[ExI] robot riding a motorcycle

spike spike66 at att.net
Sat Oct 31 23:55:28 UTC 2015


Subject: [Bulk] Re: [ExI] [Bulk] Robot riding a motorcycle

On 29 October 2015 at 16:18, spike  wrote:
<snip>
>>>... OK then, now I am in territory I don't know nearly as well, the
sensors...

>>...Hmmm --  I think there is a bit more involved than a few sensors and
processors.  :)... BillK

>...Well sure, but using the Raspberry class processor in a Lego Mindstorm
robot,  you can make a robot balance on two wheels side by side (like a
Segway.)  I have a vague notion the control system needed to ride a
motorcycle isn't all that difficult...spike 


OK I have been pondering this notion and realize that even fitting the
ro-bike with visual sensors would be overkill.  I am thinking of a race to
the bottom for cost; here's what I have so far.

I had in mind those modern racy road bikes that can go over a quarter the
speed of sound, but that wouldn't be necessary at all, or even desirable for
an indoor 1/8 mile.  The bikes they use for that are simple dirt bikes with
street tires, and there is no advantage to having a modern one: any bike
built in the last 40 years will work fine.  Here are some examples:

https://www.google.com/search?q=dirt+bike+with+street+tires&biw=1008&bih=642
&tbm=isch&imgil=kMhnQJH_DGNrFM%253A%253B8sbheUUFyUme8M%253Bhttp%25253A%25252
F%25252Fwww.mychinamoto.com%25252Fforums%25252Fshowthread.php%25253F1203-Chi
naV-s-Galaxy-XTR250-amp-TGR250!&source=iu&pf=m&fir=kMhnQJH_DGNrFM%253A%252C8
sbheUUFyUme8M%252C_&usg=__KqcXdktZ0wBg2wnONPqpY_1q394%3D&ved=0CCYQyjdqFQoTCP
f5idnq7cgCFRT_YwodC8oL6A&ei=tEo1VvfPBZT-jwOLlK_ADg#imgrc=kMhnQJH_DGNrFM%3A&u
sg=__KqcXdktZ0wBg2wnONPqpY_1q394%3D

Last time I was over at the San Jose Indoor Eighth, I noticed plenty of old
junkyard dogs in the chase.  Since these things are light, the distance is
short and the top speed is not very high (a typical interstate freeway speed
is as fast as they ever get going on that short track) we could use ordinary
street tires rather than those spendy racing tires.  That feature whacks off
a lot of cost, and even reduces the cost of the actuators, since they can be
smaller lighter devices, turning a much lighter front end at lower speeds.
With those kinds of forces and this short a race, we could use pneumatic
actuators running on a couple of fire-extinguisher bottles with compressed
air at 2k PSI.

But here's the cool part: it occurred to me we wouldn't need to do all that
image processing, because we could set up the two endpoints as short-range
radio transmitters.  All we need to let the ro-bike figure out where it is
on the track is compare the direction and relative strength of the two
signals, cool!  That would be waaaay easier to do, and probably cheaper,
than optical processing (and I have a better notion of how to get that
done.)

So now we are down to any dirt bike regardless of how hard it's been ridden,
used is fine, a set of ordinary cheapy street tires (or the sticky compound
Dunlops would work fine) linear actuators for the shifter, clutch, steering
and front brake, rotational actuators for throttle and if we really want to
get into the poor-man's racing aspect, just make it a no-shift rule: run the
whole race in second gear, which eliminates two actuators and a buttload of
code, probably two MEMS laser gyros, a dual-frequency radio receiver of some
sort and a microprocessor. 

Reasoning: an open-class dirt bike can easily loft the front wheel in second
gear with me on the back, so it could do it even easier with no one on
board.  These bikes can go faster in second gear than the short track can
manage.  The point at which the front wheel comes up is the best the bike
can accelerate anyway, so... run the whole race in second gear, cool.

Here's an even more fun part.  Take an expert motorcycle rider who has never
tried any 1/8th mile track, observe that it is not at all obvious what path
to take to get around those two cones in the least amount of time.  Does she
slow down more and make a tighter turn around the cones, or does she keep
herspeed up and accept the longer path?  It takes a lot of practice to see
what is the optimal path, so we could let a ro-bike experiment and discover
what technique works best.  This would be an example of a robot playing a
human game and teaching the carbon units the best way.  Cool!

Everything I have imagined here could be done with a 4 digit number of bucks
and a lot of free volunteer time, which brings up a new question.  If this
game is this cheap and technologically not so terribly difficult, someone
somewhere should have thought of it and is doing it already, a tech company,
a university mechanical engineering class or a private club.  Who and where?
If not, we need to talk to Stanford and have them get on this before
Berkeley comes along and whoops their Cardinal asses, then go to Berkeley
with the same story.  We could have a cool robike showdown, right here at
the San Jose Fairgrounds, and you know the locals will give a Hamilton to
see that.  The arena holds enough proles we could even make money on it,
which makes the whole notion morally justifiable: it disguises a fun
crazy-ass notion as an ordinary perfectly understandable get-rich-quick
scheme.

spike

  




More information about the extropy-chat mailing list