Early in my career, I was offered a job by someone who wanted a convenient method to measure the potential amount of cut timber from logs carried on railroad cars before being purchased and used in a lumber mill. I did not take the job because the best tools I could think of at the time (in about 1970) involved photography and the use of planimeters in a primarily manual method. It is interesting to note that such a need is used as an example on page 7 of this book. Although I did not take the job at the time, I'd be more willing now to tackle such a job with the assistance of this text and software combination.
Fall detection devices have been
made before [2,5,7], but they are not
extensively used because frustration
with false positives and false negatives.
A false positive is false alarm, and a
false negative is a failure to send an
alarm that should have been sent. The
reason for this problem is that the
commercial fall detectors [2] use
accelerometers [5] to detect falls, and
accelerometers are noisy. This means
that they are overly sensitive. For
example, if someone were to sneeze, the high acceleration could set off the alarm [1]. If the
thresholds were to be set higher to avoid having sneezes set off alarms, then some real falls
may not be detected, instead causing false negatives. As a result, existing detectors have large
error rates [2].