A novel motion-reconstruction method for inertial sensors with constraints

Area of Science:

• Robotics and Motion Capture
• Sensor Data Fusion
• Calibration Techniques

Background:

• Inertial Measurement Units (IMUs) are crucial for motion reconstruction, but suffer from accumulated deterministic and stochastic errors.
• Existing calibration methods often fail to fully address sensor misalignment in a common reference system.
• Sensor fusion is commonly used to mitigate stochastic errors, but deterministic errors, particularly misalignment, remain a challenge.

Purpose of the Study:

• To present a novel motion reconstruction method that corrects IMU data using optimization and calibration polynomials.
• To address deterministic errors in IMUs, specifically sensor misalignment, through advanced calibration techniques.
• To minimize deviations from motion constraints for improved position and orientation accuracy.

Main Methods:

• Developed a novel motion reconstruction approach employing optimization with correction polynomials for IMU data.
• Implemented gyrometer and accelerometer calibration using an industrial manipulator to correct deterministic errors like misalignment.
• Conducted experiments with constant orientation and simultaneous translation/rotation using an industrial manipulator and five individual IMUs.

Main Results:

• Achieved an average decrease of 95% in maximum position error compared to standard methods.
• Reduced average position error over the measurement duration by nearly 90%.
• Demonstrated significant improvements in motion reconstruction accuracy for both static and dynamic scenarios.

Conclusions:

• The proposed calibration and optimization methods effectively minimize deterministic and stochastic errors in IMU-based motion reconstruction.
• The novel approach offers substantial improvements in position, velocity, and orientation accuracy.
• The method is applicable to experiments starting and ending at standstill, enhancing its practical utility.