Uncertainty-Based Bounding Box Fusion for Stable Object Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current sensor fusion methods for autonomous vehicles do not adequately consider the uncertainty of sensor data, leading to unstable and inaccurate fused bounding boxes that can 'jump' around objects, affecting tracking performance.
Innovation Solution
Generate a fused bounding box based on uncertainty by comparing parameters of multiple sensor-based bounding boxes, using refined or consolidation fusion techniques based on the discrepancy between them, to account for and minimize sensor uncertainty.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If sensor fusion is performed without considering uncertainty, then processing speed is improved, but tracking stability deteriorates
Solution Approach 1:
The patent changes the parameter consideration by explicitly incorporating uncertainty parameters (covariance matrices) into the fusion process. When uncertainty is high, the system switches to a more conservative fusion approach that maintains stability, while when uncertainty is low, it uses a faster approach that maintains processing speed.
2Device complexity
If sensor fusion is performed without considering uncertainty, then device complexity is reduced, but measurement precision deteriorates
Solution Approach 1:
The system performs preliminary assessment of uncertainty levels before executing the fusion operation. By pre-evaluating the uncertainty of sensor data, the system can select the appropriate fusion strategy in advance, avoiding the need for complex real-time uncertainty analysis during fusion while still achieving precise results.
3Stability of the object's composition
If conservative estimate is used for fused bounding box, then tracking stability is improved, but area of bounding box increases
Solution Approach 1:
The patent implements a dynamic approach where the conservativeness of the bounding box estimate adjusts based on uncertainty levels. When uncertainty is high, a more conservative (larger) bounding box is used to maintain stability. When uncertainty is low, the bounding box becomes tighter, reducing area while maintaining stability. This dynamic adjustment resolves the contradiction between stability and area.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
This document describes techniques and systems (104) for generating a fused object bounding box (122) based on uncertainty. At least two bounding boxes (118, 120), each associated with a different sensor, is generated. A fused center point and yaw angle as well as length, width, and velocity can be found by mixing the distributions of the parameters from each bounding box. A discrepancy between the center points of each bounding box can be used to determine whether to refine the fused bounding box (e.g., find an intersection between at least two bounding boxes) or consolidate the fused bounding box (e.g., find a union between at least two bounding boxes). This results in the fused bounding box having a confidence level of the uncertainty associated with the fused bounding box. In this manner, better estimations of the uncertainty of the fused bounding box may be achieved to improve tracking performance of a sensor fusion system.