data for different features. Each data collection service may also be defined to collect data for multiple features. Controller 50 may include a reference algorithm 54, a data collection strategy 58, and a storage strategy 60, or these elements may be provided in one or more other controllers.

[0056] Based on several aspects, an application mechanism is used to define customizable feature-specific data collection strategies. This mechanism consists of two parts: 1) a data structure collection request, for example exemplified by data request 22; and 2) a locally stored collection strategy, such as data collection strategy 58. The data collection system 10, applied according to the design principles of the on-demand feature set, uses minimal data carrying most of the information.

[0057] Data structure collection requests include the following: Request =<req-id, strgy-id, f-list> req-id: Indicates the request, where: Potentially, a service / device may collect data for multiple characteristics.

[0058] strgy-id: The configuration of the data collection service corresponding to the entry defined in data collection strategy 58.

[0059] f-list: A list of features used to collect data.

[0060] According to an example: r = <10, {diff, (v-spd, v-loc, f-image, o-list), ref-algo}, {sbda, lca}, local>, the request will collect data on the features of both side blind spot warning and lane change warning when both are activated, collect data when a difference occurs between the feature output and the reference algorithm 54 ref-algo, and store the data on local memory.

[0061] Data collection strategy 58: Define what data to collect, when to collect data, and how to collect data.

[0062] Define a set of strategies for a feature or a group of features, which are predetermined during vehicle design. This set of strategies can be stored locally using a data collection service and can be updated in a specified state, such as when vehicle 12 is parked. Example features of the collection strategies are defined in a multi-table format, as shown in Table 1 below.

[0063] Table 1: .

[0064] Reference Figure 3 And refer again Figure 1 and 2 Data collection activation is based on activation policy 70. Activation policy 70 is referenced... Figure 2 Implemented in the control logic device 56 shown and described, the control logic device sends an activation signal 72 to trigger data collection. The control logic device 56 executes a reference algorithm 54, which is either built into the control logic device 56 or loaded on demand; it is either locally stored or loaded and subsequently removed. The reference algorithm 54 is executed by an actuator 74 in the control logic device 56 and produces a reference output 76. A comparator device 78 compares the reference output 76 with an on-board system output 80 generated by the on-board system, such as one of the scenarios in data scenario group 40, for example, interactions during a motorcycle, a weather event such as a snow event scenario 26, a rain event scenario, or a night driving event scenario 28, and generates an activation signal 72 to activate and deactivate data collection according to a data collection strategy 58.

[0065] According to Example 1: Reference Algorithm 54 is defined as a nighttime scenario, such as nighttime driving event scenario 28, where data collection strategy 58 defines a scenario where data is collected every 10 seconds using raw image velocity and position detection. When the image's onboard system output 80 is nighttime and the timer output used for collection reaches 10 seconds, an activation signal 72 is generated to trigger the data collection operation.

[0066] According to Example 2: Reference Algorithm 54 includes features of a one-look-only (YOLO) object detection algorithm for real-time detection and recognition of various objects in a scene. Object detection in YOLO is performed as a regression analysis problem, where all objects are detected in a single algorithm run, providing class probabilities for the detected scene image. Data collection strategy 58 compares different raw image output locations on the vehicle. When Reference Algorithm 54 produces detection results different from those of the vehicle system output 80, an activation signal 72 is generated to trigger data collection.

[0067] The deployment of reference algorithm 54 is a microservice. Each reference algorithm 54 is implemented as a separately deployable application, wherein any reference algorithm 54 is loaded by executor 74 when the data collection service begins. Executor 74 in control logic device 56 is responsible for loading the reference algorithm 54 and ensuring its proper configuration. For each reference algorithm 54, its implementation is serialized, stored locally, or received from a remote server such as remote server 42. Reference algorithm 54 is uniquely identified by algo_id (as shown in Table 2 below) and follows a uniform structure. The structure includes: all inputs required by algo_id and their data types and sizes, all outputs produced and their types and sizes, and the required execution environment, such as infrastructure services, available memory, processor type and frequency, etc. When all inputs and environment are ready, executor 74 invokes the executable program of reference algorithm 54.

[0068] Table 2: .

[0069] refer to Figure 4 And refer again Figures 1 to 3 For reference Figure 1 The data collected by the defined data scenario group 40 must be consistent and clear enough to reflect real-world conditions. Data is initially stored in one or more caches in an onboard memory such as local storage device 64, and then written in blocks to permanent memory such as remote storage device 66. All data is appended with a timestamp from a global clock, ensuring the entire system sees the same time at any given moment. To manage the data scenario group 40, the data collection system 10, based on the on-demand feature set, may include a lock-free, wait-free storage management feature 82. Depending on several aspects, the lock-free, wait-free storage management feature 82 can be configured as a ping-pong buffer mechanism 84 for modifying lock-free and wait-free data recordings, as this approach improves recording while preventing data corruption.

[0070] The modified ping-pong buffer mechanism 84 includes two buffers, defined as a first buffer 86 having a first buffer size and a second buffer 88 having a second buffer size of the same as the first buffer size, and the size of each of the first buffer 86 and the second buffer 88 is designed to store data for all received or required signals, defined as described above. Figure 3In the collection strategy discussed, at any given time, such as in the first operating configuration 84a, there exists a cache, for example, a first cache 86 marked for read function 90, and another cache, for example, a second cache 88 marked for write function 92. A recorder, such as recorder device 62, stores the collected data in the operable read cache and writes the data in parallel to the operable write cache, to local storage device 64, or to permanent storage such as remote storage device 66. When collecting new data in the read cache or dumping data into the write cache, whichever is more recently completed, the markings for reading and writing are switched. At the time of switching, the second operating configuration 84b provides the opposite cache in the cache, for example, the second cache 88 marked for read function 94, and another cache, for example, the first cache 86 marked for write function 96.

[0071] The on-demand feature set data collection system 10 disclosed herein allows for flexible configuration of the data collection system to collect data on any request and in different scenarios. The on-demand feature set data collection system 10 can be retargeted to collect different types of data for different features without rebuilding and refreshing the controller software.

[0072] Reference Figure 3 The described controller 50 is a non-general-purpose electronic control device that includes a pre-programmed digital computer or processor, memory or non-transitory computer-readable medium for storing data such as control logic, software applications, instructions, computer code, data, lookup tables, etc., and a transceiver or input / output port. Computer-readable medium includes any type of media accessible by a computer, such as read-only memory (ROM), random access memory (RAM), hard disk drive, compact disc (CD), digital video disc (DVD), or any other type of memory. Non-transitory computer-readable medium excludes wired, wireless, optical, or other communication links that transmit transient electrical signals or other signals. Non-transitory computer-readable medium includes media in which data can be permanently stored and media in which data can be stored and later rewritten, such as rewritable optical discs or erasable memory devices. Computer code includes any type of program code, including source code, object code, and executable code.

[0073] The data collection system 10 for on-demand feature sets disclosed herein can be implemented as a standalone device deployed to autonomous vehicles (AVs) or advanced driver assistance systems (ADAS) equipped with sensing capabilities to collect real-world data for various scenarios that are difficult or impossible to create (and sometimes even known) during the vehicle's design time. The system of this disclosure allows for the on-demand collection of data relevant only to the target features using various strategies, including activation conditions, data selection, collection duration, and data storage.

[0074] The data collection system 10 for on-demand feature sets disclosed herein offers several advantages. These include a data collection system that includes on-demand data collection triggered by several customizable collection strategies. At least one algorithmic decision is used to enable the collection of only relevant and meaningful data for one or a set of target features. A microservices-based architecture enables the on-demand deployment and activation of strategies and algorithms for the data. Continuous data collection is supported by lock-free and wait-free storage management.

[0075] The description in this disclosure is merely exemplary in nature, and changes that do not depart from the spirit and scope of this disclosure are intended to be within its scope. Such changes should not be considered as a departure from the spirit and scope of this disclosure.

Claims

1. A data collection system for on-demand feature sets of autonomous vehicles, comprising: The autonomous vehicle is equipped with at least one data collection device, which collects data defining multiple different data scenarios around the autonomous vehicle. and A controller, which communicates with the at least one data collection device, has a processor and a memory, the memory including instructions that program the processor to: The data collection unit receives data collected by the at least one data collection device. A portion of the data received by the data collection unit is combined into a data scene group, and the data scene group is stored in the memory; Receive a user-submitted data collection request related to a feature, wherein the data collection request includes a predetermined format, including a request identifier, a feature identifier associated with the feature, and a data collection strategy identifier; Based on the data collection policy identifier of the data collection request, a data collection policy is determined. This policy defines the data collection trigger event, data collection type, data collection duration, and a reference algorithm. The reference algorithm defines how the data is collected, including the inputs required by the algorithm, the data type of the inputs, the size of the inputs, the outputs generated by the algorithm, the output type, the output size, and the execution environment required to execute the algorithm. The execution environment includes infrastructure services, available memory, and the type of memory. By executing the data collection strategy, data from the feature set is collected from the data scenario group, wherein executing the data collection strategy includes: The reference algorithm is executed based on the data collection trigger event. Using the reference algorithm, data in the feature set is collected based on the data collection type; and Data in the feature set is collected during the data collection duration; and The autonomous vehicle operates based at least in part on data from the feature set; The memory includes storage capacity allocated for the data in the feature set, and a storage strategy for storing the data in the feature set.

2. The system according to claim 1, wherein, The storage capacity includes lock-free, wait-free storage management features designed to store data in the feature set.

3. The system according to claim 2, wherein, The lock-free, wait-free storage management feature is configured as a modified ping-pong buffer mechanism comprising two caches, the two caches defining a first cache having a first cache size and a second cache having a second cache size having the same second cache size as the first cache, wherein at any time, one of the first cache or the second cache is marked for read functionality, and the other of the first cache or the second cache is marked for write functionality.

4. The system according to claim 1 further includes a recorder device, which is triggered when predetermined data conditions are met.

5. The system according to claim 1, wherein, The data scenario group includes multiple data scenarios, which are retrieved from a remote server on an "on-demand" basis and returned to the controller via a first remote request. The first remote request is forwarded to the remote server via a wireless device, which also transmits a second remote request to the remote server.

6. The system according to claim 1, wherein, The data defines at least one of the following scenarios: snowfall event scenario, nighttime driving event scenario, rural driving event scenario, urban driving event scenario, vehicle accident event scenario, residential street scenario, and highway scenario.

7. The system according to claim 1, wherein, The at least one data collection device includes at least one of a forward-facing camera, a rear-facing camera, or a radar.

8. The system according to claim 1, wherein, The at least one data collection device includes a lidar.

9. The system according to claim 1, wherein, The reference algorithm includes object detection features for real-time detection and recognition of objects in a scene.

10. The system according to claim 1, wherein, The reference algorithm defines applications that are deployed separately.

11. A method for performing feature-related on-demand data collection in an autonomous vehicle, the method comprising: Provide at least one data collection device equipped with an autonomous vehicle, the at least one data collection device collecting data defining multiple different data scenarios around the autonomous vehicle; The data collection unit receives data collected by the at least one data collection device. The data received by the data collection unit is partially combined into a data scenario group, and the data scenario group is stored in either the data collection unit of the autonomous vehicle or a remote server. Receive a data collection request submitted by a user related to the feature, wherein the data collection request includes a predetermined format, including a request identifier, a feature identifier associated with the feature, and a data collection strategy identifier; and A data collection strategy is determined based on the data collection strategy identifier of the data collection request. The data collection strategy defines the data collection trigger event, data collection type, data collection duration, and reference algorithm. The reference algorithm defines how data is collected, including the input required by the reference algorithm, the data type of the input, the size of the input data, the output generated by the reference algorithm, the output type, the output size, and the execution environment required to execute the reference algorithm. The execution environment includes infrastructure services, available memory, and memory type. By executing the data collection strategy, data from the feature set is collected from the data scenario group, wherein the steps of executing the data collection strategy include: The reference algorithm is executed based on the data collection trigger event. Using the reference algorithm, data in the feature set is collected based on the data collection type; and Data from the feature set is collected during the data collection duration; The autonomous vehicle operates based at least in part on data from the feature set.

12. The method of claim 11, further comprising: Allocate storage devices for the data in the feature set; A storage strategy is adopted to store the data in the feature set in the storage device; as well as After receiving a data collection request from the user, data in the feature set is retrieved from the storage device as needed.

13. A method for performing feature-related on-demand data collection in an autonomous vehicle, the method comprising: Provide at least one data collection device equipped with an autonomous vehicle, the at least one data collection device collecting data defining multiple different data scenarios around the autonomous vehicle; Receive data collected by the at least one data collection device; A portion of the data received by the data collection unit is combined into a data scenario group, and the data scenario group is stored in either the data collection unit of the autonomous vehicle or a remote server. Receive a data collection request submitted by a user related to the feature, wherein the data collection request includes a predetermined format, the predetermined format including a request identifier, a feature identifier associated with the feature, and a data collection strategy identifier; A data collection strategy is determined based on the data collection strategy identifier of the data collection request. The data collection strategy defines the data collection trigger event, data collection type, data collection duration, and reference algorithm. The reference algorithm defines how to collect data, including the input required by the reference algorithm, the data type of the input, the size of the input data, the output generated by the reference algorithm, the size of the output, and the execution environment required to execute the reference algorithm. The execution environment includes infrastructure services, available memory, and memory type. When the reference algorithm is not stored locally in the autonomous vehicle, the reference algorithm is retrieved from the remote server; By executing the data collection strategy, data from the feature set is collected from the data scenario group, wherein executing the data collection strategy includes: The reference algorithm is executed based on the data collection trigger event. Using the reference algorithm, data in the feature set is collected based on the data collection type; and Data from the feature set is collected during the data collection duration; The autonomous vehicle operates based at least in part on data from the feature set.

Images