Subject:
|
Re: moving in a straight line with two motors?
|
Newsgroups:
|
lugnet.robotics
|
Date:
|
Sat, 14 Nov 1998 03:18:44 GMT
|
Viewed:
|
2205 times
|
| |
|
|
Daniel Miller wrote in message ...
> 1) You've been playing too much Carmageddon.
What?
>
> 2) Control theory is relatively new (it paces the development of
> computers). There are quite a few good textbooks available, but a
> professor is a much greater help. Familiarity with calculus and dynamics
> is required.
>
> Not to say I'm an expert (I got a "c" in AAE464 and I was grateful) but it
> should be pretty possible to implement proportional and derivative
> control. Integral control may be harder... but it would be needed to
> eliminate steady-state error (i.e. keep it straight!). Lead-lag control
> would be damned impressive.
This is all fine and well, but I was thinking about something
a little less complex. The RCX doesn't have the power to perform
image processing and sophisticated feedback control. What I was
trying to impress on designers was the fact that bots can't be
depended on to "stay the path" without providing some type of
reaction control. Furthermore, this reaction control doesn't
doesn't have to be complex. For example, while the angle sensor
might be useful for steering, the bot is going to have to
recalibrate itself on occasion against a "beacon". For example,
maybe a light which the light sensor can be home in on. Another
simple example consists of these bots that follow a line. These
bots close the loop using a light sensor to determine when the
bot is no longer going the right direction.
Unfortunately, I find that the RCX lacks the sensor inputs to
really build something truly sophisticated. After all, Fred
Martin's Crickets have four sensor inputs, which is a far cry
better than three. In fact, I think it would be ideal if the
RCX had four of each port. Most of the applications I have
played with left me wanting for one more of each port at one
time or another. I have been following the work people have
been doing to expand the port capability. Unfortunately, I
do not have the time to build these devices. Who wants to
build me one and sell it to me (of course there would have
to be some kind of guarantee that the thing works)?
>
> Here's an exercise for someone with an RCX: Create a chassis that
> consists of one drive motor to the wheels and an angle sensor geared to a
> 2-d inverted pendulum on top. Write a program that moves the chassis to
> keep the pendulum straight vertical.
>
> And finally: It's quite possible to create a car that drives itself. It
> still needs human control for things like parking. Same goes for
> aircraft; the main purpose of the pilot in a current production A320 is to
> park the airplane and make the passengers feel better en route.
>
> There are some cases where the control system has failed. On a flight
> from Warsaw to New York, the pilot told the passengers they could see
> Ireland out the starboard side. The ensuing rush to the starboard
> windows was too much for the autopilot, which disengaged. Later, the
> cause of the crash was determined to be a problem with the flight
> computer; it seems there were too many Poles on the right hand plane.
> Sorry, controls engineering joke.
>
> Another: "Welcome passengers to the first flight of the Boeing 2707, the
> world's first totally robotically controlled airplane. Please do not be
> alarmed; the systems have been tested to the tightest imaginable
> tolerances and we guarantee that nothing can go wrong go wrong go wrong go
> wrong..."
>
> Daniel "Dan'l" Miller Senior, School of Aeronautics and
> danielmi@expert.cc.purdue.edu Astronautics, Purdue, Indiana
> danielmi@cartoon.ecn.purdue.edu
>
|
|
Message has 1 Reply:
 | | Re: moving in a straight line with two motors?
|
| (...) Image processing? Proportional control pushes back proportional to deflection (i.e. it simulates a spring). Derivative control pushes back proportional to rate of deflection (i.e. it simulates dynamic friction). Integral control pushes back (...) (26 years ago, 14-Nov-98, to lugnet.robotics)
|
Message is in Reply To:
| | Re: moving in a straight line with two motors?
|
| (...) 1) You've been playing too much Carmageddon. 2) Control theory is relatively new (it paces the development of computers). There are quite a few good textbooks available, but a professor is a much greater help. Familiarity with calculus and (...) (26 years ago, 14-Nov-98, to lugnet.robotics)
|
12 Messages in This Thread:
  
 
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
