A set of cameras for levelling stuff when building

Aug 01 2017 Published by under Uncategorized

Say your are building a simple fence. What is not so simple is that every pillar must be in level and vertical.
The normal way of doing this is to have a spirit level, laser line level, water level or even theodolite.

Why not have a (self levelling) camera or two. or three?

Since the camera knows its orientation and its lenses and in turn optical aberations it can say what is level and in line and not.
Point out the pillar you want vertical and let the software in the camera do some image recognition and tell you when it is. Add another camera looking at the same pillar, but from another angle and the pillar’s angle is shown to you in all directions.

Now each camera has 3, or more, of the pillars in view and can tell you both if the pillars are vertical and if they are aligned and if they are level.

Use the same camera and software to help you with attaching the pickets equidistance.

If you give the camera some clues about distance, for instance by attaching or showing a ruler at some clever positions, it can return measures directly without you having to take out the laser range finder, carpenter’s/folding ruler or tape measure.

If you give the camera(s) more information with calibration points, like balls mounted on a stick stuck steadily into the ground, you can move the cameras around and the system will still know where it is.

Focus and focus point is a problem as optical aberrations might change when refocusing.

No responses yet

Measuring a room, a house, a factory, a statue or anything

Nov 16 2011 Published by under Uncategorized

Here comes an idea I have had for 20 years or more.

Create a wand with a laser distance meter.  The update rate must be quite high.

Have a device to put somewhere.  This device knows the position and orientation of the wand.  The wand sends data to the device.

As long as we keep the measuring fast we can “paint” the surroundings and store the data.  We can then do some smart calculations and figure out what the room/object looks like.

If we want to measure more than the distance between the wand and the device allows we can set the device in “moving mode” after putting the wand down; move the device, and then set the device in “measuring mode” again.  This way we can move the measuring through a whole multi floor factory or from a building to another.

If we have computer power enough, we can do it all real time.  It would be super to be able to see how one paints the objects to see if there are any forgotten areas.

There should be a way to help the computer calculate the data (afterwards).  One can give it hints about what is a corner and what is a pipe.  Or a nose on a statue.

Another solution which probably requires more computing power, is to do it with a (film-)camera.  Or a stereo camera to make the calculations easier.  Hold the camera in your hands, on a pole, on a vehicle or in a ball to through around.
Does the Microsoft Kinect solve this problem? It measures a room alright but at a high enough definition? At least it can be used as a proof of concept.

No responses yet