Unified Data Processing for Autonomous Driving Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In autonomous driving systems, the current method of independent data processing leads to increased system overheads and reduced performance efficiency due to repeated data copying from memory to the graphics processing unit, causing operations to be suspended and failing to meet real-time demands.
Innovation Solution
A data processing method where a data processing party simultaneously acquires and processes data for multiple receiving parties based on pre-received processing mode information, storing the processed data in a storage area for each party, thereby reducing the need for multiple copies and improving efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If each data receiving party copies data from memory to graphics processing unit separately, then each process can process data independently, but system overheads sharply increase and performance efficiency is reduced
Solution Approach 1:
The patent merges multiple independent data copying operations into a single unified copying operation. A data processing party simultaneously acquires data for multiple data receiving parties and processes it according to their respective processing mode information, eliminating redundant copying operations and reducing system overheads while maintaining independent processing capabilities.
Solution Approach 2:
The data processing party performs multiple functions: it simultaneously serves multiple data receiving parties, handles different processing modes (such as motion compensation processing), and manages data distribution to multiple parties. This multi-functional approach reduces the need for separate processing paths for each receiving party.
2Reliability
If data is copied multiple times from memory to graphics processing unit, then each receiving party gets processed data, but operations are suspended and real-time performance is not met
Solution Approach 1:
The data processing party receives processing mode information from each data receiving party in advance and prepares processing parameters beforehand. This preliminary preparation allows the system to execute data processing operations continuously without suspension, as all necessary processing instructions are ready before data copying begins.
Solution Approach 2:
The system implements continuous data processing by having the data processing party simultaneously process data for multiple receiving parties in one operation. This eliminates the stop-start nature of separate processing operations, ensuring continuous useful action and meeting real-time performance requirements.
3Adaptability or versatility
If independent processing procedures are used for each receiving party, then data processing can be customized per party, but system overheads increase due to repeated copying operations
Solution Approach 1:
The patent combines multiple customized processing operations into a single unified processing framework. The data processing party receives different processing mode information from each receiving party and applies appropriate processing (such as motion compensation) within a single simultaneous operation, maintaining customization while reducing system overhead.
Data Source
AI summary
The present application discloses a data processing method and apparatus. A specific implementation of the method includes: simultaneously acquiring, by a data processing party, data to be used by a plurality of data receiving parties, and processing the to-be-used data according to processing mode information that is received in advance from each data receiving party and that indicates a processing mode for the to-be-used data, to obtain processed to-be-used data corresponding to the data receiving party; and storing the processed to-be-used data into a storage area corresponding to the data receiving party. An independent data processing party for simultaneously acquiring data that needs to be processed by each data receiving party is provided, so that the data that needs to be processed by each data receiving party is acquired at a time.