Readme file
SERIES C
 Applied Statistics
Bayesian disclosure risk assessment: predicting small frequencies in contingency tables, by J. J. Forster and E. L. Webb, pages 551-570
General Description of the Data
The data described in the paper and used to model the hand trajectories was collected at the Human Motion Simulation Laboratory (HuMoSim) at the University of Michigan in 1999. The set of experiments described here concentrated on the motions of standing people performing reaches to 30 different shelves. The shelves are located on 3 towers (in front of the subject, to the right side and 45 degrees to the right), at 5 levels ranging from eye to ankle level and 2 depths (near and far). The subjects were asked to move a box with both hands, a vertical cylinder with the right hand and a horizontal cylinder with the right hand. The hand motion began in front of a person on a small home shelf with the subject delivering the object to the target shelf.
The subjects were selected to provide a means to assess the effects of anthropometry, gender, and age on the motions. The subjects ranged from very short to very tall and from 20 to 70 years of age. Ten were male and ten were female. All were right handed. A total of 2031 motions were performed by the group of 20 subjects. Some motions were replicated. Some motions were lost due to data collection errors. After removing these cases, 2019 cases were left as presented here.The data discussed here is a subset of a larger sequence of experiments --- see www.humosim.org for more.
The motion capture was achieved with an optical reflective marker system (Qualysis MacReflex) recording at 25 Hz. Our particular interest is in the movement of a marker attached to the right wrist, although
21 locations on the body were tracked in total. We choose to model the motion of the right wrist rather than hand because the hand motion will contain some rotation mixed in with the translation. Thus the data here consist only of right wrist trajectories.
Format of the Data
The file "preds.txt" contains target and subject information about each trajectory file. There are 2019 trajectory files with names like e05d21h2.loc where:
All files start with e, end with .loc and have a "d" (for deliver)
05 indicates subject number 5 (ranges 1-20)
21 indicates the target number (ranges 1-30)
h indicates the object moved was a horizontal cylinder (v=vertical
and b=two-handed box move)
2 indicates the replicate number (only some cases are replicated)
In the trajectory files, there are 4 columns. The first three columns
indicate the (x,y,z) coordinates of the right wrist. x is to the right, y is forward and z is up. The units are in cm and the origin is roughly between the feet of the subject in the rest position. The targets are shelves to place the object so some variation in the final wrist location is expected because the placement of the objects is not precisely controlled. The fourth column is an indicator variable where 1 indicates the frames we used for the data analysis. The hand is roughly stationary at the beginning and end of the reach and so it is necessary to trim this data. One could do this in several ways so we have included the excess data that we did not use so that others can experiment with other trimming methods.
The file "preds.txt" has the following columns:
* The trajectory file name
* The subject number (1-20)
* The target number (1-30)
* The object moved (h, v or b)
* The replicate number
* The sex (0=male, 1=female)
* The age in years
* The weight in kg
* The height in cm
R code
The file, bezfit.R contains some R functions that were used in the analysis of this data.
Acknowledgements
We acknowledge the vital support of Don Chaffin who founded the HuMoSim laboratory at the Center for Ergonomics at the University of Michigan and provided many years of support to the research presented in this paper. We are also grateful for the support of our colleagues and the industrial sponsors of HuMoSim.
Julian Faraway
Department of Mathematical Sciences
University of Bath
Claverton Down
Bath
BA2 7AY
UK
E-mail: jjf23@bath.ac.uk
|
