Subject:
|
Re: forwarded message from Chris Robson - about compasses
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Thu, 29 Aug 2002 14:34:47 GMT
|
Viewed:
|
839 times
|
| |
|
|
This is the Dinsmore compass which was mentioned in lugnet.robotics a year
or two ago.
I evaluated this device back then and found it to be less than what I would
consider adequate for robotic navigation use.
It is designed to be used in applications like automotive where the user is
not looking for "navigation" class accuracy. In fact, they have one version
which only outputs the 8 directions, N, NE, E, SE, S, SW, W and NW as suited
to a rear view mirror display common in a lot of US model cars. The "analog
version" is alos pretty poor. It's outputs are nominally sine and cosine,
but in reality aren't close enough to those functions to get better than
about 20 degree accuracy.
The device is based on a conventional north seeking mechanical compass. The
bearing system is poor and relies on a continuous source of vibration to
overcome the inherent "stiction" associated with the bearing's friction. (A
system which works fine in a car with continuous engine noise being
transmitted through the fabric of the structure.) In my experimentation,
slow rotation of the device could pull it up to 30 degrees away from north
before the needle would swing to a new position. Being mechanical, there is
the attendant overswing problem to deal with too.
The electrical outputs (for the analog version) are derived from two hall
effect sensors which detect the needle position. Their output is not
"nicely" related to the needle position. The output is also seriously
compromised if the device is angled. The needle stays level while the
sensors tilt with the
body of the device which dramatically changes their relative orientation to
the needle's magnetic field which they are trying to measure.
I was unable to build an RCX compatible compass sensor of enough accuracy to
enable a platform to navigate around the sides of a 4 foot square and return
to its starting point. (I have done this test with other compass
technologies and have easily succeeded.)
JB
In lugnet.robotics, Russell Nelson <nelson@crynwr.com> writes:
> Can somebody do me a favor and look at this company's sensors to see
> if they really are appropriate for Lego Robotics?
> -russ
|
|
Message is in Reply To:
2 Messages in This Thread:
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
