Vehicle Sensor Data Fusion via Segmented Processing Zones
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing driving-environment sensing systems for vehicles face challenges in optimizing data fusion from multiple sensors, leading to inefficiencies in processing and robustness, particularly in automated driving applications where a larger number of sensors generate more data.
Innovation Solution
A system and method that utilize a complex sensor model to analyze data from various sensors, creating data objects which are then processed by a driving-environment modeling unit to generate a simpler model for control commands, allowing for improved data fusion and scalability, with separate dynamic and static models for handling different types of environmental data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a larger number of sensors are used to capture the driving environment, then the completeness and accuracy of environmental sensing is improved, but the computational complexity and data processing burden increase significantly
Solution Approach 1:
The patent divides the driving environment into multiple zones (first zone closer to vehicle, second zone farther away) and processes sensor data from each zone separately using different levels of complexity. This segmentation allows the system to handle comprehensive multi-sensor data without processing all data at maximum detail, reducing overall computational burden while maintaining sensing accuracy.
Solution Approach 2:
Different processing strategies are applied to different spatial regions: the first zone (closer to vehicle) receives more intensive processing with higher precision object detection and classification, while the second zone (farther away) uses simplified processing. This local differentiation optimizes computational resources by applying complexity only where necessary for safety-critical near-field detection.
2Measurement precision
If complex sensor models are used to analyze measurement data, then the precision of object detection and classification is improved, but the processing time and computational load increase
Solution Approach 1:
The patent segments object detection into two stages: initial detection using simplified models for rapid identification, followed by detailed classification using complex sensor models only for objects in the critical first zone. This staged approach maintains high detection precision for nearby objects while reducing overall processing time through early filtering.
Solution Approach 2:
The system applies full complex sensor model processing selectively to only those objects and regions where maximum precision is critical (first zone, potential hazards), rather than uniformly processing all sensor data at maximum detail. This partial application of complex processing reduces computational load while maintaining precision where it matters most.
3Reliability
If all sensor data is processed in detail, then the robustness of driving environment capture is improved, but the system's scalability and real-time performance deteriorate
Solution Approach 1:
The patent implements a hierarchical processing architecture that segments data flow into parallel streams: critical safety-related data from the first zone undergoes detailed robust processing, while less critical data from the second zone uses streamlined processing. This segmentation enables the system to maintain robustness for essential functions while achieving real-time performance through parallel execution.
Solution Approach 2:
The patent introduces intermediate processing stages that filter and prioritize sensor data before detailed analysis. These intermediary steps identify and flag only the most critical objects and events requiring full robust processing, allowing the majority of data to be handled through faster pathways while maintaining system-wide reliability through selective detailed analysis.
Data Source
AI summary
A system and a corresponding method for sensing the driving environment of a vehicle, the system having a plurality of sensors for capturing objects in the area surrounding the vehicle. The system has a plurality of evaluation units, an evaluation unit being bijectively associated in each case with a specific sensor and being adapted for analyzing a measurement performed by the sensor using a sensor model and for creating a data object to describe a particular object captured by the sensor. In addition, at least one driving-environment modeling unit is provided that is linked to each evaluation unit and is adapted for computing a driving environment model to describe the environment of the vehicle on the basis of the data objects. An evaluation unit and the at least one driving-environment modeling unit are thereby designed in a way that makes the sensor model more complex than the driving environment model.