SLAM Feature Filtering Using IMU Prediction for Mobile Localization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional techniques for determining stationary features in autonomous mobile devices are computationally intensive, leading to increased latency and reduced performance in simultaneous localization and mapping (SLAM) systems, which affects the accuracy and speed of autonomous movement.
Innovation Solution
A method that uses stereo images from a pair of cameras to identify stationary features by comparing predicted positions based on inertial sensor data with actual positions, applying a threshold value to determine feature stability and reducing the processing load by designating moving features with bounding techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional techniques are used to determine stationary features, then comprehensive feature detection is achieved, but computational load increases and latency increases
Solution Approach 1:
The patent segments the feature detection process into multiple stages: initial feature detection from stereo images, prediction of feature positions using inertial sensor data, comparison with actual positions, and classification as stationary or moving. This segmentation allows the system to process features efficiently by handling different types of features through different computational paths, reducing overall processing latency while maintaining detection accuracy.
Solution Approach 2:
The patent applies preliminary action by using inertial sensor data to predict the positions of features before comparing them with actual detected positions. This prediction step is performed in advance to pre-filter and classify features, enabling the system to identify stationary features more quickly without compromising the comprehensiveness of feature detection.
2Measurement precision
If traditional techniques are used to determine stationary features, then all features are processed, but processing load and computational intensity increase
Solution Approach 1:
The patent extracts and separates moving features from stationary features by comparing predicted positions (based on inertial sensor data) with actual detected positions. By taking out the moving features through this comparison process, the system processes only the essential stationary features for SLAM operations, significantly reducing computational power requirements while maintaining position accuracy.
Solution Approach 2:
The patent applies partial action by not processing all detected features equally. Instead, it uses a threshold-based comparison approach where only features that meet certain criteria (stationary features within the threshold distance) are processed further for SLAM. This selective processing reduces computational load while maintaining sufficient position accuracy for autonomous navigation.
3Reliability
If all detected features are used for SLAM, then comprehensive mapping is achieved, but errors and non-physical solutions increase
Solution Approach 1:
The patent implements feedback by continuously comparing predicted feature positions (from inertial sensors) with actual detected positions, using the difference (error) to classify features as stationary or moving. This feedback mechanism ensures that only reliable stationary features are used for SLAM, reducing non-physical solutions while maintaining mapping completeness through continuous validation.
Solution Approach 2:
The patent changes the parameter of feature selection by introducing a threshold value for position comparison. Features are classified based on whether their position deviation from predicted positions exceeds this threshold. This parameter-based filtering ensures that only features with sufficient reliability (stationary features) are included in SLAM processing, reducing errors while maintaining comprehensive mapping coverage.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Significantly reduces errors in position accuracy and non-physical solutions, improving the overall performance of SLAM systems by focusing on stationary features, with observed reductions in errors by approximately 50% and non-physical solutions by one-third.
Implementation Method 1
data from an inertial measurement unit (IMU) indicative of a first orientation of the pair of cameras at the first time and a second orientation of the pair of cameras at the second time
Implementation Method 2
A first pair of images are acquired at a first time from a pair of cameras... A second pair of images are obtained at a second time from the pair of cameras
Data Source
AI summary
An autonomous mobile device moves through a physical space using simultaneous localization and mapping (SLAM) techniques. SLAM processes images from cameras to determine localization and trajectory of the device based on features that are assumed to be stationary. SLAM performance is improved by removing moving features from consideration. A first position of a feature at a first time and data from an inertial sensor are used to determine a predicted position at a second time. The predicted position is compared to a second position of the feature at the second time. This comparison takes into consideration an assumed Gaussian error distribution of how the positions are determined. If the predicted position differs from the second position by less than a threshold value, the feature may be determined to be stationary. The stationary features are then processed using SLAM to determine the localization and trajectory information.