Sensor Track Fusion Using Distance and Overlap Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor information fusion technologies struggle to maintain reliability and accuracy in complex driving scenarios such as low-speed merging and turning vehicles, as they are designed primarily for constant-speed situations with sufficient vehicle spacing.
Innovation Solution
A sensor information fusion method that selects target tracks based on a combined cost calculation, considering both distance and overlapping areas between tracks, using weight values adjusted according to the relative areas and distances, to improve fusion accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sensor information fusion is performed using conventional methods designed for constant-speed vehicles with sufficient spacing, then fusion reliability is maintained in simple driving conditions, but fusion accuracy deteriorates in complex situations such as low-speed merging and turning vehicles
Solution Approach 1:
The patent applies dynamics by making the association threshold dynamic rather than fixed. The threshold is adjusted based on the relative velocity between the ego vehicle and target vehicles, and the distance to target vehicles. This allows the fusion system to adapt to different driving scenarios (constant-speed vs. low-speed merging/turning), resolving the contradiction between maintaining reliability in simple conditions and achieving accuracy in complex conditions.
Solution Approach 2:
The patent changes key parameters (association threshold, weight values) based on driving conditions. By modifying the threshold parameter according to relative velocity and distance, and adjusting weight values for different sensor types based on scenario, the system achieves both reliability in simple conditions and accuracy in complex conditions like low-speed merging and turning vehicles.
2Ease of operation
If a fixed association threshold is used for track fusion, then processing simplicity is maintained, but fusion performance deteriorates in varying driving conditions
Solution Approach 1:
The patent transforms the fixed threshold into a dynamic threshold that automatically adjusts based on relative velocity and distance parameters. This maintains processing simplicity through automated adaptation while significantly improving fusion performance across varying driving conditions, eliminating the need for manual threshold tuning.
Solution Approach 2:
The system performs self-adjustment of the association threshold based on real-time sensor data (relative velocity, distance). The fusion algorithm automatically selects appropriate threshold values and weight parameters without external intervention, maintaining simplicity while improving reliability through condition-based adaptation.
Data Source
AI summary
A sensor information fusion method of an embodiment includes selecting target tracks having distances from a reference track within a predetermined distance from among tracks obtained by a plurality of sensors, and selecting a target fusion track to be fused with the reference track on the basis of sums of first calculation values calculated according to distances between the selected target tracks and the reference track and second calculation values calculated according to overlapping areas of the target tracks and the reference track.