Estimation of upper-limb orientation based on accelerometer and gyroscope measurements

Area of Science:

• Biomechanics
• Sensor Technology
• Medical Device Engineering

Background:

• Accurate estimation of upper-limb orientation is crucial for various applications, including clinical assessment of neurological conditions and performance analysis in sports.
• Existing sensor systems often require a high number of components, increasing obtrusiveness, cost, and data bandwidth requirements.
• There is a need for efficient and accurate measurement systems, particularly for quantifying movement amplitudes up to 15 Hz, as required for assessing conditions like tremor in multiple sclerosis.

Purpose of the Study:

• To propose a novel solution for estimating upper-limb orientation using a minimized number of miniature accelerometers and gyroscopes.
• To develop a sensor system that delivers known accuracy over a defined frequency range, focusing on applications requiring up to 15 Hz movement quantification.
• To reduce sensor count for a less obtrusive, more cost-effective, and bandwidth-efficient measurement system.

Main Methods:

• Development of a novel sensor design based on composite filters to minimize sensor count while maintaining accuracy.
• Implementation of a simple estimator structure to provide insight into fundamental accuracy limitations.
• Quantification of performance for an estimator designed for shoulder, upper arm, lower arm, and hand orientations.

Main Results:

• The proposed system achieves good estimation accuracy for upper-limb orientations up to 15 Hz.
• The novel design utilizes a reduced total sensor count of 18, compared to 24 sensors required by conventional architectures.
• The composite filter-based approach enables accurate measurement of movement amplitude at higher frequencies with fewer sensors.

Conclusions:

• The developed sensor system offers an accurate and efficient solution for estimating upper-limb orientation, particularly for applications requiring high-frequency movement analysis.
• Minimizing sensor count through novel filter designs leads to practical advantages in terms of cost, obtrusiveness, and wireless data transmission.
• This approach provides a viable method for quantifying tremor in multiple sclerosis patients and has broader applications in sports training and virtual reality.