Moving-Vehicle Camera Pose Fusion for Long-Term Drift Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing camera pose estimation methods in advanced driver assistance systems (ADAS) suffer from long-term drift and inaccuracies, particularly in environments where ground-view and overhead images are used for localization and mapping.
Innovation Solution
A combined SLAM and G2O pose estimation method that integrates simultaneous localization and mapping (SLAM) with ground-to-overhead (G2O) pose estimation, utilizing a loss function to minimize drift by combining SLAM and G2O poses through a pose graph optimization, and includes spatial bounds and change comparisons to filter out erroneous G2O poses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If SLAM algorithm is used for camera pose estimation, then localization capability is improved, but long-term drift occurs causing accuracy degradation
Solution Approach 1:
The patent introduces an intermediary mapping between ground-view images and overhead images as a mediator to bridge the SLAM localization system with a reference frame. This mapping relationship serves as a correction mechanism that prevents long-term drift by periodically aligning the SLAM pose estimates with the overhead image reference, thus resolving the contradiction between localization capability and pose accuracy.
Solution Approach 2:
The patent implements a feedback mechanism where the pose graph optimization uses both SLAM poses and G2O poses to correct drift. The system continuously refines pose estimates by comparing SLAM-derived poses with G2O poses from overhead image matching, creating a feedback loop that maintains accuracy while preserving localization capability.
2Measurement precision
If G2O pose estimation is used, then mapping accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the pose estimation process into two distinct components: SLAM-based localization for real-time tracking and G2O-based mapping for accuracy refinement. By dividing the computational task into these segments, the system can leverage the strengths of both approaches while managing complexity through structured processing stages.
Solution Approach 2:
The patent merges SLAM pose estimates with G2O pose estimates through a unified pose graph optimization framework. This combination allows the system to benefit from both real-time localization efficiency and mapping accuracy, integrating the computational advantages of both methods into a cohesive system.
3Measurement precision
If multiple pose estimation methods are combined, then overall accuracy is improved, but system complexity increases
Solution Approach 1:
The patent creates a universal pose graph optimization framework that can handle multiple pose estimation methods (SLAM and G2O) within a single unified system. This multi-functional approach allows the system to process both localization and mapping data through a common optimization pipeline, improving accuracy while managing system complexity through standardization.
Data Source
AI summary
A computer includes a processor and a memory, and the memory stores instructions executable by the processor to determine a first set of SLAM poses of a camera with respect to an environment by performing a simultaneous localization and mapping (SLAM) algorithm, determine a second set of G2O poses of the camera based on a plurality of ground-view images from the camera and an overhead image depicting the environment, and determine a third set of final poses of the camera by minimizing a loss function derived from a pose graph of the final poses. The loss function is based on the SLAM poses in the first set and the G2O poses in the second set.