QUICK SEARCH:   [advanced] Author: Keyword(s):
Home     Help     Feedback     Subscriptions     Archive     Search     Table of Contents

First published online June 16, 2005
Journal of Experimental Biology 208, 2503-2514 (2005)
Published by The Company of Biologists 2005
doi: 10.1242/jeb.01658

This Article Summary Full Text Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Pfau, T. Articles by Wilson, A. M. Search for Related Content PubMed PubMed Citation Articles by Pfau, T. Articles by Wilson, A. M.

A method for deriving displacement data during cyclical movement using an inertial sensor

Thilo Pfau^1,*, Thomas H. Witte¹ and Alan M. Wilson^1,2

¹ Structure and Motion Laboratory, The Royal Veterinary College, Hawkshead Lane, Hatfield, Hertfordshire, AL9 7TA, UK
² Structure and Motion Laboratory, Royal National Orthopaedic Hospital, Brockley Hill, Stanmore, Middlesex, HA7 4LP, UK

View larger version (24K): [in a new window]   Fig. 1. Schematic drawing of the data collection set-up. The inertial sensor is mounted over the thoracic spinous processes (the withers) of the horse with the motion analysis marker wand attached. Cables are run from the inertial sensor to a laptop computer (via the safety harness mast) as well as from the optical motion capture cameras to a second laptop computer.

View larger version (127K): [in a new window]   Fig. 2. The harness employed for attachment of the inertial sensor to the thoracic spinous processes of the subject animal. The wands and markers for 3D optical motion capture are also shown.

View larger version (16K): [in a new window]   Fig. 3. Schematic drawing of the coordinate systems used. Orientation of the inertial sensor is expressed as a sequence of roll, pitch and heading from the `sensor' into the `earth' system. In order to compare optical motion capture and inertial sensor output, the `global' coordinate system is used.

View larger version (28K): [in a new window]   Fig. 4. Displacement data for x (craniocaudal), y (lateral) and z (dorsoventral) movement for optical motion capture (blue) and inertial sensor (red) for a series of strides at canter (9 m s-1). In the integration process, a context window of one stride to each side of the current stride has been used for the mean subtraction.

View larger version (43K): [in a new window]   Fig. 5. Optical motion capture (blue) and inertial sensor (red) outputs for x, y and z displacements and roll, pitch and heading angles of the sensor mounted over the thoracic spinous processes of a horse during treadmill locomotion (mean ± S.D.). Data are shown for the mean gait cycle (stride) at (A,B) walk (1.4 m s-1), (C,D) trot (3.5 m s-1) and (E,F) left lead canter (9.0 m s-1). For the x, y and z displacements and for the heading angle, the stride-to-stride differences, as indicated by the standard deviations, are bigger than the differences between the two methods (difference between the mean traces). For roll and pitch orientation, however, there is a small offset between the mean traces of the two methods, which increases from walk to trot and canter. The biggest inter-stride variability can be found in the y direction for both methods and across all gaits. This demonstrates the tendency of the horse to drift from side to side while performing on the treadmill.

View larger version (19K): [in a new window]   Fig. 6. Motion analysis (blue) and inertial sensor (red) outputs for x, y and z displacements after high-pass filtering (mean ± S.D.). Data are shown for a stride at (A) walk (1.4 m s-1), (B) trot (3.5 m s-1) and (C) left lead canter (9.0 m s-1). Smaller standard deviations (compared with the unfiltered displacements in Fig. 5), especially in the y direction, show that high-pass filtering is a valid means of eliminating non-cyclical inter-stride differences of the displacements.

View larger version (22K): [in a new window]   Fig. 7. Histograms of frame-wise difference between motion analysis and inertial sensor for roll, pitch and heading angles and x, y and z displacements for the whole data set. Histograms are shown for (A) walk (1.4 m s-1), (B) trot (3.5 m s-1) and (C) left lead canter (9.0 m s-1). The spread of the histograms is generally increasing from walk to trot and canter. In addition, there is an offset in roll and pitch orientation, which increased with speed.

View larger version (21K): [in a new window]   Fig. 8. Histograms of frame-wise difference between motion analysis and inertial sensor x, y and z displacements after high-pass filtering. Histograms are shown for (A) walk (1.4 m s-1), (B) trot (3.5 m s-1) and (C) left lead canter (9.0 m s-1). The reduced spread of the histograms (compared with the unfiltered displacement error histograms in Fig. 7) demonstrates the ability of the inertial sensor integration process to capture the cyclical components of the displacements.

View larger version (13K): [in a new window]   Fig. 9. Ranges of x, y and z displacements are shown as a function of speed and gait for a horse exercising on a treadmill. Ranges are calculated as the difference between maximum and minimum values within the mean stride. Error bars represent ±1 S.D., and the number of strides in each category is given. Walk is represented by open circles, trot by triangles and canter by diamonds.