Classifications

• • G—PHYSICS
  • G07—CHECKING-DEVICES
  • G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
  • G07C5/00—Registering or indicating the working of vehicles
  • G07C5/008—Registering or indicating the working of vehicles communicating information to a remotely located station
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
  • G06F16/90—Details of database functions independent of the retrieved data types
  • G06F16/906—Clustering; Classification
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N20/00—Machine learning
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N5/00—Computing arrangements using knowledge-based models
  • G06N5/04—Inference or reasoning models
• • G—PHYSICS
  • G07—CHECKING-DEVICES
  • G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
  • G07C5/00—Registering or indicating the working of vehicles
  • G07C5/08—Registering or indicating performance data other than driving, working, idle, or waiting time, with or without registering driving, working, idle or waiting time
  • G07C5/0808—Diagnosing performance data
• • G—PHYSICS
  • G07—CHECKING-DEVICES
  • G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
  • G07C5/00—Registering or indicating the working of vehicles
  • G07C5/08—Registering or indicating performance data other than driving, working, idle, or waiting time, with or without registering driving, working, idle or waiting time
  • G07C5/0841—Registering performance data
  • G07C5/085—Registering performance data using electronic data carriers

Definitions

• the present disclosure relates generally to automotive data processing, and particularly to methods and systems for metadata generation and exporting in vehicles.
• Autonomous cars and other vehicles generate and process high volumes of data, sometimes on the order of Terabytes, in the course of a day of driving.
• Sources of data in an autonomous car comprise, for example, cameras and other sensors, Advanced Driver- Assistance Systems (ADAS), Telematics Control Units (TCU), Infotainment systems, as well as various Electronics Control Units (ECUs).
• ADAS Advanced Driver- Assistance Systems
• TCU Telematics Control Unit
• ECUs Electronics Control Units
• copious volumes of data are uploaded on a regular basis to a remote processor.
• An embodiment that is described herein provides an automotive data processing system including a storage subsystem and a processor.
• the storage subsystem is disposed in a vehicle and is configured to store at least data produced by one or more data sources of the vehicle.
• the processor is installed in a vehicle and is configured to apply, to the data stored in the storage subsystem or that is en route to be stored in the storage subsystem, at least one model that identifies one or more specified features-of-interest in the data, so as to generate metadata that tags occurrences of the specified features-of-interest in the stored data, and to export at least part of the metadata to an external system that is external to the vehicle.
• the processor is configured to identify occurrences of a same feature-of-interest in the data produced by two or more different ones of the data sources, and, in generating the metadata, to tag the data so as to link to one another the occurrences of the same feature-of-interest appearing in data produced by the two or more different ones of the data sources.
• the data produced by the two or more ones of the data sources are of a same type of data.
• the data produced by the two or more ones of the data sources are of different types of data.
• the two or more ones of the data sources are disposed at different locations in the vehicle.
• the processor is further configured to receive from the external system, in response to the exported metadata, a request for one or more selected portions of the data relating to the features-of-interest, and to provide, to the external system, at least the selected portions of the data but less than all the data.
• the processor is configured to provide to the external system two or more selected portions of the data, which were produced by two or more different ones of the data sources and which are associated with a same feature-of-interest.
• the processor is configured to select the part of the metadata for exporting, in accordance with a selection criterion.
• the processor is configured to export the metadata without concurrently exporting data corresponding to the metadata.
• the storage subsystem is configured to store at least data produced by multiple independent data sources in the vehicle. In another embodiment the storage subsystem is configured to store at least data produced by multiple data sources installed at different locations in the vehicle. In an embodiment, the storage subsystem is configured to store two or more different types of the data.
• the processor is configured to receive the model from the external system.
• the processor is configured to export the metadata while the vehicle is connected to a docking station.
• the processor is configured to export the metadata independently of whether the vehicle is connected to a docking station.
• the model includes an Artificial Intelligence (AI) inference model
• the processor is configured to generate the metadata by applying the AI inference model to the data.
• the processor is configured to receive the AI inference model pre-trained from the external system, and to update a training of the AI inference model in the vehicle.
• AI Artificial Intelligence
• the storage subsystem includes a centralized storage device that is configured to store the data produced by a plurality of the data sources.
• the centralized storage device is configured to store the data produced by two or more ones of the data sources that are of different types.
• the processor is further configured to apply to the data a second model different from the first model, the second model identifying one or more second features-of-interest in the data, to add to the metadata second occurrences of the second features- of-interest in accordance with the second model, and to provide to the external system one or more of (i) metadata relating only to the occurrences of the features-of-interest, (ii) metadata relating only to the second occurrences of the second features-of-interest, and (iii) metadata relating both to the occurrences and to the second occurrences.
• the processor is configured to store the metadata in structured form in the storage subsystem.
• an automotive data processing method including storing, in a storage subsystem disposed in a vehicle, at least data produced by one or more data sources of the vehicle.
• a processor installed in a vehicle at least one model, which identifies one or more specified features-of-interest in the data, is applied to the data stored in the storage subsystem or that is en route to be stored in the storage subsystem, so as to generate metadata that tags occurrences of the specified features-of-interest in the stored data.
• At least part of the metadata is exported to an external system that is external to the vehicle.
• an automotive data processing system including a plurality of data sources and a processor.
• the data sources are installed in a vehicle and are configured to generate data.
• the processor is installed in the vehicle and is configured to collect the data generated by the data sources, to store the data locally in a storage device disposed in the vehicle, without exporting the data outside the vehicle, to identify one or more portions-of-interest in the locally stored data, and, while the vehicle is connected to a docking station that is coupled to an external system located remotely from the docking station, to upload to the external system at least the identified portions-of-interest and less than all the locally stored data.
• the processor is configured to collect and store the data, and identify the portions-of-interest, independently of whether the vehicle is connected to a docking station.
• the processor is configured to collect the data from two or more different ones of the data sources that produce a same type of data.
• the processor is configured to collect the data from two or more different ones of the data sources that produce different types of data.
• the processor is configured to collect the data from two or more different ones of the data sources that are disposed at different locations in the vehicle.
• the processor is configured to identify and upload two or more portions-of-interest, which were produced by two or more different ones of the data sources and which are associated with a same feature-of-interest.
• an automotive data processing method including generating data by a plurality of data sources installed in a vehicle.
• the data generated by the data sources is collected and stored locally in a storage device disposed in the vehicle, without exporting the data outside the vehicle, one or more portions-of-interest of the locally stored data are identified, and, while the vehicle is connected to a docking station that is coupled to an external system located remotely from the docking station, at least the identified portions-of- interest, and less than all the locally stored data, is uploaded to the external system.
• an automotive data processing system including multiple data sources, a packet network and a processor.
• the data sources are distributed at different locations in a vehicle and are configured to generate data.
• the packet network is disposed in the vehicle and is configured to transfer the data from the multiple data sources to a central storage location in the vehicle.
• the processor is installed in the vehicle and is configured to generate metadata corresponding to the data transferred from the multiple data sources and stored at the central storage location, and to transfer one or more selected portions of the sensor data to an external system that is external to the vehicle, the selected portions being selected based on the metadata corresponding to the data transferred from the multiple data sources and stored at the central storage location.
• the multiple data sources include two or more different data sources that are configured to produce different types of data.
• the data produced by the two or more ones of the data sources are of different types of data.
• the processor is configured to select the one or more portions of the sensor data, at least partly in response to an indication of one or more features-of-interest received from the external system.
• the processor is configured to generate the metadata by applying an Artificial Intelligence (AI) inference model to the data stored at the central storage location.
• AI Artificial Intelligence
• the processor in generating the metadata, is configured to tag one or more occurrences of one or more features-of-interest in the data.
• the processor in generating the metadata, is configured to tag occurrences of a same feature-of-interest in two or more different types of the data, or in data generated by two or more different data sources.
• an automotive data processing method including generating data by multiple data sources distributed at different locations in a vehicle, and transferring the data from the multiple data sources to a central storage location in the vehicle, over a packet network disposed in the vehicle.
• metadata is generated corresponding to the data transferred from the multiple data sources and stored at the central storage location, and one or more selected portions of the sensor data are transferred to an external system that is external to the vehicle, the selected portions being selected based on the metadata corresponding to the data transferred from the multiple data sources and stored at the central storage location.
• an automotive data analysis system including an interface for communicating with a vehicle, and a computer.
• the computer is configured to define a model that identifies one or more specified features in data produced by one or more data sources of the vehicle, to provide the model to a processor installed in the vehicle, to receive metadata from the processor in the vehicle, wherein the metadata was generated by the processor using the model and tags occurrences of the specified features in the data, and to analyze the received metadata.
• the interface is configured to receive the metadata while the vehicle is connected to a docking station. In another embodiment, the interface is configured to receive the metadata from the vehicle independently of whether the vehicle is connected to a docking station.
• the model includes an Artificial Intelligence (AI) inference model
• the computer is configured to train the AI inference model and to provide the trained AI inference model to the processor in the vehicle.
• the computer is configured to send to the processor in the vehicle, in response to the received metadata, a request for one or more selected portions of the stored data, to receive the requested one or more selected portions from the processor, and to analyze the one or more selected portions of the data.
• AI Artificial Intelligence
• the computer is configured to request two or more selected portions of the stored data, which were produced by two or more different ones of the data sources. In yet another embodiment, the computer is configured to request two or more selected portions of the stored data, which are associated with a same feature-of-interest.
• an automotive data analysis method including, in a computer external to a vehicle, defining a model that identifies one or more specified features in data produced by one or more data sources of the vehicle, and providing the model to a processor installed in the vehicle.
• Metadata is received in the computer external to the vehicle, from the processor in the vehicle, wherein the metadata was generated by the processor using the model and tags occurrences of the specified features in the data.
• the received metadata is analyzed.
• FIG. 1 is a block diagram that schematically illustrates an automotive data processing system, in accordance with an embodiment that is described herein;
• Fig. 2 is a flow chart that schematically illustrates a method for metadata generation and processing in the automotive system of Fig. 1, in accordance with an embodiment that is described herein.
• An autonomous car may generate one or more video feeds from video cameras, sensor data of various kinds streaming from sensors disposed at various locations in vehicle, radar and lidar information, driver behavior information, information from infotainment subsystems and user applications, and location and navigation information such as GPS data and maps.
• an autonomous car may generate on the order of four Terabytes of data.
• the data produced in a vehicle during operation can provide extremely valuable information if analyzed correctly.
• GPS data and car driving data can be used for updating commute prediction models; camera footage can be used to optimize the autonomous driving capabilities and safety procedures of autonomous cars; driving behavior information can be used by insurance companies to determine an insurance premium for the driver, to name only a few examples.
• an automotive data processing system is deployed in a vehicle.
• the automotive data processing system comprises a storage subsystem coupled to a processor.
• the storage subsystem stores raw data that is produced by the various data sources.
• the processor pre-processes the raw data by identifying specified features-of-interest, and generates metadata that tags occurrences of the specified features-of-interest in the raw data.
• the metadata or at least a selected part thereof, is exported from the vehicle to the external system.
• the external system typically analyzes the metadata to identify specific portions of the raw data that are of interest, and retrieves these specific portions from the vehicle for analysis.
• the automotive data processing system is configured to communicate with the external system while the vehicle is connected to a docking station.
• the processor stores the metadata locally in the storage subsystem on an ongoing basis. When the vehicle is docked, i.e., parked and connected to a docking station, the processor transmits the accumulated metadata to the external system.
• the in-vehicle processor identifies the features-of- interest and generates the metadata by applying a pre-trained Artificial Intelligence (AI) inference model to the raw data.
• AI Artificial Intelligence
• the AI model may be supplied, for example, by the external system.
• the processor applies several different AI models, e.g., for identifying different features-of-interest, or for identifying the same features-of-interest in different types of raw data.
• the same feature-of-interest may appear in two or more different streams of raw data that are produced by two or more data sources, e.g., sensors that are of the same type but are disposed at different locations in the vehicle, or sensors of different type that are disposed at the same location or different locations in the vehicle.
• the in-vehicle processor identifies the feature-of-interest (e.g., object or event) in the different streams of raw data, and generates metadata that tags portions of the raw data which are common to and shared by the occurrences of the same feature- of-interest in the different streams. Further aspects of generating and processing metadata relating to different media objects are addressed in U.S. Patent Application Publication 2020/0042548, entitled “Metadata Generation for Multiple Object Types” whose disclosure is incorporated herein by reference.
• the metadata is several orders of magnitude smaller than the raw data, simpler and has a structured format.
• the metadata can be transmitted to the external system over a short period of time using modest bandwidth.
• the specific portions of raw data, which the external system subsequently requests based on the metadata, are also much smaller than the overall volume of raw data. Therefore, the disclosed technique provides an optimal “division of labor” that (i) overcomes the communication bottleneck between the in- vehicle system and the external system, and (ii) does not compromise the quality of data analysis.
• Fig. 1 is a block diagram that schematically illustrates an automotive data processing system 20, in accordance with an embodiment that is described herein.
• System 20 is installed in a vehicle, and comprises various sensors 24, multiple Electronics Control Units (ECUs) 32, an Advanced Driver- Assistance System (ADAS) 28, an infotainment system 30 and a central computer 34.
• ECUs Electronics Control Units
• ADAS Advanced Driver- Assistance System
• Sensors 24 may comprise, for example, video cameras, velocity sensors, accelerometers, audio sensors, infra-red sensors, radar sensors, lidar sensors, ultrasonic sensors, rangefinders or other proximity sensors, or any other suitable sensor type.
• each ECU 32 (sometimes referred to as a “zone ECU”) is connected to the sensors installed in a respective zone of the vehicle.
• Each ECU 32 typically controls its respective sensors 24 and collects data from the sensors.
• one or more of sensors 24, e.g., image sensors are connected directly to ADAS 28, not via ECUs 32.
• any element in system 20 that produces data is considered a “data source.”
• ECUs 32, ADAS 28, infotainment system 30 and central computer 34 are regarded as examples of “electronic subsystem” of the vehicle.
• a certain system component may serve both as a data source and as an electronic subsystem.
• the various electronic subsystems of system 20 are deployed at various different locations in the vehicle, and communicate over a packet network installed in the vehicle.
• the packet network comprises an Ethernet network, but other suitable network protocols can also be used.
• the network comprises multiple Ethernet links 36, and one or more Ethernet switches 40.
• the bit rate used in the network may be 10G bits per second (lOGbps) in accordance with IEEE 802.3ch, 1000Mbps in accordance with IEEE 802.3bp, 100Mbps in accordance with IEEE 802.3bw, 10Mbps in accordance with IEEE 802.3 cg(10Base-T Is), or any other suitable bit rate.
• Links 36 may comprise, for example, twisted-pair copper links or any other type of link suitable for Ethernet communication.
• system 20 further comprises a centralized storage device 44 (also referred to as a “storage subsystem”) and a processor 48.
• a centralized storage device 44 also referred to as a “storage subsystem”
• processor 48 a processor
• Storage device 44 stores raw data 45 that is provided by the various data sources in the vehicle (e.g., sensors and/or electronic subsystems).
• processor 48 identifies specified features-of-interest in raw data 45, generates metadata 46 that tags occurrences of the specified features-of-interest in the raw data, and stores metadata 46 in storage device 45.
• processor 48 identifies the features-of-interest by applying a computational Artificial Intelligence (AI) engine to raw data 45.
• AI computational Artificial Intelligence
• Processor 48 exports at least part of the metadata to an external system that is external to the vehicle, e.g., to a cloud-based data processing system or another suitable computer that is external to the vehicle.
• centralized storage device 44 comprises a Solid State Drive (SSD) that is connected directly to the packet network installed in the vehicle.
• SSD Solid State Drive
• centralized storage device 44 comprises two interfaces that are connected via respective links 36 to two different ports of two different switches 40. This scheme provides a degree of redundancy, since failure of a single interface (or of the corresponding link) will not disconnect centralized storage device 44 from the network.
• processor 48 and storage device 44 communicate with one another over a fast computer bus, such as a Peripheral Component Interconnect express (PCIe) bus.
• PCIe Peripheral Component Interconnect express
• processor 48 may be connected independently to the packet network, in which case processor 48 and storage device 44 communicate with one another over the network.
• Processor 48 is configured to communicate with a docking station or any other suitable interface for transferring data (not seen in the figure) to which the vehicle connects, using a data link 52.
• the data sources send raw data 45 to centralized storage device 44 over the network.
• the data sources and centralized storage device 44 may communicate using any suitable protocol, such as the Non Volatile Memory express over Fabrics (NVMe-oF) protocol or NVMe-over-TCP.
• NVMe-oF Non Volatile Memory express over Fabrics
• NVMe-over-TCP NVMe-over-TCP
• the configuration of communication system 20 and its components, such as of the various devices and electronic subsystems and/or of centralized storage device 44, as shown in Fig. 1, are example configurations that are depicted solely for the sake of clarity. In alternative embodiments, any other suitable configurations can be used.
• system 20 may comprise other suitable types of electronic subsystems and/or devices, laid out and connected in any other suitable scheme.
• the packet network (including links 36 and switches 40) may have any other suitable topology.
• system 20 and its components may be implemented using dedicated hardware or firmware, such as using hard-wired or programmable logic, e.g., in an Application-Specific Integrated Circuit (ASIC) or Field-Programmable Gate Array (FPGA). Additionally or alternatively, some functions of the components of system 20, e.g., of ECUs 32, storage device 44 and/or processor 48, may be implemented in software and/or using a combination of hardware and software elements. Elements that are not mandatory for understanding of the disclosed techniques have been omitted from the figure for the sake of clarity.
• ASIC Application-Specific Integrated Circuit
• FPGA Field-Programmable Gate Array
• ECUs 32, storage device 44 and/or processor 48 may be implemented in one or more programmable processors, which are programmed in software to carry out the functions described herein.
• the software may be downloaded to any of the processors in electronic form, over a network, for example, or it may, alternatively or additionally, be provided and/or stored on non-transitory tangible media, such as magnetic, optical, or electronic memory.
• Raw data 45 may comprise, for example, video data, audio data, sensor data, radar and/or lidar data, driver behavior data, data from infotainment subsystems and/or user applications, location and/or navigation data such as GPS data and maps, and/or any other suitable type of data.
• the raw data may be unstructured.
• the raw data is typically of multiple different types (modalities), and typically originates from multiple independent data sources of multiple different types that are installed at multiple different locations in the vehicle.
• multiple streams of raw data are generated by multiple sensors of the same type (e.g., multiple video cameras or multiple lidars, for example), each of which sensors is disposed at a different location in the vehicle.
• multiple instances of the same type of sensor are disposed at the same location in the vehicle to provide redundant streams of the same type of raw data.
• Processor 48 scans raw data 45 in order to identify one or more specified features-of-interest, generates a database of metadata 46 that tags occurrences of the specified features-of-interest in the raw data, and stores metadata 46 in storage device 45.
• processor 48 may search for any suitable features-of-interest, such as, for example:
• ⁇ Objects or environmental conditions in the vicinity of the vehicle that are visible in video or detectable in sensor data e.g., an obstacle, a traffic sign, another vehicle or a pedestrian approaching the vehicle.
• ⁇ Events relating to driver behavior e.g., vehicle slowing down, speeding up in response to a traffic light, swerving to avoid an accident or obstacle, or reacting to a traffic sign).
• processor 48 may identify any other suitable features that might be of interest for analysis.
• a certain feature-of-interest occurs in raw data of multiple data sources.
• the multiple data sources may be of different types (e.g., video, audio, LIDAR, radar etc.), or multiple instances of the same type of raw data (e.g., video captured at different locations in the vehicle).
• a road accident, or a mechanical breakdown or failure may be visible in video feeds, audible in audio data, sensed by various sensors and/or reported by one or more electronic subsystems.
• processor 48 tags the occurrence of such a feature-of-interest (e.g., object or event) using the same tag, or otherwise in a way that links the occurrences in the data of the different sensors to one another.
• processor 48 may use various techniques for identifying the specified features-of-interest in raw data 45.
• processor 48 runs a computational Artificial Intelligence (AI) engine that is configured to compile and apply pre-trained AI inference models that identify features-of-interest in raw data 45.
• processor 48 applies to the raw data several different AI models, e.g., for identifying different features-of-interest, or for identifying the same feature-of-interest in different types of raw data.
• AI Artificial Intelligence
• processor 48 receives the pre-trained Artificial Intelligence (AI) model or models from the external system, e.g., via data link 52 while the vehicle is connected to a docking station. During on-going operation of the vehicle, processor 48 applies the pre-trained AI model or models to raw data 45 provided by the various data sources.
• AI Artificial Intelligence
• Processor 48 identifies occurrences of the features-of-interest in raw data 45 using the AI model or models, and generates metadata 46 that tags these occurrences.
• Processor 48 may apply AI models to raw data already stored in storage deice 44, and/or to raw data that is en route to be stored in storage device 44.
• Metadata 46 is stored temporarily in storage device 44.
• processor 48 transmits metadata 46 that is computed in the vehicle to the external system, typically without transferring all of the raw data.
• processor 48 may transmit metadata 46 to the external system independently of whether the vehicle is connected to a docking station, e.g., over a suitable wireless link such as a cellular link, mobile Wi-Fi connectivity coupling or other suitable wireless link, during vehicle operation.
• the external system may send an AI model to processor 48 independently of whether the vehicle is connected to a docking station, e.g., over a suitable wireless link.
• communication via a docking station e.g., receiving an AI model, and exporting metadata and/or requested portions of raw data is regarded as an offline process.
• processor 48 transmits all of metadata 46 to the external system. In other embodiments, processor 48 selects only part of metadata 46 using a selection criterion, and sends only the selected part of the metadata to the external system.
• the disclosed techniques can be used for analyzing a wide variety of features in various use-cases. Consider, for example, a use-case in which the external system analyzes the behavior of an autonomous vehicle when animals appear on or around the road. To perform this sort of analysis, in an embodiment, the external system trains an AI model that recognizes animals in raw video data. Processor 48 receives the pre-trained AI model, compiles it and applies it to the video camera feeds provided by the various cameras in the vehicle.
• processor 48 Based on the identified occurrences of the feature-of-interest (appearances of animals in the videos), processor 48 generates metadata that tags the video frames in which animals appear. Tagged frames may be located in video streams from two or more video sensors disposed at different locations in the vehicle. Processor 48 uploads the metadata to the external system for initial analysis.
• the external system may then request portions-of-interest of raw data 45 that are related to the identified occurrences, e.g., video excerpts that start slightly before appearance of an animal.
• the requested portions-of-interest of raw data need not necessarily be the exact ones in which the occurrences were identified.
• the external system may log the times at which animals were identified in the video, and request sensor data that correspond to these times.
• processor 48 uploads portions-of- interest of raw data without being requested to do so by the external system, e.g., along with the metadata.
• the external system analyzes the behavior of an autonomous vehicle when pedestrians are present, in rainy conditions.
• the external system provides processor 48 with multiple pre-trained AI models - e.g., a model for recognizing rainy scenes in video, another model for recognizing pedestrians in video, and possibly other models for recognizing rainy conditions in the outputs of other types of sensors such as environmental-condition sensors.
• processor 48 uses these models to generate metadata that tags time intervals in the video in which the scene is rainy, time intervals in the video showing pedestrians, and possibly in the raw data of other sensors.
• the external system may later find the intersection between the various types of intervals (i.e., identify time intervals of rainy conditions containing pedestrians), and request sensor data and/or subsystem data corresponding to these time intervals.
• Fig. 2 is a flow chart that schematically illustrates a method for metadata generation and processing, in accordance with an embodiment that is described herein.
• the left- hand side of the figure depicts actions performed by the external system (e.g., cloud-based data processing system).
• the right-hand side of the figure depicts actions performed by system 20 in the vehicle.
• the external system trains an AI inference model for identifying a certain feature-of-interest in raw data 45 of a certain type (e.g., identifying a human in video data, identifying a mechanical breakdown in audio data or sensor data, etc.).
• the trained model is provided to processor 48, for example (although not necessarily) while the vehicle is connected to a docking station.
• processor 48 compiles the AI model to run on the processor’s computational AI engine.
• processor 48 may update the training of the AI model to account for local inputs, e.g., perform additional training using data acquired in the specific vehicle. Such an update can be useful, for example, to better fit the area typically driven by the specific vehicle. Additional training of this sort may be performed at any stage, e.g., while the vehicle is connected to a docking station and/or while driving.
• processor 48 identifies occurrences of the feature-of-interest in raw data 45 by applying the AI model.
• processor 48 At a metadata generation operation 76, processor 48 generates metadata 46, which tags the identified occurrences. Each tag in metadata 46 typically indicates the feature- of-interest being identified, and the location of the occurrence in the raw data (e.g., frame number in video, or other location depending on the type of raw data). Processor 48 may apply AI models to raw data already stored in storage device 44, and/or to raw data that is en route to be stored in storage device 44.
• Metadata generation may be applied to multiple sets of raw data, and may involve collating tags of the same feature-of-interest (e.g., object or event) that appears in different sets of raw data supplied by different sensors of the same type or different types.
• feature-of-interest e.g., object or event
• processor 48 stores metadata 46 in storage device 44.
• processor 48 notifies the external system that metadata 46 is available.
• the external system fetches metadata 46 from storage device 44 of the vehicle.
• metadata 46 may be “pulled” from the vehicle by the external system or “pushed” to the external system on initiative of the vehicle. For example, processor 48 may push metadata to the external system periodically, or when the vehicle is connected to a docking station.
• the external system runs a suitable initial analysis on the metadata.
• the initial analysis typically determines, based on the metadata, one or more portions-of-interest of raw data 45 that pertains to the feature-of-interest.
• the external system fetches the portions-of-interest of raw data 45 from storage device 44 of the vehicle. The external system then analyzes the portions-of-interest of raw data, at an analytics operation 100.
• the process described above may be repeated with a different AI model that processor 48 re-maps to the computational AI engine and applies so as to tag different features-of-interest (or the same features-of-interest in a different type of raw data).
• Processor 48 may add the new metadata to the existing metadata database, or generate and store a separate database of metadata 46 in storage device 44.

Landscapes

• Engineering & Computer Science (AREA)
• General Physics & Mathematics (AREA)
• Physics & Mathematics (AREA)
• Theoretical Computer Science (AREA)
• Software Systems (AREA)
• Data Mining & Analysis (AREA)
• General Engineering & Computer Science (AREA)
• Artificial Intelligence (AREA)
• Evolutionary Computation (AREA)
• Computing Systems (AREA)
• Mathematical Physics (AREA)
• Databases & Information Systems (AREA)
• Computational Linguistics (AREA)
• Medical Informatics (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Traffic Control Systems (AREA)

Priority Applications (3)

Applications Claiming Priority (2)

Publications (2)

Family

ID=74095929

Family Applications (1)

Country Status (5)

Families Citing this family (3)

Citations (2)

Family Cites Families (136)

• 2020
  • 2020-12-08 CN CN202080078835.2A patent/CN114730324A/zh active Pending
  • 2020-12-08 JP JP2022521079A patent/JP7658958B2/ja active Active
  • 2020-12-08 US US17/114,534 patent/US12183125B2/en active Active
  • 2020-12-08 WO PCT/IB2020/061625 patent/WO2021116899A1/en not_active Ceased
  • 2020-12-08 EP EP20829989.1A patent/EP4073665A1/en active Pending
• 2024
  • 2024-11-11 US US18/942,828 patent/US20250069445A1/en active Pending

Patent Citations (2)

Also Published As

Similar Documents

Legal Events