Visual Inertial Odometry Scale Drift Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current visual inertial odometry systems for vehicle localization face challenges in semi-controlled environments with varied scenarios, as inertial measurements alone are insufficient for determining scale and depth, leading to scale drift and unobservable camera pose, especially at constant velocities, and rely on stationary features that may be scarce, making them unreliable for accurate vehicle positioning.
Innovation Solution
The method combines visual feature points from lane markers or traffic signs with inertial sensor measurements and satellite navigation to determine motion trajectory and camera pose relative to a ground plane, using robust perception algorithms like lane marker and traffic sign detectors to provide reliable scale information and correct for scale drift, enabling accurate vehicle positioning without relying on GPS or external maps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If inertial sensor measurements are used for vehicle localization, then the system can operate without GPS or external maps, but scale drift and unobservable camera pose occur especially at constant velocities
Solution Approach 1:
The patent introduces visual features from lane markers and traffic signs as intermediary elements that mediate between inertial sensors and the vehicle positioning system. These features provide external reference points that allow the system to observe scale and depth information, correcting the scale drift inherent in pure inertial navigation while maintaining GPS-independent operation.
Solution Approach 2:
The patent replaces reliance on mechanical inertial sensing alone with a vision-based measurement system that uses optical features (lane markers, traffic signs) to determine scale and position. This substitution introduces visual perception capabilities that can observe absolute scale information, eliminating the scale drift problem of inertial systems.
2Adaptability or versatility
If visual inertial odometry is used for vehicle positioning, then the system can provide localization in semi-controlled environments, but stationary features may be scarce making the system unreliable
Solution Approach 1:
The patent creates a universal positioning system that can operate across diverse environments by implementing multiple feature detection modes. The system can detect and utilize both lane markers (common on roads) and traffic signs (present at intersections and key locations), allowing it to adapt to different driving scenarios and maintain reliability whether features are sparse or abundant.
Solution Approach 2:
The patent changes the parameters of visual feature detection by switching between different object types (lane markers vs. traffic signs) based on environmental conditions. When lane markers are unavailable, the system transitions to detecting traffic signs, and adjusts detection parameters accordingly to maintain positioning reliability across varying environmental contexts.
3Measurement precision
If feature points from lane markers or traffic signs are used to determine motion trajectory, then accurate scale information can be obtained, but the system complexity increases
Solution Approach 1:
The patent segments the visual processing task into distinct modules: lane marker detection, traffic sign detection, feature point extraction, and trajectory calculation. Each module handles a specific aspect of the problem, making the overall complex system manageable and maintainable while achieving high measurement precision through specialized processing at each stage.
Data Source
AI summary
A method is described. The method includes obtaining a plurality of images. The method also includes detecting an object in the plurality of images. The method further includes determining a plurality of feature points on the object. The feature points have an established relationship to each other based on an object type. The method additionally includes determining a motion trajectory and a camera pose relative to a ground plane using the plurality of feature points.