Need advice on best approach for Camera Rig

[sbc350]

My first post and I'm a total newb in the sphere of robotics and electronics.

I'm in the process of constructing a simple video camera rig that will operate in a similar fashion to a mini camera crane with both manual, preplanned and recorded paths.

I'm looking for a solution that will offer very smooth movement and will be relatively simple for someone of my inexperience to implement.

From a small amount of testing with a RC servo and Arduino, I can see that standard servos are probably unsuitable. I've looked at OpenServo (which looks ideal but more DIY) and Dynamixel.

Can anyone suggest any other alternatives using serial servos that I may have overlooked?

[Orac]

Hi,

The driver for this kind of project is scale !

Different technologies are needed when jumping from our hobby scales up higher and higher and could mean moving from hobby servos up through steppers and screw threads, on to larger motors, gearboxes and encoders.

I suspect that before anyone could offer any useful advice, a more thorough description is needed, including scales, degrees of freedom, etc

Thanks

[sbc350]

This project will essentially be a pan and tilt head on a jib that can elevate and possibly rotate. Jib length will be around 1m and cameras will be no larger than a DSLR, so mid range RC servos should be fine. Although it will be tripod mountable, my primary goal is to use this gadget hand-held.

Ideally I need a servo solution that will offer smooth accel and decel, hence my interest in OpenServo. Only modest accuracy in repeatability is necessary initially.

[i-Bot]

I would avoid standard servos. They are most likely not powerful enough also you will need something like openservo to tune the servo parameters especially for the jib.

Open servo is scalable, so you can add different output drivers for larger motors.

Even small DSLR cameras seem to weigh about 0.5Kg. The pan/tilt will probably weigh the same again. I assume the jib is counterbalanced. You need to decide the speed, and acceleration/deceleration required to determine the motor torque.
Also take care on resolution of encoders if using a long jib.

[sbc350]

Thanks for the advice.

Servocity have a range of pan/tilt mechanisms utilising RC servos. The servos they use are heavier duty ones, and for some of their pan mechanisms, they also gear them down substantially. I was thinking of adopting a similar approach, since high speed camera movement is unlikely to be necessary for my application.

Rather than drive the jib directly, I was thinking of using a servo to offset a small weight attached to the counterbalance. The only problem I can foresee with this, is that I may need to somehow brake the jib as it approaches its balance position.

Whilst seeking a solution that can offer smooth motion, OpenServo seems to offer the most elegant (in terms of curves) and inexpensive option.

The whole robotics, motion control, electronics and microcontroller stuff is all totally over my head, but I'm finding this forum a rich source of information, even though you often seem to be speaking a different language and I probably only understand a fraction of it!

[robosavvy]

In the past we've supplied Robotis servos for a similar application in a UK company.

They used RX-28s and a USB2Dynamixel connected to a PC to control the servos. You could also use a CM-700 controller if you don't want to use the PC.

They had first chosen RX-10 servos but they did not meet their needs and ended exchanging them for RX-28s (which can hold a stall torque of 28kgf.cm).

We did not develop any of the application so the only details w ehave is the parts they used (which we supplied).


Regards
RoboSavvy

[sbc350]

That's interesting. My only concern with the Robotis servos is just how smooth and seamless movement can be compared to an OpenServo.

OpenServo appears to be able to take care of the ease-in/out curves and effectively blend one into the other, which is possibly the most important aspect of my project.

I suspect that any other serial servo will require more work in scripting cosine or quadratic curves on the microcontroller, or am I mistaken in thinking this?

I noticed a thread within the forum where some guys were porting the OpenServo code to Robotis servos. That's an interesting possibility, but most of what they talked about went way over my head - I need a solution that is relatively simple - like me!

What does really appeal to me with the Robotis servos is the fact that I could use them without modification and the AX12s seem to offer good value for what appears to be a well constructed servo, and the mounting options are far superior to a standard RC servo.