Inertial Odometry Retroactive Sensor Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current inertial navigation systems face significant challenges in accurately compensating for sensor biases, particularly gyroscope biases, which lead to erroneous gravity compensation and subsequent positioning errors due to the integration of these errors over time, and existing methods like Kalman filtering are inefficient in real-time correction and do not provide a continuous propagation solution.
Innovation Solution
The method involves pre-integrating inertial sensor data to generate temporally continuous error propagation models, which are then used to calculate compensation gradients for pose parameters, allowing for retroactive calibration of inertial measurement units (IMUs) by predicting changes in pose parameters and minimizing residuals through optimization techniques, incorporating environmental cues and aiding information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If Kalman filtering is used for real-time sensor error compensation, then sensor bias correction can be performed continuously, but the computational complexity increases and real-time performance deteriorates
Solution Approach 1:
The patent segments the sensor error compensation process into discrete integration intervals between keyframes. Instead of continuous Kalman filtering, the system integrates sensor errors only at keyframe transitions, reducing computational load while maintaining correction effectiveness through the formula: δx_i = ∫(f(t) - b(t))dt from t_{i-1} to t_i
Solution Approach 2:
The patent performs preliminary integration of sensor measurements and error terms during the time interval between keyframes, storing accumulated values that are then processed retroactively. This allows the system to prepare compensation data in advance without requiring complex real-time calculations during keyframe processing
2Measurement precision
If sensor integration is performed continuously to reduce positioning error, then navigation accuracy improves, but error propagation and drift accumulate over time
Solution Approach 1:
The patent implements periodic resetting of error integration at keyframe boundaries. By establishing keyframes at regular intervals and resetting the integration state at each keyframe, the system periodically eliminates accumulated errors through retroactive calibration, preventing long-term drift while maintaining high measurement precision during integration intervals
Solution Approach 2:
The patent uses retroactive calibration to create a feedback mechanism where end-to-end pose constraints from keyframes are used to infer and correct sensor biases. The calibration parameters inferred from keyframe comparisons are then fed back to compensate sensor readings throughout the integration interval, creating a closed-loop error correction system
3Productivity
If retroactive calibration is performed using keyframe constraints, then computational efficiency improves, but the frequency of calibration updates decreases
Solution Approach 1:
The patent implements a dynamic keyframe selection strategy where the system adapts the frequency and timing of keyframe creation based on navigation conditions. During periods of high dynamic motion or significant error accumulation, keyframes are created more frequently to increase calibration update rate, while during stable conditions, fewer keyframes are used to maintain computational efficiency
Data Source
AI summary
Systems and methods for determining pose parameters of an inertial measurement unit (IMU) sensor include collecting measurement data generated by IMU sensors, using a processor to temporally integrate the measurement data, including any errors, generating a temporally continuous error propagation model, and temporally integrating the model to generate one or more compensation gradients for said pose parameters.