High-precision motion capture systems can cost more than $10,000, while affordable inertial alternatives are wireless-only, keep raw sensor data locked away and offer no scope for tuning the fusion algorithm. The purpose of this study is to design, build and validate an affordable, wired system that gives researchers unrestricted access to raw IMU data for custom sensor fusion.
A 12-node capture suit was built around MEMS inertial measurement units linked through a deterministic RS-485 wired network, with a central controller streaming raw accelerometer, gyroscope and magnetometer data at 100 Hz. Orientation was computed using a quaternion extended Kalman filter tuned transparently via Allan-variance characterization. Accuracy was assessed at 60 Hz against a Vicon T40S optical reference using a servo-driven turntable, and through walking and squatting trials with five participants, evaluated via RMSE, Pearson correlation and Bland-Altman analysis.
Against the Vicon T40S, the system produced RMSE values of 1.8° for pitch and 2.1° for roll, with strong linear correlation (Pearson r above 0.99) and clinically acceptable Bland–Altman limits. Yaw accuracy was weaker, at 6.2–7.1° RMSE, a direct trade-off from omitting adaptive magnetic rejection to preserve real-time stability. Hardware costs fell by 38–43% and installed cost by 53–58% relative to the Vicon T40S.
Most affordable inertial motion capture systems are wireless and release only preprocessed pose data through closed pipelines. This system instead combines a deterministic wired multi-node network, a fully transparent and adjustable Kalman filter and rigorous validation against an optical gold standard, at an installed cost under $5,000.
