Using sensory data fusion methods for infant body posture assessment

Infant body posture assessment is very important for monitoring the development of motor skills. The currently utilized sensory-supported measurement systems are often subject to shortcomings, such as system complexity or inaccuracy. This paper presents a novel approach integrating a combination of a pressure mattress along with inertial and magnetic measurement units on infant trunk and arms to diminish the effects of such drawbacks. The proposed multi-sensor system is validated by comparison of kinematic parameters, such as head movement, arm workspace area, and spectral arc length of hand velocity data to results of a referential optoelectronic measurement system. Integration of sensory data fusion methods provides high system accuracy and reliability. Acquired results confirm that the proposed system can be used as an adequate substitution to expensive optoelectronic measurement systems, as the arm kinematic estimation errors are in the range of 2 cm. Also presented are the results of simplified system versions with only 1 IMU per arm. Such approaches however result in lower system complexity, but also lower accuracy. Nevertheless results imply that the arm posture and workspace assessment is still reliable enough for frequent practical use. By taking into account all the evaluation results, the proposed measurement system could importantly contribute to the field of sensory-supported infant body posture assessment.