Abstract
dc:description.abstractKathylee Pinnock Branford Manual wheelchairs are essential to mobility and independence, yet accurately characterizing wheelchair motion in real-world environments remains challenging. Inertial measurement units (IMUs) provide a practical means of estimating wheelchair kinematics, but existing trajectory-reconstruction studies have largely been limited to short, controlled trials. This report developed and evaluated a four-IMU framework comprising one sensor on each rear wheel and two sensors on the wheelchair frame. Three orientation algorithms (VQF, Mahony AHRS, and Madgwick AHRS), two accelerometer ranges (±2 g and ± 4 g), and calibrated and uncalibrated sensor data were evaluated during indoor motion-capture trials and a outdoor campus route. Indoor trajectories were most accurately reconstructed using frame-based Madgwick AHRS, with two dimensional position RMSE as low as 0.116 m for the fused estimate in the repeated-turning trial. Outdoors, the ±2 g configuration produced substantially lower trajectory and slope errors than the ±4 g configuration. For the representative calibrated ±2 g trial, ChairL-VQF produced the lowest GPS-referenced two-dimensional position RMSE (8.67 m), while ChairR-VQF produced the lowest full-trial relative loop-closure error (0.29%). Terrain and ramp slope were estimated most accurately using ChairR-Madgwick AHRS, with RMSE values of 1.21° and 2.22°, respectively. Calibration had little effect on the ±2 g results but was essential for the ±4 g configuration, for which uncalibrated data produced implausible trajectory estimates. These findings support the use of calibrated ±2 g sensors, with VQF prioritized for long-duration trajectory reconstruction and Madgwick AHRS prioritized for slope estimation.
Degree
thesis:*- Name thesis:degree_name
- MS
- Level thesis:degree_level
- Problem/Project Report
- Discipline thesis:degree_discipline
- Mechanical and Aerospace Engineering
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Pinnock Branford, Kathylee
- Contributors dc:contributor
-
- Stephen M. Cain
- Nicholas Szczecinski
- Natalia Schmid
Subjects
dc:subject × 10Identifiers
dc:identifier.*- Identifier
-
<p><a href="https://orcid.org/0009-0002-2991-7384">https://orcid.org/0009-0002-2991-7384</a></p>
https://researchrepository.wvu.edu/etd/13492