Level/Flatness using laser/webcam/python

This technique is new to me.

It all seems fairly straightforward

You tube Video https://www.youtube.com/watch?v=hnHjrz_inQU

A brief search on here [laser] returned no similar thread.

Thoughts?

Regards.

Frank C.

Bryan and a heap of contributors are doing a great job. on Discord

The fundamental problem they face is that the LASER beam is affected by thermal gradients, turbulence and convection currents in the air. (Even having somebody stainding near it is enough to cause a beam deviation of several hundred microns over a few metres).

I designed & built a LASER guided cutting tool in the late 1980s (using a quadrant photodiode as the target). It was easily capable of micron level resolution, however the beam deviation due to air currents within the machine enclosure rendered it useless.

Andy G interesting from 1st hand experience.

We used to use an older version of this Renishaw laser machine tool calibration system :-

https://www.renishaw.com/en/xk10-alignment-laser-system--44377

It works in a similar way but bounces the laser back with mirrors to use interferometry to calculate errors to 4 or 5 decimal places. The requirement to avoid draughts and keeping temperature stable was the same.

Ah, but did your cutting tool or KW56's Renishaw calibrator work in the same way as the Youtube leveller? I think a cutting tool guidance system is much harder than a leveller because a guidance system has to work fast and accurately, more-or-less in real-time? The leveller only has to be accurate; it can take as long as it needs to get a good answer.

As I understood it, a rather wobbly laser beam is scanned to fill the whole room with a plane of light. The test girder is set up just below the plane, such that a digital camera sensor interrupts the scan as it sweeps the room. Thus the sensor, with pixels about 4 microns apart, detects a sequence of hits spread across several vertical lines of pixels either side of the horizontal plane.

Which pixels are lit up by hundreds of scans, are recorded by a computer. They're distributed evenly either side of a mean, and the computer calculates the mean. Air movement will spread the distribution but not alter the mean, and, although the mean's certainty will reduced, this can be corrected by taking more samples. Basically, the leveller relies on the software (Python) to achieve accuracy by identifying where the majority of a large number of photons hit the sensor. The method gets more accuracy than the pixel spacing suggests, and is less vulnerable to air currents because it averages them out.

The main problem I see is the time averaging takes to get results. Hard to see how it could guide a tool unless the laser scan, sensor and computer were all super-fast.

Looking at the Renishaw XK10 blurb, I suspect it works in the same way as the Youtube leveller. 'The laser plane is aligned and datumed at positions 1, 2 and 3. The deviation measured at position 4, in combination with the axis length, is used to calculate the vertical angle between the two axes.'

Much better engineered and the XK10 does range correction and self-checking, but at root there's also a laser plane, measurement of a deviation and lots of software running on a powerful Windows 10 box. I guess the deviation is calculated in the same way as the Youtube leveller does it, but the numbers are crunched faster and more elaborately from multiple sensors if required. Similar, but faster, easier to use, considerably more capable, and reliably accurate. (Perhaps 100 times better than Youtube!)

Bouncing lasers off the moon is harder than I thought. I imagined it being done with a red pen laser. Nope. A grown-up laser is fired through a big telescope to collimate the beam, about 3.5 metres diameter on earth. After travelling 380,000km to a reflector thoughtfully left behind by a moon mission, the laser beam is 2km across. Then, the tiny amount of reflected laser light that gets back to the telescope has to be separated from all the rest. It's all very high-tech. I expect they had to use an Arduino!

Dave

Dave

Edited By SillyOldDuffer on 31/03/2023 15:42:04

We used to use an older version of this Renishaw laser machine tool calibration system

Ditto.

I had a Renishaw ML10 laser + EC10 environmental compensation unit + a big box of optics to play with at my last-but-one employment. Beam movement due to draughts was an issue, particularly when working with long distances - the return beam could be seen wandering about in big, draughty machine shops. Long distances also showed the return beam had increased from a bright approx. 1.5mm dot that left the laser unit to a dim 5p coin sized return beam. - accurate alignment was essential to get a useable signal strength & in those circumstances a drought shop was a nightmare. No Windows 10 powerbox back then - the Renishaw software we used was DOS based & ran (via a half width ISA interface card) on an Amstrad ALT 286 monochrome laptop & a dot matrix printer !

We did have a commercial system very similar to the system in the video demonstrated at one point - probably 30 years ago. The demo was done in our works levelling a floor borer bed . The tripod mounted laser set up a level plane & there was a sensor on a base that also had a digital readout that could be zeroed. After that, as the sensor was moved about the bed next to each levelling point it displayed how high or low that point was relative to the set zero point & the levelling bolt adjusted to bring the readout to zero. It was impressively quick to use, but the boss was not impressed enough to buy one !

Impressive to see what some (clever) people can accomplish by creatively re-purposing relatively cheap, off the shelf consumer items.

Nigel B.

My tool guidance effort was a real-time servo system - no averaging.

I was more successful with a profile measurement gauge that used a linescan camera illuminated by a LASER working over ~200mm. Sub-pixel resolution is easily possible through interpolation - similarly real-time edge detection from a video signal.

Averaging will help remove random effects, but can't account for systematic errors. Hopefully it will get them what they want. When I dabbled in this professionally, anything like this was done in the 'Constant Temperature Room' which (as the name suggests) was maintained at a tightly controlled temperature 24/7 so you didn't get problems with thermal gradients due to temperature differences between equipment and the air.

The beam divergence of LASER pointers is pretty poor.

The Renishaw ML 10 and it's replacement are interferometers. It can only measure distance between optical elements. Hewlett Packard made a similar system. It used a dual wavelength laser. Both use Helium Neon gas lasers with wavelength stabilisation but minute adjustment of the cavity length. Typically this was done with a heater to expand the material. Angular and flatness measurements rely on distance comparisins with special optics. For full accuarcy they have to be compensated for the local speed of light which is affected by air pressure and temperature. I've used both the HP and Renishaw systems.
Completely different to the beam position sensing being described in the video.

I have contributed to the discord group and also the starter post at Mycncuk. The results of my experiments with webcam/laser and Webcam/wire combinations favour wire for straight line measurement. Repeatable accuracy was obtained with wire and webcam, alignment to a few microns. Tensioned wire is affected by wind and draughts so it must be done in still air. It is also affected by vibration from say nearby machinery.

Catenary sag can be calculated here. **LINK**
Over say 2.5 metres not a lot with fine music wire correctly weighted to a bit below the stretch creep point.
Use the wire specification from below. Effectively you can almost ignore the weight but you need the tension and length.

Fine music wire can support a lot you will be surprised, check here. **LINK**

Leitz Wetzlar made wire alignment microscopes, one of which was electrically controlled, so I was told.. I have not been able to locate a photo or specification for that one?
I have a purely optical one with a split image prism to centre the image on the way, It came up in a web discussion and the owner had no further use for it, I do. It is graduated to 2 microns. It will be interesting to test it out against gauge blocks. when it arrives by mail. I will use it to compare with other methods.

I do however really like the concept of laser only for scanning a large planar area where wire becomes less practical. It is well worth following the Discord group. there are quite a few minds working on the problem.

Oh! A special thanks to the generous British person that reached out to me and sent a very nice webcam module to me in Australia, You know who you are. It was a while back but i still remember it.

.

I don’t think I have ever seen the electrically controlled one, John … but this may be of interest:

.

.

MichaelG.

.

Source: __ https://archive.org/details/pdfy-8Bodc5JEPZ0RGXru/mode/2up

PDF download should be available, but it seems to be struggling today.

Edited By Michael Gilligan on 01/04/2023 20:20:10

Hi Michael
Thanks for the info.
You are amazing you always know where to look.

They must have updated it. Apparently there is a later one as well. Maybe one will turn up!
I don't speak German. It would help if I knew the actual German naming convention for it.

This is the one I have on the way.
It will be very interesting to test it out.

Very interesting, John, thanks for sharing the photos.

I don’t speak German, but fortunately most technical German is very literal and therefore translation is quite simple.

Fluchtlinienprüfer translates to Flight Line Tester

and it doesn’t take too much imagination to realise the conceptual similarity between a flight-line and a taut wire.

We did a lot of this rudimentary translation when I worked at KODAK in the early 1970s, but it’s much easier now … I just turn to DeepL when I see a tantalising word, instead of hauling-out the dictionary.

MichaelG.

.

Edit: __ and there is even a Wikipedia page about them:

https://de.wikipedia.org/wiki/Fluchtlinienprüfer

Edited By Michael Gilligan on 02/04/2023 05:34:28