Subject:
|
Re: LIDARStation and LIDARDisp application finished
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Tue, 19 Feb 2002 23:55:07 GMT
|
Original-From:
|
Chris 'Xenon' Hanson <XENON@3avoidspamDNATURE.COM>
|
Reply-To:
|
xenon@+NoSpam+3dnature.com
|
Viewed:
|
1008 times
|
| |
|
|
Andy Gombos wrote:
> > I'd be very interested in seeing a two-axis controlled pan/tilt head
> > with an ultrasonic sensor on it. Such a device could perform a rudimentary
> > 3D scan of a hemisphere of space. Alas, it would require so many IO ports
> > that putting it onto a mobile robot base would be nearly impossible.
> Not really. I believe this enough to try it soon. :) Two motors to turn
> and tilt the camera, and one for a differential drive setup(possibly using
> the reverse differential device). All the sensor ports would be used
> however - 1 for the ultrasonic sensor and 2 for the rotation sensors. A
> datapoint is logged for every tick of the rotation sensor, then a 3d map of
> the area is generated. This would require some simple 3d programming on the
> computer - easy things like simply transforming a polar coordinate into
> cartesian, then into 3d.
I'm highly interested in seeing your results. The RCX obviously is not
brainy enough to process and use the 3D data, but a host computer could be.
I'd be interested in helping with the host software on a project like this.
I've thought about ways in which one could use fewer motors and rotation
sensors on the scanning head, thus saving more inputs for navigation sensors
or other uses. I am trying to conceive of a single-output, single-input
fully mechanical panoramic helical scanning sensor head. I have not found the
mechanical design skill and proper lego parts to execute this. I think maybe
turntables might be the key.
A design of this type would work primarily with optical sensors, as it
would rely on having the sensor face upward, and deflect off of an inclined
mirror rotating about the vertical axis. The inclination of the mirror would
mechanically change once per revolution, either discretely or continuously.
A 'sync' sensor would trigger once per revolution to provide sampling
registration of all samples collected during the previous sweep. The
rotational drive would have mechanical limit stops that would halt the
rotation at the two ends of the multi-revolution scan. The drive motor would
need to be protected against stall. If designed properly, the mechanical stops
should halt the mechanism with the sync sensor triggered so that the RCX can
detect end-of-scan limits by seeing the sync sensor held on.
The sync sensor, a simple boolean switch, could be wired in series with the
optical sensor itself, saving one input. The sync trigger could be detected by
a strong out-of-range value returned by the optical sensor when the sync sensor
closes the circuit.
This would introduce a single 'blind spot' strip in the panoramic scan. If
this was unacceptable, a second sensor input could be dedicated to the sync
trigger. Perhaps it could be ganged onto the input of another sensor in the
locomotion system in a way that ensures that both sensors won't be used
simultaneously.
This would leave two outputs and two inputs left over for locomotion systems
and sensors.
On the plus side, the wired portion of the optical sensor never rotates,
solving the cable winding problem. It also allows for more complex and more
bulky optical apparatus, such as the RCX's own IR emitters.
> Andy
Chris - Xenon
--
Chris Hanson | Xenon@3DNature.com | I've got friends in low latitudes!
New WCS 5 Demo Version! http://www.3DNature.com/demo/
"There is no Truth. There is only Perception. To Perceive is to Exist." - Xen
|
|
Message is in Reply To:
 | | Re: LIDARStation and LIDARDisp application finished
|
| (...) Not really. I believe this enough to try it soon. :) Two motors to turn and tilt the camera, and one for a differential drive setup(possibly using the reverse differential device). All the sensor ports would be used however - 1 for the (...) (22 years ago, 19-Feb-02, to lugnet.robotics)
|
5 Messages in This Thread:
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
