Joint Angle Tracking with Single IMU and UKF
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for tracking the motion of multi-segment limbs over extended periods using inertial sensors face challenges such as integration drift, high expense, and complexity due to the need for multiple sensors, and poor performance in nonlinear dynamics with existing filtering techniques like the Kalman filter.
Innovation Solution
A technique using a single inertial measurement unit with a triaxial gyroscope and accelerometer, employing a nonlinear state space estimator like the unscented Kalman filter to estimate joint angles, which reduces the need for multiple sensors and handles nonlinear dynamics effectively by incorporating a kinematic model and state space evolution equations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple inertial sensors are placed at each segment of the limb, then measurement accuracy is improved, but device complexity and expense increase
Solution Approach 1:
The patent divides the multi-segment limb into distinct segments (e.g., thigh, shank, foot) and places a single inertial sensor unit at the distal end. The system then computationally segments the motion data to estimate orientations of multiple segments, replacing the need for physical segmentation with multiple sensors.
Solution Approach 2:
The patent introduces a computational intermediary (the state estimator and tracking filter) that processes inertial data from a single sensor to derive joint angle information. This computational mediator replaces the need for multiple physical sensors by mathematically reconstructing segment orientations from distal measurements.
2Device complexity
If traditional integration methods are used to estimate orientation from inertial data, then computational simplicity is maintained, but integration drift accumulates quickly
Solution Approach 1:
The patent implements a feedback mechanism through the tracking filter that continuously compares estimated joint angles with expected kinematic relationships. This feedback loop corrects drift by adjusting estimates based on the dynamic model, preventing error accumulation while maintaining computational efficiency.
Solution Approach 2:
The patent changes the computational approach from direct integration of raw inertial data to a state-space estimation framework that uses a dynamic model. This parameter change transforms the problem from accumulating integration errors to solving a constrained optimization problem that maintains reliability over extended periods.
3Adaptability or versatility
If the extended Kalman filter is used for nonlinear dynamics, then filtering capability is improved, but performance deteriorates due to linearization errors and computational burden
Solution Approach 1:
The patent substitutes the extended Kalman filter's linearization approach with a unscented Kalman filter that uses sigma points to capture nonlinear dynamics more accurately. This replacement eliminates the need for Jacobian matrix calculations and provides better tracking accuracy in highly nonlinear joint angle dynamics.
Data Source
AI summary
A method for estimating joint angles of a multi-segment limb from inertial sensor data accurately estimates and tracks the orientations of multiple segments of the limb as a function of time using data from a single inertial measurement unit worn at the distal end of the limb. Estimated joint angles are computed from measured inertial data as a function of time in a single step using a nonlinear state space estimator. The estimator preferably includes a tracking filter such as an unscented Kalman filter or particle filter. The nonlinear state space estimator incorporates state space evolution equations based on a kinematic model of the multi-segment limb.