IMU Data Compression Using Remainder Rounding for Low-Power Transmission
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current inertial measurement unit (IMU) and application processor (AP) systems face challenges with excessive power consumption and buffer overflow due to high sampling rates and low update rates, which limit the flexibility in data transmission and accuracy in motion sensing applications.
Innovation Solution
The system employs a method of rounding velocity and orientation increments at the IMU, producing remainder values that are added to subsequent increments, along with signal pre-conditioning and variable quantization, and uses a FIFO buffer to transmit data on demand, allowing for flexible data compression and reduced power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high sampling rate (2-4 kHz) is used at the ADC to prevent aliasing and coning, then measurement precision is improved, but use of energy increases and data transmission bandwidth requirements increase
Solution Approach 1:
The patent segments the data processing pipeline into two distinct stages: high-rate sampling at the ADC for accurate motion detection, and low-rate SDI integration for application-level updates. This segmentation allows the system to maintain measurement precision at the sensing stage while reducing energy consumption and transmission bandwidth at the processing stage by transmitting only the integrated motion increments at lower rates (e.g., 24 Hz or 1 Hz).
Solution Approach 2:
The patent introduces an intermediary processing stage (the SDI unit) between the high-rate ADC sampling and the application processor. This intermediary performs strap-down integration to compute motion increments (Δq and Δv) from the high-rate samples, effectively decoupling the sampling rate from the transmission rate. The intermediary allows the system to benefit from high-rate sampling for accuracy while transmitting at lower rates to reduce power consumption.
2Use of energy by moving object
If low update rate is used to preserve power in consumer grade devices, then use of energy is reduced, but buffer overflow occurs when the AP is too busy to read out the buffer
Solution Approach 1:
The patent implements a dynamic buffer management system where the FIFO buffer size and update rate can be adjusted based on system conditions. The buffer is configured to accommodate variable transmission rates, allowing it to store accumulated motion increments when the AP is busy and then transmit them when ready. This dynamic approach prevents buffer overflow while maintaining low power consumption, as the system can operate at low update rates normally and temporarily buffer data during high-load periods.
3Use of energy by moving object
If SDI rate is limited to match low update rate for power management, then use of energy is reduced, but flexibility in data transmission is lost
Solution Approach 1:
The patent implements periodic SDI integration at the maximum sampling rate, accumulating motion increments over multiple sampling periods. The integrated results are then transmitted periodically at the application-required update rate (e.g., 24 Hz or 1 Hz). This periodic action allows the SDI unit to operate continuously at high rate for accurate integration while transmitting at variable lower rates, maintaining flexibility in data transmission without increasing average power consumption.
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
A method, controller and system in accordance with various aspects of the present disclosure facilitate reduced energy consumption in a motion sensing device having an inertial measurement unit (IMU), with a strap down integration unit, and an application processing unit (AP). The system and method include rounding velocity increments and orientation increments at the inertial measurement unit, thereby producing a remainder values. The remainder values are added to subsequent velocity increments and orientation increments prior to rounding of those values, and so on. In this way, while motion granularity is slightly decreased, there is no drift of integration errors over time.