Sensor IC Posture Calculation via Dual-Frequency Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing inertial navigation systems face accuracy issues when calculating the posture of moving objects with large variations in angular velocity, such as humans, due to low computation frequencies and high sample rejection rates, leading to degraded precision.
Innovation Solution
An IC for sensors is developed that includes a detection unit, AD conversion unit, and posture variation calculating unit, which operates at a higher frequency than the host terminal, allowing for accurate posture calculation by integrating angular velocity data at a higher input frequency and reducing CPU burden by maintaining a lower input frequency for posture updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If posture computation is performed at low frequency with larger order of Taylor expansion, then computation precision is improved, but the number of unused samples increases and accuracy deteriorates for objects with large angular velocity variation
Solution Approach 1:
The posture computation process is segmented into two distinct frequency domains: high-frequency posture variation calculation (using small order Taylor expansion) and low-frequency posture correction (using large order Taylor expansion). This segmentation allows each computation to be optimized for its specific frequency range, preventing the degradation that occurs when a single low-frequency computation is used for high-frequency variations.
Solution Approach 2:
The invention dynamically changes the computation parameters (Taylor expansion order and sampling frequency) based on the frequency characteristics of the motion being measured. For high-frequency posture variations, small order expansion at high frequency is used; for low-frequency corrections, large order expansion at low frequency is applied. This parameter adaptation resolves the contradiction between precision and reliability.
2Measurement precision
If angular velocity data is integrated at high frequency, then integration accuracy is improved, but CPU burden increases
Solution Approach 1:
The invention introduces an intermediary computational structure where high-frequency angular velocity data is first processed through small order Taylor expansion to generate posture variations, which are then fed into a low-frequency correction stage. This intermediary process allows high-frequency data to be utilized for accuracy while the final correction at low frequency reduces the overall computational burden on the CPU.
Data Source
AI summary
An IC for a sensor includes a detection unit which detects an angular velocity signal of a moving object based on a signal from a sensor element, an AD conversion unit which converts an analog signal from the detection unit into a digital signal, and a posture variation calculating unit which calculates a variation in posture of the moving object during a predetermined period based on the signal from the AD conversion unit.