h a l f b a k e r yNot from concentrate.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Using a modulating (pulsed) laser pointer as a guide, this small toy car pursues the laser dot projected on the floor or as far as it can on vertical surfaces. Equipped with a simple 360° photosensor with which to see the target dot, the car has enough intelligence to back off from a wall or obstacle
and maneuver to continue pursuing the dot. The car slows as it nears the dot, speeds up as the dot moves away and stops if it cannot see the dot. Skillful drivers know where and how far in front of the moving car to play the light for deft control of the car direction and speed.
The behavior is like a cat chasing a laser dot but more dogged and without the inclination to grow bored of the pursuit.
When racing, each participating car is keyed to a different laser pulse pattern.
To help protect children's eyes, a tilt switch is included in the laser pointer that prevents operation of the laser when pointed at or above the horizontal position.
Not one of these: Photopopper Photovore BEAM robot.
http://www.hobbytron.net/photopopper.html
This link is to a simple autonomous sunlight seeking toy and not a laser target seeking toy. [bristolz, Oct 04 2004, last modified Oct 05 2004]
[link]
|
| |
send it under the fridge, send it under the fridge..yay - good game.. |
|
| |
With this you could launch a Dot Come company. |
|
| |
If you could get your robot to mate with a Roomba you might have something.
No, I mean sell videos of robots having sex... |
|
| |
I like this lots. Fly above the car in an open-sided helicopter with your laser pointer for maximum enjoyment - hours of fun, at least until the batteries run out (on the pointer or the car - hopefully the helicopter runs off a more significant power source...). |
|
| |
Hmm. Good point. Maybe a highly collimated regular LED light source could work. |
|
| |
I thought about that but the driver needs to be able to see the dot. |
|
| |
Maybe with two simple 360° photosensors instead of one. |
|
| |
"Like a cat....but more dogged" - I like your style.
|
|
| |
What happens if it catches the dot by the way? Perhaps the sport could be made more extreme by programming the robot to trace the laser back to source and then attack the pointer wielder. |
|
| |
[stupop], thanks! Haven't seen you for a while. |
|
| |
It stops moving when it catches the dot. |
|
| |
But only briefly, before it turns and attacks the pointer wielder. Right? |
|
| |
Bristolz], since the car can only follow on horizontal
surfaces, the laser device could have a tilt switch shut-off
mechanism at the horizontal position, assuming the laser
source is always above the plane the car is traveling on.
Certainly, it's not
fool proof but a start. |
|
| |
Also, I'd love to see this little thing do surpise stunts. It
might be more fun if it rotated just one optical sensor
attachment, giving nuances more like a real cat does
when looking
for the elusive "sprite". This would be where a "random
stunt generator" would enhance the fun. Perhaps
ultimately the stunts could build in magnitude, climaxing
in an explosion of parts...of course easily reasembled for
the next round of torment! + |
|
| |
The mercury switch in the pointer is a good idea (and added, thanks!) |
|
| |
Maybe I am making this much more difficult than it needs to be but I see some challenges with this idea. I'm imagining an oddball blend of proportional and linear control. As mentioned in the idea, the speed is controlled proportionally by how far in front of the car the dot lays but the steering is linear with the car turning to follow the dot directly. This mix of methods might make the thing difficult to drive but I can't imagine how else it would work. Ideally, the way the whole thing worked would just seem natural and easy but that's likely ignoring reality. |
|
| |
Because the dot would be wiggly from the operator's natural hand motion, there would probably need to be some sort of target dampening, or averaging, built into the steering control so that it wouldn't be "nervous." Maybe race models would have less dampening or maybe the dampening would be adjustable. Also, maybe the acceleration/deceleration proportionality curves would need to be adjustable, too. |
|
| |
To answer an earlier question posed by [half], I think that the distance the dot is from the car could be measured by calculating the angle of the dot in the sensor's field of view against the known height of the sensor from the floor. |
|
| |
Inspirations or clarifying thoughts welcome. |
|
| |
The latter could be done easily with triangulation and a
signal emitted from the pointer. But maybe you were
already assuming that? |
|
| |
As far as the steering/speed issue goes, I believe this
could be moderated just fine through the car's processor.
Turns could be attenuated to the personality of the user.
The car could make soft arching turns with a calm user
and faster harder turns following the assessed motions of
a
wild child. :b |
|
| |
[Tiger Lily]: The distance the dot is from the car I was assuming would be calculated at the car and not take into account the pointers position at all. |
|
| |
[UB] Yes, R/C cars are pretty amazing. The town I live in is R/C car crazy. |
|
| |
[disappearing half]: I meant spherical but now realize that even that isn't right. It needs to have maybe a 220° field of view horizontally and . . . I'll say, maybe 30-45° vertically. Rearward, it only needs crude detection just to signal to the car that it needs to turn around to reacquire the dot. |
|
| |
I see better now. I think I was still in abuse-prevention
mode and contemplating safety mechanisms to
incorporate into the pointer. |
|
| |
Would it be more economical, production wise, though
technologically more crude, to make the pointer the
source of directional processing through a series of tilt
switches? The laser, of course, would still be needed for
the driver's
reference. |
|
| |
Sorry for the disappearing act. Didn't think you saw that. I posted that at the same time I got your message which cleared it up indirectly. |
|
| |
I think that the blend of control methodologies that you suggest is good. To my way of thinking, detecting the location of the dot is the most complex part of this. For driving, the sophistication of real PID control might not be necessary. You might be able to get away with using a fairly basic scheme that just looks up motor speed and steering angle based on the 2 angles that represent the location of the dot. You'd likely be able to get away with a pretty lightweight processor and minimal circuitry if you're shooting for cheap, non-configurable, basic functionality. |
|
| |
I may be visualising this wrong, but wouldn't you have a problem with the car continually having to make time consuming three-point turns as the laser dot flicked to either side of it? Or would the car move with equal ease in any direction along the floor, without having to turn the body? Perhaps it could have mouse balls for wheels (as opposed to your regular cat which has mouse balls for dinner). |
|
| |
What would happen if you put it in a room full of mirrors? |
|
| |
I expect to see this baked in the next year... very good! + |
|
| |
[UB]: I think most indoor R/C cars can hit a wall with a lot of force and survive. My son has a little "Nikko" car that really moves and it has dented up our base mouldings many a time with no ill effects to the car (alas). |
|
| |
As for the mirrors [jivet...], I think trying to confuse it would be part of the fun. |
|
| |
[stupop]: I think it would only do a three-point after hitting an obstacle, like the aforementioned wall, otherwise it would just do turns. Once you got good with it you could even have it slide around turns or do "donuts." |
|
| |
So how would the car find the dot again? Nobodiy has
actually spelled that one out. Photo sensors and angles:
sounds nice, works in theory, but not so easy when you
try to build one. |
|
| |
I don't know, exactly. There are weapons systems that can do it so I know it's not impossible. What's your suggestion? |
|
| |
My suggestion is that this is hard to do cheap. Weapons
systems are expensive. Toy "laser" tag units have wide
angle IR sensors that pick up IR from a gun with a
narrowish beam IR emitter. I'm guessing you'll have to use
several narrow detectors, but I'm not sure any are
available that will pick up the light bouncing off the floor.
As a project for MIT, this seems pretty simple, but
manufacturing it and selling it for less than $40 is another
story. All this aside, I love the idea. |
|
| |
Of course, you could always just stick stray cats in them,
with a hole so they can see out. Special treadmills inside
would pick up the cat's intended direction and drive the
car accordingly. Or just Flintstone style. Would help
solve domestic cat overpopulation. |
|
| |
Did I mention that these sell for $450,000 each? |
|
| |
I used the example of a laser guided missile only to illustrate that this is technically feasible. Weapons systems have to be incredibly rugged and mission critical reliable whereas this toy doesn't. Also, the challenge of discriminating a modulated laser dot from the rest of the typical indoor environment is a much less difficult task than distinguishing a laser dot from several miles away in an outdoor, sunlit environment while traveling at 3500 mph. |
|
| |
With this toy I think that, even if all color info is ignored, just the difference in luminance and the laser modulation should make the target readily distinguishable to cheap vision systems of low resolution. Maybe a crummy 50¢ CCD from a webcam would be enough. |
|
| |
One for the price of two! |
|
| |
"Maybe a crummy 50¢ CCD from a webcam would be enough". With a sort of conical (maybe parabolic) mirror positioned, point down, above the CCD (facing up)? Also, a bandpass optical filter specific to the laser frequency would probably be helpful. Pixel location might map directly to distance and angle. Not really sure how fast a CCD can be scanned. Wonder if the modulation could be accurately detected with something so simple? |
|
| |
On steering "nervousness" & delta a: take cues from nature. Like one of the big cats on a long chase going through hike - trot - lope - run - terminal persuit modes, where a shorter chase might consist of just stalk - explode. Reactions become more extreme as distance shortens. |
|
| |
Ok, first we get rid of the horse (horseless carriage) and now Bris is wanting to get rid of the cat? I mean, come on, what other use is a cat if you can't drive it mad with a laser pointer? |
|
| |