fredm@media-lab.MEDIA.MIT.EDU (Fred Martin) (10/16/90)
robots must do is climb an inclinated plane. Thus, I'm soliciting all of you robot hackers for ideas on inclination sensors. One obvious such thing is a mercury switch. In fact, I plan on giving these to all of my student robot-builders. But, mercury switches tend to give you only one bit of data, so either you need a lot of them poised at strange angles, or you only get crude inclination data. Another idea is to have a pendulum-like rod mounted on a freely rotating shaft. At all times, the pendulum would swing to point down, and one could use an optical shaft encoder to measure the rotation of the shaft. That's about as far as we got before the ideas got pretty random. Any suggestions? I'd be particularly interested in a low-cost, commercially-available device (wouldn't we all?), although anything reasonably easy to construct would be wonderful. Thanks. - Fred Martin MIT Media Lab
cphoenix@csli.Stanford.EDU (Chris Phoenix) (10/16/90)
In article <3717@media-lab.MEDIA.MIT.EDU> fredm@media-lab.MEDIA.MIT.EDU (Fred Martin) writes: >robots must do is climb an inclinated plane. Thus, I'm soliciting >all of you robot hackers for ideas on inclination sensors. Use a free-swinging pendulum with a white bob, and beam infra-red rangefinders at it. I don't know how precise you want to get, but if you just want to know level/not level and which way to turn to point up a hill, this should work with as few as 3 IR transmitter/receiver pairs. You might want to feed the receiver outputs pairwise into comparators (op amps?) and then send those outputs into thresholders (D-A converters?) (So I'm not an EE.) You might also try mounting one transmitter on the pendulum, and receivers around it. This will give you some data. I have no idea if it will be enough. My experience with IR is that it has lousy range, about 12-16" with our set-up (not a problem here) and was rather non-linear over most of its range (I don't know if it's a problem). Our set-up used the output to drive a voltage-controlled oscillator, then counted the time it took to get N ticks off the oscillator. Good luck... -- War is a little naked kid running along a road and screaming because the napalm hurts so bad. War is young men in body bags -- theirs and ours. And the dying doesn't necessarily have anything to do with baseball, apple pie and the Grand Old Flag. -- Mike Royko
llw@corwin.eng.yale.edu (Louis L. Whitcomb) (10/16/90)
Greetings: So I suppose ring gyro lasers are out. Sigh. The pendulum works well when the vehicle's max acceleration is low enough not to disturb the reading. You can get low cost (~$5) 2 axis potentiometer joystick innards from some junk electronics houses (Edmund Scientific used to sell just the thing) and mount it upside down with a lever arm and bob attached for a 2-axis inclinometer. Cheap a/d and you are done. -Louis. -- ............................................................................ Louis L. Whitcomb llw@corwin.eng.yale.edu ph: (203) 432-4311 Yale Robotics Laboratory fx: (203) 432-7481 Department of Electrical Engineering, 1968 Yale Station, New Haven, CT 06520
martins@hplhasm.hpl.hp.com (Henrique Martins) (10/17/90)
There is a company selling a "digital carpenter's level". Look for it in your local hardware store. It replaces the bubble with a LCD readout that gives you the angle of the level. I may be able to dig out some information on it. Henrique ============================================================================== Henrique A. S. Martins Hewlett-Packard Laboratories 3500 Deer Creek Road Phone: (415)-857-2763 Palo Alto, CA 94304, USA Internet: martins@hplabs.hpl.hp.com ==============================================================================
barts@cyber.Eng.Sun.COM (Bart Smaalders) (10/17/90)
In article <3717@media-lab.MEDIA.MIT.EDU> fredm@media-lab.MEDIA.MIT.EDU (Fred Martin) writes: robots must do is climb an inclinated plane. Thus, I'm soliciting all of you robot hackers for ideas on inclination sensors. Use a radio shack cheapo joystick. Hang it upside down w/ a few ounces of weight epoxyed on the handle. - Bart (barts@Eng.Sun.Com) -- & @ @ @ @ @ @ @ _@__ | |
jjensen@convex.UUCP (James Jensen) (10/17/90)
In article <3717@media-lab.MEDIA.MIT.EDU> fredm@media-lab.MEDIA.MIT.EDU (Fred Martin) writes: >robots must do is climb an inclinated plane. Thus, I'm soliciting >all of you robot hackers for ideas on inclination sensors. > > - Fred Martin > MIT Media Lab How about some strain gauges with weights on them? Jim Jensen aka jjensen@convex.com
hbg6@citek.mcdphx.mot.com (10/17/90)
>In article <3717@media-lab.MEDIA.MIT.EDU> fredm@media-lab.MEDIA.MIT.EDU (Fred Martin) writes: >>robots must do is climb an inclinated plane. Thus, I'm soliciting >>all of you robot hackers for ideas on inclination sensors. > Set up an array of IR LED and detector pairs so that the pendulum only breaks one of the beams. Run the signals from each detector into a decimal to BCD converter and read this value directly into the computer. Personaly, I like the joystick idea best since it would provide both pitch and roll measurement but the IR LED approach gets rid of the A/D converter. HOWEVER: if you used an A/D converter with an 8 to 1 input mux you could measure pitch, roll, battery voltage, radio link signal strength, drive motor temperature, and three other analog parameters with the one chip. John ..................................................................... reply to 'from' address; hbg6@citek.mcdphx.mot.com NOT the 'sender' line address! Someday my sysadm will decide this is a 'real' problem. :-) ..................................................................... All opinions expressed are mine and not Motorolas, their loss. .....................................................................
hollombe@ttidca.TTI.COM (The Polymath) (10/18/90)
In article <3717@media-lab.MEDIA.MIT.EDU> fredm@media-lab.MEDIA.MIT.EDU (Fred Martin) writes: }... I'm soliciting }all of you robot hackers for ideas on inclination sensors. }Another idea is to have a pendulum-like rod mounted on a freely }rotating shaft. At all times, the pendulum would swing to point down, }and one could use an optical shaft encoder to measure the rotation of }the shaft. You'll need something to damp the pendulum. Otherwise, acceleration will set it swinging and cause erratic readings. I suggest a very simple pneumatic piston-and-cylinder damper. You could probably build one from scrap. I don't know what cheap optical encoders cost these days (the ones we use cost about $800), but linear pots are cheap and ADCs are off-the-shelf chips. -- The Polymath (aka: Jerry Hollombe, M.A., CDP, aka: hollombe@ttidca.tti.com) Head Robot Wrangler at Citicorp(+)TTI Illegitimis non 3100 Ocean Park Blvd. (213) 450-9111, x2483 Carborundum Santa Monica, CA 90405 {csun | philabs | psivax}!ttidca!hollombe
mgobbi@cs.ubc.ca (Mike Gobbi) (10/18/90)
A couple ideas for detecting inclination immediately pop to mind. The first (and easiest) method would be similar to your "optical shaft encoder" idea. Take a variable resistor (the rotating kind) attached to the pendulum. A voltage is applied through the resistor and then fed into an Analog to Digital Converter. For instance, using a 100K resistor (set so that with the pendulum vertically downward resistance = 50K increasing counterclockwise) and an opposing resister (at 50K, increasing clockwise), connect the leads from the voltage source to one resister. From the other end of the resister connect both to the second resistor and the ADC. Finally, connect the free end of the second resistor to ground. What we have is a voltage divider, which passes (given a 12-volt supply) 12 volts when the device is rotated 180 degrees forward and 0 volts when rotated 180 degrees backward. Once fed through (say) an 8-bit ADC you get out values from 0 to 255, with 128 being horizontal. ADC * 360 / 256 - 180 = angle robot is inclined FORWARD. <<I'm not an electronics person, so I realize that this explanation is rather greek. Forgive me.>> -- __ /..\ In quest of knowledge.... --mm--mm-- Mike Gobbi
wiml@milton.u.washington.edu (William Lewis) (10/18/90)
In article <3717@media-lab.MEDIA.MIT.EDU> fredm@media-lab.MEDIA.MIT.EDU (Fred Martin) writes: >robots must do is climb an inclinated plane. Thus, I'm soliciting >all of you robot hackers for ideas on inclination sensors. Disclaimer: I've never built one of the following, so it may or may not be feasible. But it's got to be easier than attaching IR rangefinders to a pendulum bob =8) I have, several places, seen designs inclinometers (is that a word?) that work by having two C-shaped parallel plates in a thin container, partially full with liquid. As the angle of the container changes, and therefore the angle of the liquid's surface with the plates, the electrical characteristics will change, depending on the liquid you use. You could use a conductor and measure resistance, or a dielectric and measure capacitance (say, an oscillator controlled by the cpacitance on the plates could run a counter, which is cleared and read at timed intervals to make a sort of frequency counter). The only problem I can think of offhand is sloshing, but you'd get swinging effects in a pendulum, too. Choosing the right liquid (whatever it may be) should reduce that. A diagram: _ | | | | <-- plates / / . . . / / . . <--liquid level _____/ / |______/ >One obvious such thing is a mercury switch. In fact, I plan on giving >these to all of my student robot-builders. But, mercury switches tend >to give you only one bit of data, so either you need a lot of them >poised at strange angles, or you only get crude inclination data. Or, one contact could be resistive (carbon rods?) and the mercury could cover the contact to different heights depending on the angle, making a measurable changing resistance. Same idea as above, except the contact is being resistive instead of the liquid... -- wiml@milton.acs.washington.edu Seattle, Washington (William Lewis) | 47 41' 15" N 122 42' 58" W "These 2 cents will cost the net thousands upon thousands of dollars to send everywhere. Are you sure you want to do this?"
smithc@motcid.UUCP (Chris Smith) (10/18/90)
Tried to e-mail...oh well. In article <3717@media-lab.MEDIA.MIT.EDU> you write: >robots must do is climb an inclinated plane. Thus, I'm soliciting >all of you robot hackers for ideas on inclination sensors. > >That's about as far as we got before the ideas got pretty random. Any >suggestions? I'd be particularly interested in a low-cost, >commercially-available device (wouldn't we all?), although anything >reasonably easy to construct would be wonderful. Disclaimer: I'm a software engineer. With that out of the way, here's my suggestion: Could you use an artificial horizon from an airplane. It seems to me that some units would provide encoded readings for auto-pilots, the parts should be available "off the shelf", and they should also provide inclination in two dimensions, giving you the ability to determine the overall attitude of the platform rather than just front/back inclination. I have no idea what the cost of such a unit is, but it could save you development time. Just an idea. -- o-------------------------------------------------------------------------o o Chris Smith @ Motorola Inc. uunet!motcid!smithc o o "Our great computers fill these hallowed halls..." - N. Peart o o---------------------My thoughts, not my employer's----------------------o -- o-------------------------------------------------------------------------o o Chris Smith @ Motorola Inc. uunet!motcid!smithc o o "Our great computers fill these hallowed halls..." o o---------------------My thoughts, not my employer's----------------------o
gerry@frc2.frc.ri.cmu.edu (Gerry Roston) (10/18/90)
The answer to the orignal query depends on the following: whether or or not there is a deadline to meet and whether or not there is some level of robustness/accuracy required. If the answer to both questions is no, then the idead previously suggested will be fine, and you may learn something from your failures. However, if this is a "real" robot, you can not afford the engineering time required to set up such a device. Furthermore, the acuracy and usability will be questionable. If your need for angle sensing is low bandwidth, say < 1 Hz, a standard clinometer from Sperry or Schavitz is probably the answer. If you need high precision/high bandwidth, the answer would be a IRS (Inertial Reference System). Several companies produce these, but they are outrageously expensive. Gerry (the person who is developing the dead-reckoning for the CMU Ambler) -- gerry roston, field robotics center robotics institute, carnegie mellon university pittsburgh, pennsylvania, 15213 (412) 268-6557 gerry@cs.cmu.edu
hbg6@citek.mcdphx.mot.com (10/19/90)
In article <107247@convex.convex.com> jjensen@convex.UUCP (James Jensen) writes: >In article <3717@media-lab.MEDIA.MIT.EDU> fredm@media-lab.MEDIA.MIT.EDU (Fred Martin) writes: >>robots must do is climb an inclinated plane. Thus, I'm soliciting >>all of you robot hackers for ideas on inclination sensors. > >How about some strain gauges with weights on them? > I guess that if you had a weight, hanging from two strain gauges, on two links running at 45 degrees from the weight to the gauges it would work. The ratio of the two gauge readings would indicate inclination. I think you could even resolve 360 degrees using rigid links. ======= ======= gauges [] \ / [] \ / \links/ \ / +-----+ | | weight | | +-----+ Can you get strain guages sensitive enough that you could use a very light weight? I would think that acceleration with a relatively heavy weight could destroy the strain guage. (I guess I'l have to get out the Omega strain guage book :-) ) John ..................................................................... reply to 'from' address; hbg6@citek.mcdphx.mot.com NOT the 'sender' line address! Someday my sysadm will decide this is a 'real' problem. :-) ..................................................................... All opinions expressed are mine and not Motorolas, their loss. .....................................................................
nagle@well.sf.ca.us (John Nagle) (10/19/90)
An excellent 2D inclination sensor is manufactured by Etak, Inc, in Menlo Park, California, but it is not generally available for retail sale. They will sometimes sell them to researchers, along with a 2-axis rate gyro and appropriate electronics, for about $200 each. The device is normally used as part of the Etak vehicle navigation system. The inclinometer is a small cylinder of liquid, about 15mm across, equipped with four capacitance level sensing plates. The liquid has been carefully chosen to have the right poise to make the device relatively insensitive to vibration. Etak has the electronics which causes the output to be provided as a variable audio frequency, for convenience in digital handling. I have one of these, and this is definitely the way to go. The whole unit runs on 5VDC. John Nagle
rogers@ux.acs.umn.edu (Brynn Rogers) (10/28/90)
Why aren't there any used/surplus gryos available from airplanes that would be cheap enough for robots, but maybe not accurate enough for planes anymore? A friend of mine had a Gryo for his RC Helecopter, it was less than $100 or so, but it was used only for the tail rotor (it was mechanically linked to it). Are there any even moderatly inexpensive gyros? [ Honeywell had some Ring Laser Gyros that were small enough for air to air missles, and big ones that give inertal positioning to trident subs, but the pricing would be far out of reach for anyone but the military. Though Boeing 757 and 767 jets use commercial ones] [[ I used to work at Honeywell S&RC ]] Brynn rogers@ux.acs.umn.edu