Automating camera focus (pan/tilt/turn) of telescope

Thread Starter

RogueRose

Joined Oct 10, 2014
375
Here is the setup of the item that needs to have remote image finding and focusing. A telescope (with camera on lense) is mounted on a tripod or some other base along with another very HD camera (possibly 1-6 units for a 360 degree view) that does not have telescopic focus. The HD cameras views a larger area and is used to view a large portion of the sky in order to direct the telescope to the proper area. The camera's know where the "center point" or "home" is for the scope, how far the scope currently is from "home". The other camera's have known fields of view relative to the "home" location and may have 36,000 pixels on the horizontal axis from home/center all the way around and back to home (6 camera's at ~6000 x 3400, X,Y resolution). The scope and "focusing camera" turn independent of the 6 field of view (FOV) camera's. The FOV's images are used to find images, select the target and then the scope & "targeting camera" turns to that sector using the camera to help identify the location better.

The output of the cameras are on screens and when an area of interest is identified on the screen it can be selected (mouse or by touch screen/pen) and the motors need to turn in the proper manner to focus the scope to point at the intended area. There are similar systems for telescopes such as this, but they don't really do the same thing - a specific "address" can be selected (usually related to other known stars) and the scope will find it, but this has severe limitations and can't be used to look at oddities in the sky (comets, asteroids, space stations and "UFO's" - no joke..). So when looking at these things, it is imperative to be able to select a point in the sky to immediately focus upon and move the scope to that area. This is to be done with a remote connection with the scope placed 50 meters to 5,000 miles away with the data transferred back over the internet. The video/image transfer is not a problem but setting up the focusing is the largest hurdle.

When the image is selected and let's say it is 8000 pixels left and 1300 up, the device needs to know how the motors/actuators need move to position the scope to that exact location as quickly as possible. I'm sure this could be done by using directional control, holding the left arrow until under the target then the up arrow to move it vertically, but I'm sure this can be automated to a large degree.

I need to figure out what motors or actuators would be best suited for this application for very precise and fast movement of the scope. I am thinking that stepper motors might be ideal for this, though I've only read about them. Since they are used for printers/plotters, it seems they are capable of very small incremental changes and this can be even further reduced with proper gearing controlling the scopes movement (reduction gear from the motor to the "movement transmission"). So if a stepper motor can be programmed for each "pulse" that it moves 1 pixel, then this would be very easy to program.

The calculations of the distance from current center of the scope/camera to the new "target" can be plotted in an X,Y calculation and whatever the X.Y equals, then it gives that number of pulses to the motors to acquire the new image. from there the scope can either auto focus or manually focus (remote manipulation). If the movement is off, it may be necessary to have another camera in parallel with the scope with something like 2-8x optical zoom (as compared to the 400-2000x telescope) to give a more resolved image on which to target more accurately.

The weight of the scope seems to vary from unit to unit (carbon fiber are much lighter but 3-4x as much) but the weight looks to be from 21-46lbs depending upon which unit is selected, and another 4-8lbs for camera's, so I am looking at ~25-55lbs mounted on the pod that need to be able to rotate 360 (or +/- 180 from center but ideally freely rotate 360 continuously to track targets) as well as tilt from horizontal to about 75 degrees (90 would be ideal but that might be very difficult - maybe having retractable legs on tripod would allow for vertical viewing by tilting entire pod in one direction...).

So what I'm asking people here is what would they suggest be used for the motor to pan/turn the scope horizontally 360 and what for the tilt (vertical movement). I was wondering if something like a linear actuator might be more useful in the tilt implementation.

Now all of this needs to be controlled electronically somehow and I'm not sure what device would do the best job. If this can work with the pulses as described, say these are 12v motors, could I have something that sends X number of pulses to get to the proper position or some other similar method? This and the proper motor mechanism are the biggest sticking points I'm having and need any suggestions that may be relevant to this scenario. What would you suggest?
 

Thread Starter

RogueRose

Joined Oct 10, 2014
375
Is this post in the wrong forum, not clear as to what I need to do, or does no one know what is needed for the certain parts of the project?
 

DNA Robotics

Joined Jun 13, 2014
647
Your project has a lot of different parts.

I wrote a windows program once for a store surveillance system to aim a camera at an object that is anywhere on the screen. Say you are interested in an object in the upper right corner of the screen. You move the cursor to that position and right click. You calculate how many screen pixels that position is from the center of the screen. You send one stepper motor signals to move the camera right that many pixels and the other stepper motor signals to move up that many pixels. Then that object will be in the center of your screen. How many step signals you sent to the motor drivers depends on your gear ratios. For instance, it might take 23 steps per pixel.
 

DNA Robotics

Joined Jun 13, 2014
647
Regarding auto focus:

I used DirectX to work with screen memory. In your case you are looking at light objects with a black background and want sharp edges. You examine pixels starting from the center of the screen that now contains your light colored object, looking for a fast transition from light to dark. If there are many gray pixels at the edge of your object, it is out of focus. Send signals to your focus stepper motor to turn left & right until there are as few gray pixels as possible. (in focus)
 

DNA Robotics

Joined Jun 13, 2014
647
Regarding auto tracking:

Assuming you have a round light object in the center of the screen. You examine pixels starting from the center of the screen looking for the transition from light to dark. You can tell if it is off center and send signals to the stepper motors to move the camera.

In the DirectX software development kit (SDK), there is or was the source code for a sample program "DIBLOOK". That may be a good place to start. DIB is a device independent bitmap.
 
Top