Cybran's LASER Guided Project

[Cybran]

What sensor is used for laser detection? I seem to come across standard opto-interrupters and such most of the time, so I'm wondering if I could just use an opto-interrupter with a laser (red?) filter on the lens. However, If there is another "correct" sensor that I could buy that would be designed to detect laser light, it would be great if I could hear it.

Anyway, the design or whatever is going into a (hopefully) laser guided cart (dubbed T1ER, for odd reasons). The idea was to have a simple robotic cart act a "walking portfolio", as well as to carry my books and gear. While the original followed a infrared beacon placed in my backpack, there were several issues with this, most notably doors. If the T1ER was always behind me (following the beacon), then it would never come in front of me while I held the door open without resorting to simple remote control.
Eventually I hit upon this idea: the T1ER was designed after a unit from a RTS game, why not control it like a unit from a RTS game? So the laser came into play. The idea is to use a laser pointer or laser pointer cluster that reflects off the floor into 3 laser sensors, which it then uses to triangulate the laser beacon's position and distance. It will then move to where it thinks the beacon was. This way I can send it through doors while I hold them open, etc.

[SmAsH]

Wow, but isn't a laser beam a bit too narrow? Have you thought about using an ir led pulsing at a certain frequency and three "detectors" to pick up on only that frequency like a tv remote?

[Cybran]

Yes, that was the original idea. But I decided a laser reflection-detection...thing... would be the better way to go.

[waltr]

There are phototransistors and photodiodes that can detect a red LASER. Just check their data sheets for the light response wavelength. To use these you probably need some optical conditioning, lens, to define the sensing width and to amplify the reflected and scattered optical signal. A red filter will help to reduce false detections of ambient light as will modulation of the LASER. google "LASER communication" for information on modulating a LASER diode.

I won't say it can't work but it will take some experimentation.
good luck and have fun

[Cybran]

Edited Post:
Thanks.
The way I think this thing will work is (not written in minute detail) as follows: I shine a LASER pointer or cluster of LASER pointers onto the floor. 3 LASER sensors of some form mounted on the robot pick up the LASER light reflected off of the floor. It then translates this into distance between it and the light and the position of the light. I'm not sure whether I can do this with a time measurement sort of equation, where it measures the time between two pulses of LASER light it receives and then subtracts the actual time between the pulses for distance; or if I would have to try to measure the amount of Resistance from detection and translate that into distance.  You know, "This Voltage = This Distance" sort of thing.


Original post:
Thanks. Can anyone think of any major issues I might come up against? I'm trying to get this thing as finalized as I can before I head of to Radio Shack with my 50$.

[Soeren]

Hi,

Can anyone think of any major issues I might come up against?

That your idea, as outlined, won't work at all might be considered an issue by some   

I'm trying to get this thing as finalized as I can before I head of to Radio Shack with my 50$.

If you can explain what you wanna do in minute detail, you have a slim chance, if you can't, you don't.

The idea is to use a laser pointer or laser pointer cluster that reflects off the floor into 3 laser sensors, which it then uses to triangulate the laser beacon's position and distance. It will then move to where it thinks the beacon was. This way I can send it through doors while I hold them open, etc.

Assuming that both the LASER (or cluster of LASERs) is mounted stationary on your vehicle and the sensors are too, how is it gonna see any difference whether you're going through a door or not?

And why would it move at all? After all, the spot(s) will allways be at a certain distance in front of it (assuming a plain and level floor), so there's no triangulation to do (and you cannot do TOF or phase difference to triangulate with a simples setup anyway).

I'm not convinced that you thought this project through, but the real test is if you can explain it down to component level.

[chelmi]

I think I understand what the OP wants.

He want to hold a laser pointer in his hand and be able to point it to where he wants the bot to go.

If this is is want you meant, then the only solution I see is video processing on a laptop or maybe use one of these
camera with processing capabilities (CMUCam, Blackfin Camera,...)

Chelmi.

[hopslink]

The camera unit from the Wiimote might be well suited to this purpose, along with IR LEDs and laser.

[Cybran]

I think I understand what the OP wants.

He want to hold a laser pointer in his hand and be able to point it to where he wants the bot to go.



Chelmi.

Basically, yes. I figured that the three LASER sensors would pick up the LASER light reflected off of the floor. The LASER would pulse, giving it timing measurements to determine both distance and direction of where the LASER reflection pulsed. I'm figuring it could figure out the distance by measuring time from pulse one to pulse two, similar to how imaging equipment measures distance by measuring time between pulse sent and when the pulse was received. If there are any design issues, just tell me.

Apologies to Soeren, I fixed my above post to include my idea.

[Soeren]

Hi,

[...] I'm figuring it could figure out the distance by measuring time from pulse one to pulse two, similar to how imaging equipment measures distance by measuring time between pulse sent and when the pulse was received. If there are any design issues, just tell me.

There's a major design issue, which I allready mentioned.
You want to measure TOF (Time Of Flight) from say 0.5m to maybe 3m (or whatever).

The speed of light is 299,792,458 m/s and this mean that it will travel 1m in 3.34ns and a distance of 10cm (4") is 333ps (pico seconds).

You be the judge... Are you up for designing counter circuits (and input circuitry) for sub nano seconds to get a quite poor resolution of around 1' ?
To have a resolution of 10cm with just a hint of precision (~1/20), you need a signal conditioning and counter circuit capable of counting at a speed of around 20GHz (at that speed, each period and hence, the max. resolution, equates to 1.5cm, so you don't wanna drop too many counts).

[Cybran]

Well, that won't work then. Could I run this bot using voltage-type readings  (I.E. more distance= higher Resistance) from a photo-resistor like circuit?
Is there a better way to do this all together?

[Soeren]

Hi,

Could I run this bot using voltage-type readings  (I.E. more distance= higher Resistance) from a photo-resistor like circuit?

That might work under ideal circumstances in a lab, but in real life, there is way to much to throw it off - just imagine a ray of sun hitting the sensors from off center.

Is there a better way to do this all together?

Some of the options I can think of:
If you wanna use TOF, ultrasound (US) is a possibility (sound is "only" propagating at around 340 m/s). With a fairly simple setup, you could even try using the amplitude of the received US, but the receivers need to be as wide apart as possible.

Infrared (IR) with a cluster of emitters (placed on you) and once again, using the relative amplitude of the receivers.
A setup like used on photovores (angled receivers with a shade) might be possible too.

Radio, with a transmitter in your pocket - compare the AGC voltages (and make sure they're in their midrange at the typical distance they'll be used).

Camera, wear a bright colored jacket to make you stand out and make it easier to locate you.

Whatever method, use a Sharp distance sensor or similar to keep it from running into your heels.

[Cybran]

Thanks very much. Here's my solution to sunlight issue: Use a sensor that detects mostly red LASER light (which I was trying to find earlier) and/ or have the LASER pulse at maybe every 0.10 seconds. This way it can identify between sunlight (non-pulsing) and LASER light (pulsing). What are your thoughts on this?

Other wise,the US sounds like I good idea. However, I'll likely will be using this in a very crowded, noisy place. Would this interfere with the US sensor? And could I use this like my original idea; have the emitter in hand and point it where I wanted the robot to go, then turn it on for a few clicks? Thanks again.

[Soeren]

Hi,

Here's my solution to sunlight issue: Use a sensor that detects mostly red LASER light (which I was trying to find earlier) and/ or have the LASER pulse at maybe every 0.10 seconds. This way it can identify between sunlight (non-pulsing) and LASER light (pulsing). What are your thoughts on this?

Sunlight contains all colors, red included, so while color filtering will help a bit, it won't exclude sunlight.
Modulating the beam is a good idea, but you should use a much higher frequency, say 1kHz to 50kHz, to make it easier to discriminate and you can filter out DC and low frequency variations with just a capacitor.
Which exact frequency to use is a tradeoff. The higher the frequency (within the approximate range mentioned), the easier it is to filter, but the harder it is to get the needed gain from common op-amps. 10kHz is a good compromise here IMO.

Try using a 5mm red LED as the detector (akin to a photo diode), then you have the filter build in and LED's can be found with different angles, so you have a bit of built in optics for free.
If you're using a LED instead of a LASER and as such have freedom in color, go with a blue-green hue (if you can get it in a 5mm housing as well as a power device), as this color is more efficient. Use red if you want to be able to switch between LED and LASER.

Other wise,the US sounds like I good idea. However, I'll likely will be using this in a very crowded, noisy place. Would this interfere with the US sensor?

It might be a problem with false echoes. Noise could be a problem if eg. someone is rustling a bunch of keys or similar, as this produces harmonics way into the US spectrum.

And could I use this like my original idea; have the emitter in hand and point it where I wanted the robot to go, then turn it on for a few clicks?

It won't work by pointing it at a spot you want the robot to go to, then you'd be better off with the LASER, or better, a reasonably tight collimated flash light, perhaps with a modulated red 1W LED and the sensors as wide apart as possible.
I'd say, start with getting the direction bit working (i.e. to light the spot until the robot is there) and when that is working as wanted, first then think about triangulation and such.

Much of what you want to do, position determination, noise filtering and an overall rigidity of these functions will be testing your ability to code. I have seen very stable behaviour from crappy sensors, due to excellent coding and I have seen crappy code being applied to otherwise stable and perfect sensors with lousy results.

Starting off simple will give you a foundation to build on, so don't rush things, but get each stage to completion before moving on - then you won't fail (finding out something is not working isn't failing, it's getting experience).

Oh, and experiment your behind off while taking notes about what worked and what didn't (a notebook where you date your notes is gold), some experience has to be gained by actually doing rather than reading about it.

[Cybran]

Thanks again! This was incredibly helpful; I think you answered just about everything I need to know.

[Cybran]

I'm sorry to bump this thread up again, but I have more questions 
LEDS can detect light? Could you elaborate on this?
Also, how would I run a 9-volt PLC (an RCX 1.0, to be exact) off of a lawn mower battery (around 12 volts by 234 or so AMPS). I figured I could use a transformer to lower the Amperage, but that would increase the voltage, which (I think) would still burn up the PLC.
Finally, could I possibly re-name this topic to something a little more suitable?

[SmAsH]

LEDS can detect light? Could you elaborate on this?

Yes, led's can generate a small voltage if used in reverse, this voltage varies depending on the amount of light shining on

Also, how would I run a 9-volt PLC (an RCX 1.0, to be exact) off of a lawn mower battery (around 12 volts by 234 or so AMPS). I figured I could use a transformer to lower the Amperage, but that would increase the voltage, which (I think) would still burn up the PLC.

Well, you could chuck a 9v regulator in there? Amperage is not something that can be forced through somewhere, it has to be pulled, so only the amount of current being pull will come out of the battery.
You don't have to worry about frying your electronics because of too much amperage, spikes maybe.

Finally, could I possibly re-name this topic to something a little more suitable?

Are you able to edit the first post? It might only be senior members and moderators who can.

[Cybran]

Thanks Smash. You just explained one of the mysteries of electronics to . I didn't know that Amperage acted like that. I thought that if it it was too high, it would all rush through and burn up my stuff. Thanks for the other answers too!

[SmAsH]

Well, a batteries amperage is not addressed as so, it is addressed as mah "milli-ampre hours" or ah "amp hours"
Eg: a 10ah battery can supply 10A of power for one hour, 5A of power for two hours etc...
Of coarse there are many more factors for batteries but these are the main ones to know.