Building a LADAR

[Marks256]

How would one go about making a ladar? I know that ladar works by pointing an oscillating lazer at a spining mirror, and then recording the time it takes for the beem to return.

Really my initial question is, what would be used as a reciever? Ok, so maybe how does a ladar work?

 

[3iMaJ]

Quick question: Is there an application you wish to use a LIDAR for which a standard radar would be inadequate. I ask this because it is much simplier to build a standard radar.

 

[evandude]

Think for just a second about the speed of light - about 300 million meters per second. Now think about what kind of delay times you would be measuring here. That should be enough to convince you that this is not a project suitable for a hobbyist

Now, if it's a matter of pure curiosity, that's a different story.

 

[3iMaJ]

Its the speed of light applies to a radar as well.

 

[evandude]

Its the speed of light applies to a radar as well.

True, but in radar applications where you're looking at the doppler shift of the returned signals, you end up dealing with a much more manageable signal - as evidenced by the fact that you can buy a (basic) working radar speed gun as a children's toy for about $30. With LIDAR (aka LADAR, to the military), you're stuck with much more difficult timing tasks.

 

[3iMaJ]

Lidar can work just as a Radar would, one isn't inherently more difficult to build than the other, they just have different applications. Pulse compression can be used in Lidar just as it would in Radar, so it would again just come down to measuring doppler shift for ranging information.

P.S. Depending on how you build your radar (IE pulse compression or not) it can just come down to measuring the delay between transmit pulse and echo pulse and with processors the speed they are, the speed of light is a rather trivial number to deal with (same with time delay). Time resolution is very easy to buy.

 

[Marks256]

I guess lidar would work too, but i would like to stick with LADAR for now...

I guess you could say that it is just pure curiosity. Maybe make a few projects with it, but nothing major.

I would like to stay away from radar, because it uses radio, and i need to find the distance of an EXACT point. It really doesn't have to be all that accurate, maybe within a 1/2" (~1.27cm) or so...

Really i would just like to see if i could make one... I saw one going on ebay for $5000usd, and almost fainted....

 

[Nigel Goodwin]

I guess lidar would work too, but i would like to stick with LADAR for now...

I guess you could say that it is just pure curiosity. Maybe make a few projects with it, but nothing major.

I would like to stay away from radar, because it uses radio, and i need to find the distance of an EXACT point. It really doesn't have to be all that accurate, maybe within a 1/2" (~1.27cm) or so...

Really i would just like to see if i could make one... I saw one going on ebay for $5000usd, and almost fainted....

Perhaps you were planning using a Z80 to do the timing?

 

[evandude]

I guess lidar would work too, but i would like to stick with LADAR for now...

I guess you could say that it is just pure curiosity. Maybe make a few projects with it, but nothing major.

I would like to stay away from radar, because it uses radio, and i need to find the distance of an EXACT point. It really doesn't have to be all that accurate, maybe within a 1/2" (~1.27cm) or so...

Really i would just like to see if i could make one... I saw one going on ebay for $5000usd, and almost fainted....

LIDAR and LADAR are just different acronyms for the same thing...

I still think you're dramatically underestimating the difficulty here... For one thing, do you have a lot of experience with RF/microwave circuitry?
Just as some food for thought, here's a research paper abstract I came across:
"Practical low-cost high-range-resolution ladar"
http://adsabs.harvard.edu/abs/1995SPIE.2472..118S

This paper treats a practical adaptation of frequency modulation (FM) radar ranging principles to an incoherent laser radar (ladar). In the simplest sense, the ladar's laser transmitter output is amplitude modulated with a radio-frequency subcarrier which itself is linearly frequency modulated. The subcarrier signal may have a start frequency in the tens to low hundreds of megahertz and stop frequency in the hundreds of megahertz to low gigahertz. The difference between the start and stop frequency, (Delta) F, is chosen to establish the desired range resolution,(Delta) R, according to usual equation from FM radar theory, (Delta) R equals c/(2(Delta) F), where c is the velocity of light. The target-reflected light is incoherently detected with a photodiode and converted into a voltage waveform. This waveform is then mixed with an undelayed sample of the original modulation waveform. The output of the mixer is processed to remove `self-clutter' that is commonly generated in FM ranging systems and obscures the true target signals. The clutter-free waveform is then processed coherently using the discrete Fourier transform to recover target amplitude and range. A breadboard of the ladar architecture was developed around a 30-mW GaAlAs diode laser operating at 830 nm. Imagery and range responses obtained show that the theoretical range resolution of 0.25 m was attained for a (Delta) F of 600 MHz. Embodiments of this ladar are likely to be practical and economical for both military and commercial applications because low-cost continuous wave laser diodes are used, coherent optical mixing is not required, and the post- mixing processor bandwidth is low.

So, they're talking about range resolution of 25 cm (20 times worse than the 1.27cm which you considered "not that accurate") and they're dealing with a frequency range that is hundreds of MHz. Needless to say, such a system is not exactly going to be practical for a hobbyist to build, especially one without at LEAST an undergraduate college education in EE under their belt...

Not saying it's not something you should look into if you're interested, but you seem to have some VERY unrealistic goals... Especially given the knowledge that commercial units sell for thousands of dollars...

 

[Marks256]

Perhaps you were planning using a Z80 to do the timing?

I was thinking more along the lines of an Athlon or P4 running XP pro along with the .net 3.0 (and if i am feeling good enough, i just may give it 384MB of ram...

you seem to have some VERY unrealistic goals...

Well obviously! If i didn't chances are i wouldn't be part of this forum...




What else is there that can be used to find the distance of ONE point. I don't want any of this "[hardware sending robot data] Hey, guess what? There is an object about 20cm from this position! Wanna guess where it is so we don't hit it?"...

 

[dknguyen]

I don't think you can make a EM time-of-flight anything unless you can make the timing circuitry yourself, especially if you want to work in the meter range. You need dedicated hardware based timing circuitry for this kind of thing and I don't think that a software-executing device like an Athlon, no matter how fast it sounds would be able to cut it.

I don't think an Athlon is fast enough to do the time-of-flight work because even though 3Ghz sounds fast, it's really not especially considering the Athlon must run through at least a few routines to do anything (it's a software based device and software always slows everything down)- and that's if it's programmed in assembler and is dedicated to just timing. But when you're running something like Windows XP on it, it will run even slower. Just to give you an idea, let's ignore all the stuff I just said and assume the Athlon can somehow time something down to 1/3 billionth of a second. It will only have a resolution of 10cm. That's great on the km range, but since you're probably working on the meter range (probably up to a max of 10m) that's pretty horrible considering the difficulty of building the device in the first place. 10cm accuracy in 1m is abysmal considering you can get 1cm accuracy with something that costs as little as $5 with a beamwidth of 15 degrees.

If you want to redirect the beam to image then you need all sorts of precision rotating mirrors.

You could buy a laser scanner for $3000 that works up to 4m, but it's pretty unfeasible to build it yourself.
https://www.acroname.com/robotics/parts/R283-HOKUYO-LASER1.html

Anything else costs $10,000 or more I've found and in some cases doesn't scan without costing much more although the range is usually km rather than 4m.

If you really want to know where something is, just use one of the Sharp IR sensors with a pan head. The closer you get to the target the smaller the beam divergence and the greater accuracy you have which isn't too bad since when something is farther away you don't need to know exactly where it is until you get closer.

 

[3iMaJ]

I would like to stay away from radar, because it uses radio, and i need to find the distance of an EXACT point. It really doesn't have to be all that accurate, maybe within a 1/2" (~1.27cm) or so...

I'm not sure you have a good understanding of practical radar (lidar/ladar follows the same rules). Radar can resolve targets as close together as lidar, the only price you pay is with bandwidth. Consider for a second a synthetic aperature radar system (SAR), a lot of times employed from space. These types of systems can create resolution cells from space that are smaller than 15m by 15m. In fact the actual distance from your target is irrelevant to a SAR system, SNR is the governing factor. So a resolution of 1/2 an inch isn't terribly difficult to do with a radar system, it only requires a bandwidth of 12 GHZ . That would also be the minimum distance that your radar/lidar system could resolve different targets.

So in summary, lidar and radar have different uses, but when it comes down to target resolution and ranging resolution the rules are identical. Maxwell doesn't care what frequency you operate at, its all EM to him.

P.S. if you were curious what equation I used to calculate the BW with its R_res = c/(2*Bw)

Good luck!

 

[evandude]

What else is there that can be used to find the distance of ONE point. I don't want any of this "[hardware sending robot data] Hey, guess what? There is an object about 20cm from this position! Wanna guess where it is so we don't hit it?"...

You also seem to have an exaggerated idea about how accurate a robot's sensors need to be - plenty of people have made plenty of good robots without being able to measure every distance at every possible angle with 1/2" accuracy. Even if you could do it, your robot would never get anywhere, it would spend all its time scanning everything in front of it.

I don't know about you, but as a human being, I don't know the exact distance to any given point in the room - I probably could barely even guess it to within the nearest foot for objects more than 5 or 6 feet away. And yet, I think I do a pretty good job of avoiding obstacles.

Much like that, as dknguyen pointed out, who cares about exact distances at precise points when things are far away? a 15 degree beam width may seem like more than you want, but if you know where an object is to +/- 15 degrees, you can still do a reasonable job avoiding it - and if you aren't doing a good enough job, as you get closer to it, your positional accuracy will only be improving.

 

[dknguyen]

You can also get multi-beam IR sensors from Sharp now. THey have a 5 degree beamwidth (with 5 sensors beside each other) so you can actually "scroll" in 5 degree increments across a 25 degree range. They also cost about 5x the single sharp IR sensors! But $50 is pretty good all things considered- or you can use a single beam unit with a pan unit.

"Analog" are the versions you want.
https://www.sharpsma.com/Page.aspx/...b-a14f-92e2bd34f1c7/Distance_and_Air_Sensors/

These are the 3 multi-beam unit numbers available so you don't have to hunt through:

GP2Y3A001K0F 30 mA ANALOG 4cm to 30cm 16.5 ms
GP2Y3A002K0F 30 mA ANALOG 20cm to 150cm 16.5 ms
GP2Y3A003K0F 30 mA ANALOG 40cm to 300cm

The third unit has a range that is double the range of any other sensor (including single beam). Other sensors have a max range of 20-150cm. Note the minimum range increases as the maximum range does.

 

[Marks256]

All this information, and none of you noticed that i spelt laser wrong...

plenty of people have made plenty of good robots without being able to measure every distance at every possible angle with 1/2" accuracy. Even if you could do it, your robot would never get anywhere, it would spend all its time scanning everything in front of it.

Yes, but i want the accuracy...



This forum is really odd. We spend more time explaining how complex things are other than answering the initial question. Chances are VERY good that i will never even attempt to build one, i am just curious how it WOULD be put together...


never mind.

 

[dknguyen]

Well it is a lot easier to explain why things are too complicated than to explain how to build complicated things! If it could be explained on here then it wouldn't be complicated.

If you want to know how it's put together it's a transmitter, detector, timing circuitry, and spinning mirror with position feedback to tell where the laser is being directed.

(If you want accuracy you should still go with the Sharp IR since it's accurate to 1cm and dirt cheap. The LADAR you are planning would be far harder and expensive and only an accuracy of 10cm)

 

[Marks256]

Well it is a lot easier to explain why things are too complicated than to explain how to build complicated things! If it could be explained on here then it wouldn't be complicated.

You can't tell me that talking about making expensive, complicated, and downright stupid projects isn't fun. I would think any geeky person would jump on that!

(If you want accuracy you should still go with the Sharp IR since it's accurate to 1cm and dirt cheap. The LADAR you are planning would be far harder and expensive and only an accuracy of 10cm)

But i don't think that a ir sensor could detect object upto 40 feet away in the blairing sun now, can it?

 

[3iMaJ]

What is it that you're exactly interested in? The physics, signal processing, engineering, implementation...etc?

 

[Marks256]

mostly the physics and the engineering.

 

[dknguyen]

You can't tell me that talking about making expensive, complicated, and downright stupid projects isn't fun. I would think any geeky person would jump on that!

Discussing it is- typing a long one-way conversation about how to do it is not.

But i don't think that a ir sensor could detect object upto 40 feet away in the blairing sun now, can it?

Perhaps not, but what are the chances that there won't be an object 5 meters away? 10 meters away? Consider that when you are walking through a room (or outside) you don't really pay attention (as in start to avoid them) until they are within 20 feet or so.

You might start to avoid really large obstacles like walls from much farther than 20 feet, but such large obstacles are easy to detect from far away and you don't need to know their pin point location since they are so big (so sonar works). But for a small obstacle like a wet floor sign, don't try and tell me that you start adjusting your path to avoid it from 40 feet away. Chances are you'd deviate from your initial path within 5 feet of the sign.

And what are the chances you actually need to know where something is down to the sub-degree resolution when it's 40 feet away? To understand what I am talking about, stand somewhere, and point at an obstacle that you have to avoid that is some distance away. Now walk towards it WITHOUT adjusting your finger's aim at the obstacle (since as you get closer you can get more accurate information).

You certainly had no trouble avoid the obstacle even though your initial aim from 40 feet away was off. At a far distance you didn't need to really know where the obstacle was, but as you got closer it started to matter more but it also got a lot easier to get more accurate heading of where the obstacle was. In fact, chances are that you would miss the obstacle anyways if you started walking blindly toward it in a straight line from 40 feet- even moreso if you had wheels for legs.

So unless you only want to take one reading of an obstacle and then avoid it without taking any further readings such accuracy generally y isn't worth the effort or cost. But that's easy to overcome- just continue to take readings as you travel and get closer to the obstacle.

I know where you are coming from though- it's really tempting to find one be-all-end-all laser imager (a.k.a scanning laser rangefinder) that is immune to sunlight, wind, air temperature/pressure, target colour, and target softness so you can detect any and all obstacles within a 1cm to 1km radius with pinpoint accuracy. But it's a lot of money spent on something needless.