large omnidirectional "robot" for bombsight simulator

[jonbarril]

I'm a volunteer at a small air museum (Oakland Air Museum).  Someone donated several WWII bombsights, which are working.  Most museums put them in a case and that's it.  I thought it would be way cool to mount one on a platform that could roll over a map while simulating crosswind motion of an airplane in flight.  Achieving such motion requires a vehicle that can "crab" (i.e. move in a given direction while skewed).  I started googling for omnidirectional vehicles and such, and came across this robot site.

Any suggestions on how to scale up a 3 or 4 wheeled omnidirectional platform such as the one described in the tutorials to a size sufficient to hold two seated people (one "pilot" doing the steering, and one "bombardier" doing the sighting).  My background is EE, not ME, so this is a bit out of my league.  Any suggestions or pointers for where to start with plans and parts?

The platform can be low to the ground and only needs to move slowly and steadily.  Precise motion and tracking is not needed since that is what the gyro stabilized sights were supposed to deal with in a plane buffeted by wind and other more deadly factors. The platform will be used indoors and can be plugged in (no batteries).

--jon

[Steve Joblin]

conventional wisdom would be to create an omnidirectional platform that moves the people... an alternative approach, would be to keep the people stationary and move a map underneath giving the illusion of motion... check out this post (http://www.societyofrobots.com/robotforum/index.php?topic=1778.0) for omni-directional treadmill that could be controlled by motors and a joystick.  Imagine the pilot and bombardier sitting in a "cockpit" suspended over such a treadmill.  The nice about this is that there are no "boundaries" and they can fly forever!

[dunk]

i'd put a flat panel monitor under the sight and display google maps imagery on it.
after that it becomes a programming challenge rather than a mechanical one.

dunk.

[Admin]

As a mechanical engineer, I'd say the mechanical way is asking for too much trouble.

In addition to the other ideas already suggested, consider a projector hooked up to a laptop/joystick displaying a map like dunk suggested. The people can sit down on an elevated platform, with the projector(s) underneath pointing down on a horizontal screen.

[Admin]

After I just made that post, I realized a giant omni-wheel robot would be fun to drive . . .

You will need three or four big geared down (low speed high torque) DC motors, some big omni-wheels rated for the weight you expect, encoders on each wheel, wood+steel frame, and some motor drivers. All this will be controlled by a microcontroller.

All of the above will be very expensive, minus the mcu. You might want to first decide on a budget. Im thinking about $1k if you get lucky with used parts/scrap.

hope that gets you started . . .

[jonbarril]

 I wasn't expecting so much this soon. Great ideas! Thanks!  I can see the elegance of putting a flat panel under the sight.  There is, of course, the reality that these devices actually have to move in order to work correctly and demonstrate their 60+ year old sophistication -- they have azimuth and vertical sensing gyros inside.  As such, the bombsight would have to physically turn in response to pilot commands (tilt can be simulated in other ways), along with the graphics in the panel.  Of course, this could all be simulated via graphics alone, but it is compelling to actually hear the gyros inside whirring and grinding as the sight turns and the motors correct for the precession.

Another aspect of this problem is that there would likely be others standing around watching this exhibit waiting until they get a turn.  Actually seeing the vehicle inching along over a map and crabbing would provide a much more intuitive feel for what is going on, especially with a laser pointer or something showing the groundtrack, release, and impact (for an extreme progressive liberal this sounds so warlike, but it is easy to appreciate the elegance of this solution to such a compelling problem for that time).

--jon

[jonbarril]

After I just made that post, I realized a giant omni-wheel robot would be fun to drive . . .

You will need three or four big geared down (low speed high torque) DC motors, some big omni-wheels rated for the weight you expect, encoders on each wheel, wood+steel frame, and some motor drivers. All this will be controlled by a microcontroller.

...

Do the wheels really require encoders?  It seems that they could be run open loop since the pilot and bombardier working together would presumably be correcting for minor inaccuracies in the system -- just like in a real plane.  Of course, this assumes that the inaccuracies in the platform motion are reasonably small.

Would you recommend 3 or 4 wheels/motors?  Given the slow speed and controlled conditions I'm thinking that 3 wheels may be adequate and would help to keep the cost down.

I found several sources for wheels that seem capable of handling the load.  Do you have any pointers for motors such as you described?  Is there anything special about mounting them for omnidirectional use?

--jon

[Admin]

Do the wheels really require encoders?  It seems that they could be run open loop since the pilot and bombardier working together would presumably be correcting for minor inaccuracies in the system -- just like in a real plane.  Of course, this assumes that the inaccuracies in the platform motion are reasonably small.

good point . . . forget the encoder thought . . .

Would you recommend 3 or 4 wheels/motors?  Given the slow speed and controlled conditions I'm thinking that 3 wheels may be adequate and would help to keep the cost down.

Keep your options open when you are looking for parts and computing loads/torques/speed. Depending on the situation, 4 might actually be cheaper. Four is also mathematically easier to compute and is more stable. But yea, 3 will be fine for you too. I'd look into both and then decide.

I found several sources for wheels that seem capable of handling the load.  Do you have any pointers for motors such as you described?  Is there anything special about mounting them for omnidirectional use?

Try this out:
http://www.societyofrobots.com/RMF_calculator.shtml
Omni-wheels have really poor efficiency as a lot of force is lost to friction. 3 wheel designs are even less efficient than 4. The advantage to four wheel design is each motor can be much weaker (since you have more motors and better efficiency), and hence you can buy smaller/cheaper/lighter motors.

Use highschool trig to calculate force/speed vectors to guess efficiency losses. (hint, not all wheels are pointing in the direction of travel)