Vehicle Data Schema for Driver-Assistance Storage and Retrieval
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems lack a uniform format for storing data from driver-assistance systems of vehicles, making data sharing and processing between teams difficult, and there is a need for efficient data storage and retrieval solutions that support machine learning and algorithm verification.
Innovation Solution
A computer system with a defined schema comprising entities such as sensor, vehicle, log, stream, and algorithm entities, using a NoSQL database like MongoDB, with pre- and post-hooks for validation, and separate metadata and binary storage components to optimize data structure and retrieval.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If data is stored in various formats without a uniform schema, then data can be stored flexibly, but data sharing and processing between teams becomes difficult
Solution Approach 1:
The patent implements a universal data schema that can accommodate multiple data types (sensor data, annotation data, algorithm data) within a unified structure. The schema uses polymorphic data fields and type indicators to handle diverse data formats while maintaining consistent access patterns, enabling both storage flexibility and ease of processing across different teams and applications.
2Ease of operation
If all data is stored in a single database, then data retrieval is simple, but storage costs increase
Solution Approach 1:
The patent segments data storage into multiple specialized databases: a primary database for metadata and structured data, and secondary storage systems for raw sensor data and large binary objects. The unified schema spans across these segmented storage systems, allowing simple logical access while physically separating hot and cold data paths to reduce overall storage costs.
3Quantity of substance
If data is stored in a non-uniform format, then storage capacity is maximized, but machine learning and algorithm verification become difficult
Solution Approach 1:
The patent transforms raw heterogeneous data into a standardized parameterized format within the unified schema. Data is converted into structured fields with defined data types, units, and relationships, while preserving the original data through references. This parameterization enables efficient querying and processing for machine learning and algorithm verification while maintaining storage efficiency through selective data representation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A computer system for storing data of driver-assistance systems of vehicles may include: a sensor entity configured to store, for each of at least one sensor mounted on a vehicle, respective sensor configuration data comprising a respective sensor identifier; a vehicle entity configured to store respective vehicle configuration data comprising a respective vehicle identifier and a reference to at least one sensor identifier; a log entity configured to store respective logging configuration data comprising a respective log identifier and a reference to at least one vehicle identifier; a stream entity, wherein the stream entity is configured to store, for each of at least one stream, respective stream configuration data comprising a respective stream identifier, and a reference to at least one log identifier, and a sample entity configured to store respective sample configuration data, comprising a respective sample identifier and a reference to at least one stream identifier.