Sensor Data Sharing Across Processing Pipelines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor data processing systems face challenges in making accurate perception decisions due to conflicting decisions from different sensor modalities, particularly in scenarios where sensors like image sensors struggle in low-light conditions while LiDAR sensors excel, leading to redundant and conflicting data.
Innovation Solution
Implementing a system where sensor data is shared across processing pipelines at various stages to leverage the strengths of different sensor modalities, such as sharing LiDAR data with image processing pipelines for feature extraction and vice versa, allowing for feature-level fusion and final decision-making through voting or weighting schemes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If multiple independent sensor processing pipelines are used, then each sensor modality can be processed independently, but conflicting perception decisions occur and reliability decreases
Solution Approach 1:
The patent merges multiple independent sensor processing pipelines into a unified architecture where sensor data is shared across pipelines at different processing stages. Instead of completely separate pipelines that produce conflicting decisions, the system combines image sensor data and LiDAR data into a shared data structure that multiple pipelines can access and process collaboratively, eliminating conflicting decisions while maintaining independent processing capabilities.
Solution Approach 2:
The patent implements a universal sensor data sharing mechanism where a single sensor data structure serves multiple processing pipelines simultaneously. The shared sensor data can be accessed and processed by different pipelines for different sensor modalities (image, LiDAR, radar), allowing the same data to fulfill multiple functions and produce consistent perception decisions across all pipelines.
2Measurement precision
If sensor data is shared across processing pipelines, then perception decision accuracy improves, but system complexity increases
Solution Approach 1:
The patent segments the sensor data sharing system into distinct processing stages (raw data level, feature extraction level, and decision level) that can be independently implemented and managed. This segmentation allows the system to share sensor data effectively while maintaining modularity and reducing overall system complexity through structured organization of the sharing mechanism.
Solution Approach 2:
The patent introduces an intermediary sensor data sharing structure that mediates between multiple processing pipelines. This intermediary layer manages the sharing of sensor data, features, and decisions between pipelines, simplifying the complexity by providing a standardized interface and coordination mechanism rather than requiring direct peer-to-peer interactions between all pipelines.
3Reliability
If feature-level fusion is implemented, then perception accuracy in challenging conditions improves, but processing time increases
Solution Approach 1:
The patent performs preliminary feature extraction and processing of sensor data before the fusion stage. By pre-processing image and LiDAR data to extract relevant features in advance, the system reduces the computational burden during the actual fusion operation, thereby maintaining high perception accuracy in challenging conditions while minimizing additional processing time.
Solution Approach 2:
The patent implements partial feature-level fusion by selectively fusing only the most relevant features from different sensor modalities rather than processing all possible features. This selective approach maintains perception accuracy in challenging conditions while avoiding the excessive processing time that would result from comprehensive feature fusion of all available data.
Data Source
AI summary
Sensor data captured at by different sensors may be shared across different sensor processing pipelines. Sensor processing pipelines may process captured sensor data from respective sensors. Some of the sensor data that is received or processed at one sensor data processing pipeline may be provided to another sensor data processing pipeline so that subsequent processing stages at the recipient sensor processing pipeline may process the combined sensor data in order to determine a perception decision. Different types of sensor data may be shared, including raw sensor data, processed sensor data, or data derived from sensor data. A control system may perform control actions based on the perception decisions determined by the sensor processing pipelines that share sensor data.