Mobile Apparatus Ego-Motion Estimation with Adaptive SLAM Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
SLAM-based algorithms for mobile apparatuses often converge to local minima, leading to inaccurate and uncorrectable ego-motion estimations, especially in environments without loop closures, and GPS data is unreliable for correction.
Innovation Solution
A method that combines a high-accuracy but less robust SLAM-based data stream with a more robust but less precise data stream, such as acceleration sensors, to correct ego-motion estimates by comparing their offsets and switching to the more reliable stream when errors exceed a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If SLAM-based algorithms are used for ego-motion estimation, then measurement precision is improved, but reliability deteriorates due to convergence to local minima
Solution Approach 1:
The patent merges SLAM-based ego-motion estimation with alternative estimation methods (such as acceleration sensor-based or visual odometry) into a unified system. Multiple estimators work together, where the SLAM algorithm provides high precision when reliable, while alternative methods provide robustness when SLAM converges to local minima. The system combines these complementary approaches to achieve both precision and reliability simultaneously.
Solution Approach 2:
The patent implements feedback mechanisms where the system continuously monitors the reliability of SLAM-based estimates and adjusts its behavior accordingly. When reliability metrics indicate potential local minimum convergence, the system switches to or blends with alternative estimation methods. This feedback loop ensures that the system maintains reliable ego-motion estimation by adapting to changing conditions in real-time.
2Measurement precision
If loop closure algorithms are used for correction, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-establishing multiple ego-motion estimators and reliability assessment mechanisms before actual navigation begins. The system prepares alternative estimation paths and correction strategies in advance, so when SLAM encounters local minima, pre-configured alternative methods are already available for immediate deployment, reducing the need for complex real-time decision-making and algorithm switching.
3Reliability
If multiple data streams are integrated using Kalman filter, then reliability is improved, but measurement precision deteriorates due to difficulty in controlling operations on independent data streams
Solution Approach 1:
The patent applies dynamics by making the data integration approach adaptive rather than static. Instead of using a fixed Kalman filter configuration that treats all data streams uniformly, the system dynamically selects and weights different estimation methods based on current operational conditions, sensor reliability, and environmental factors. This dynamic approach allows the system to maintain high precision by selecting the most appropriate data stream for current conditions while still benefiting from the reliability of multiple sources.
Data Source
Figure 1~2
Figure 3~4
Figure 5a
AI summary
Provided is a method of determining a change in a position and/or orientation of a mobile apparatus, comprising: acquiring a first data stream indicating a first estimated change in the position and/or orientation of the mobile apparatus in respective ones of a plurality of time frames; acquiring a second data stream indicating a second estimated change in the position and/or orientation of the mobile apparatus in the respective ones of the plurality of time frames, wherein the first data stream has a higher accuracy than the second data stream and the first data stream has a lower robustness than the second data stream; and responsive to a comparison of the first indicated change with the second indicated change in the respective ones of the plurality of time frames, using the first estimated change or the second estimated change to determine the change in the position and/or orientation of the mobile apparatus.