Vehicle Pose Correction Using Semantic Contour Map Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for pose estimation in autonomous vehicles are not precise enough, particularly in scenarios with clustered landmarks, and often rely on correspondence-based approaches that are difficult to establish and prone to errors.
Innovation Solution
A method utilizing semantic contour image measurements from a monocular camera to perform correspondence-free pose correction by generating score and error images, projecting expected three-dimensional points, and applying iterative optimization to align with a sparse, semantically labeled map, allowing for precise pose estimation in six degrees of freedom without requiring depth measurements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If correspondence-based pose estimation methods are used, then the system can work with existing landmark data, but the method becomes difficult to establish and prone to errors especially with clustered landmarks
Solution Approach 1:
The patent introduces an intermediary optimization process that mediates between the initial pose estimate and the final corrected pose. This optimization acts as a bridge that avoids direct correspondence establishment between clustered landmarks, thereby reducing errors while maintaining reliability in pose estimation.
Solution Approach 2:
The system performs preliminary pose estimation using available landmark data before applying the correction optimization. This preliminary action establishes an initial pose that serves as a starting point for subsequent refinement, avoiding the need to establish correspondences from scratch and reducing complexity.
2Loss of information
If dense point clouds or captured images are used in the map, then more detailed information is available, but the map becomes less efficient for sparse representation and processing
Solution Approach 1:
The patent extracts only the essential landmark features from dense point clouds or captured images to create a sparse map representation. This extraction process removes redundant information while retaining critical pose estimation data, thereby maintaining information quality while improving processing efficiency.
Solution Approach 2:
The map data is segmented into discrete landmark features rather than processing entire dense point clouds or images. This segmentation allows the system to work with simplified representations that are more efficient to process while still containing sufficient information for accurate pose estimation.
3Adaptability or versatility
If semantic contour image measurements are used, then the method can operate in image space without depth measurements, but the pose correction must be more precisely constrained to be sufficient
Solution Approach 1:
The patent implements a feedback mechanism where the optimization process iteratively refines the pose correction based on the comparison between observed semantic contours and map features. This feedback loop ensures that the pose correction becomes sufficiently precise even though the system operates without depth measurements, maintaining adaptability while achieving required precision.
Data Source
AI summary
The invention relates to a method for correcting a pose (20) of a motor vehicle (10) in the surroundings (18) of the motor vehicle (10) by an assistance system (12) of the motor vehicle (10), the method comprising the steps of receiving first data (22) of semantic contour image measurements by an electronic computing device (14) of the assistance system (12), receiving second data (24) of an initial pose estimate by the electronic computing device (14), receiving third data (26) of semantically labeled map elements by the electronic computing device (14), generating a score image (28) and/or an error image (30) and its image derivatives based on the first data (22) by the electronic computing device (14), generating expected three-dimensional points (32) of the map elements at the initial pose estimate based on the second data (24) and third data (26) by the electronic computing device (14), comparing the expected three-dimensional points (32) of the map elements with the score image (28) and/or the error image (30) and its image derivatives to perform model-to-image alignment by the electronic computing device (14), based at least in part on projecting the expected three- dimensional points (32) into the score image (28) and/or the error image (30) and using the score image (28) and/or the error image (30) and its image derivative to perform an iterative optimization (34) by the electronic computing device (14), and transmitting the alignment's resulting pose correction (20) and pose correction uncertainty (36) to the assistance system (12). Furthermore, the invention relates to a computer program product as well as an assistance system (12).