Designing Motion Control Camera Rig

[Madjak]

Hi all,

I'm not exactly which is the best forum to post this but I figured here was a good start.

I'm just researching controllers, motors etc for a Motion Control rig to move a camera around. It will be a 7-8 axis machine and will need to reach about 4m laterally and vertically. Most motion control camera rigs are very pricey, and are designed for film cameras. With the new digital cameras being released overall weight can be reduced drastically and it opens up the possibility of building lighter weight boom arm camera rigs.

Mechanically I have a good concept on how to design the structure and I have used CNC machines before, plus I have some very tallented programers who can help on the software side, but I really need to get some help in where to start looking for good controllers and motors for the rig. Overall I'm expecting around 40-60KG's of moving mass with accuracy at the camera's location the prime concern.

I pretty much am looking for a starter kit, 8-axis controller, and range of motor sizes that I can get a for a decent price so that I can start figuring out the mechanical aspects.

Does anyone have any good links to helpful sites?

Rick

[amplexus]

Are you looking at a rigid gantry design? it would be much more accurate than than some kind of suspended cable design. Keling has nice stepper moters fairly cheap. Gecko 540's are 4 axis for nema 23 motors or you could use 7-8 203v vampire drives as in unkillable for nema 34 drives. Aluminum frame, profile linear rails and bearings, rack and pinyon drive, if you need more accuracy big ball screws probably custom out of china and not cheap. how fast do you need to go. Screw whipping will be a problem with small diameter screws at high travel rates and big screws are expensive and heavy. Linux might be easier to customize. Do you really need 8 axis? All of this will be much heavier than your estimate.
Amplexus Ender

[the4thseal]

i do a great deal of work with the rigs in question as well as design and build them. i understand what you are thinking but it will be expensive when you are done. the systems are expensive due the amortizing of all time and r&d. i do not mean to discourage you, have at it. being that you are in OZ i am not sure what you have access to? do you want to use steppers or servos , are you thinking of building a MILO? How about something like a sparrow? thoughts.

[skullworks]

I think you need to dig into EMC as it has much of what you require - but you are still facing a huge amount of customization. 9 axis exits - but not in the way you need.

You will likely need a modified kins file, and a modified GUI.

You will need some kind of teach pendant interface such as is used to program robotic moves. I work with a few robots and the "puma" might work after mods.

I'd love to build one myself for my little Sony Handy CAM. (maybe half the size that you describe)

IIRC there isn't a GUI for Puma robots yet and your application adds an additional factor of the time/rate of change that is more like the Inverse time function than "normal" feed rates.

I have an idea for a control pendant - use a LCD picture frame turned on end (vertical) and put a touch screen overlay on it - write a GUI that uses the touch pad as primary input. (Ideal was for a Puma pendant, but would work for both just load the proper GUI)

In the process a feature long ask for in EMC would be refined - Teach mode and playback.

You would be breaking ground in many areas, but you would not be alone.

BTW - Stepper design would be nice - but the cogging is a deal breaker - even a small servo might need dampers to make the motion less noticable than the change in prespective.

[Madjak]

Originally Posted by the4thseal i do a great deal of work with the rigs in question as well as design and build them. i understand what you are thinking but it will be expensive when you are done.

I figure most motion control rigs out there like Milo are designed to handle very large camera loads. With the new RED cameras you can locate most of the mass away from the moving parts of the rig. The entire system can be designed much lighter with similar deflections if the camera weight is reduced.

I have a few mates who have some a CNC workshop and often work in interesting materials like Carbon Fiber and titanium tubing. If I can keep the weight down over the entire system then I can absorb some more exotic material costs if it helps the rigidity and weight.

I've worked in 3D a long while and I'm kindof approaching this from a different tack to what Film guys would. My idea is to have a realtime 3D engine driving the arm, with keyframable coordinates. Both capturing from the rig and also "printing" to it.

We currently don't have access to any motion control rigs so have to use a "software" version... It works well for inserting people into 3D that have been filmed from multiple takes.

Rick

[Madjak]

Originally Posted by amplexus Are you looking at a rigid gantry design? Amplexus Ender

It will be a double boom arm with elbow. I have no idea if that sort of reach is even achievable but I figure I'll just start making it and see where I get! :P There are ways to spring the arms so they take their own weight, kinda like a off the wall desk lamp.

I'll try and post some links to the prelim design once it's done.

Thanks for everyones comments.

Rick

[SteveS]

Rick,

I don't know enough about Aussie geography, but you should try to find someone who can get you into a car or truck assembly plant. The robots for painting are a significant number of axis. The techs guide (train) them in the general path to get coverage and then fine tune from there.

They will be really large and extremely heavy, but could give you some pointers. You could get a closer idea on the number of axis you will need.

Steve.

[amplexus]

How much absolute accuracy do you need? Could you just build a room the size you need, mount rails on the x y and z axis, and set up another 3 axis on the camera mount. It would be easier than a fancy high tech arm.
Amplexus

[Madjak]

Originally Posted by amplexus How much absolute accuracy do you need? Could you just build a room the size you need, mount rails on the x y and z axis, and set up another 3 axis on the camera mount. It would be easier than a fancy high tech arm. Amplexus

The problem with a gantry system is in portability. The idea is the system could be set up not only in a studio space, but also on location. My original concept was a large gantry, kinda like the big shipbuilding ones, but I was a little worried about speed and overall size. Also a big gantry in a studio space is hard as the gantry can intersect with lighting hung from the ceiling and cast shadows on talent.

We're doing mostly car ad work atm and it requires a rig that can by set up by the side of a road to shoot the background plate, and then also in a studio to shoot the car / talent etc.

Rick

[Madjak]

Originally Posted by SteveS Rick, The robots for painting are a significant number of axis... Steve.

My design is very similar to a robot arm used for Car Making / painting, except very light in comparison. I did come across one years ago, but they are very complex to program for (so many axis of movement) and are designed very heavily.

I do already know the number of axis I need, dependant on the design what I don't know is the size of the motors and type (stepper or servo) and how to connect to the arm (belt, direct etc) and if they need to be geared etc.

There are so many variables it's hard to know where to start!

My idea is to get a design, and iterate it from there till I hopefully get something that works!

Rick

[Switcher]

The robots that I worked with, welding & painting automotive parts, had pendants that was very easy to program/edit axis moves. Just like the direction pad on a teach pendant for cnc.

I think Mach3 will allow you to write g-code, from jogging each axis?

How will you deal with power supply to the whole setup while out on location?

Sounds like a fun project.


.

[Madjak]

I'm unsure how to do this yet... yet another path of learning required, but I really want to be able to control the motion rig via a realtime 3D engine. So imagine having a 3D model of the arm in a 3D app that you can animate and keyframe over time. The software would talk to the 3D app and "proof" the motion to make sure it was feasible and works within the machines limits. The ease curves etc can all be controlled from this software. You then run the move, and find out from the controller what was actually achieved, which you can then tweak and re-run multiple times. This is where I need to find a controller that allows the PC to query the actual rotations of the motors through time.

There are other ways to get the camera's location in 3D space via witness cameras at known locations but that adds complexity. There must be a way to get that information from the controller. As it knows where it's motors are at any given time and if it can be captured in real time it's immediate.

Another idea is that you can set the arm to manual mode, and record actual locations and rotations of the camera that can be sent back to the software so they can be set to a timeline.

All fairly creative stuff, and I have no idea if it will work. The end concept is to be able to see a 3D model running at the same time as the filming, with the video overlayed or keyed into the 3D scene / object.

As for powering the rig on site.... Generator? batteries? Depends on the size of the motors I spose!