by Emile Orzea
Aleksey Zaytsevsky is at it again, another breakthrough in the field of camera stabilization technologies.
Over the past year, video operators have been actively switching to controllable camera mounts directly connected to an electric motor. The result might seem satisfactory, but the overall quality of this solution tuned out to be questionable, since an electric motor without a gear box is inefficient at low RPMs.
An alternative solution demonstrated by Aleksey Zaytsevsky was simple and elegant: it only required the replacement of the position sensor in the servo drive with a force sensor. As the result, the servo drive rotates the camera resting on the unstable base, but the tilt of the unstable base makes no sense. This allows the operator to rotate cameras of any mass and use all the benefits of inertial stabilization, including the possibility to use a camera with an additional counterweight.
A few prototypes of the devices being designed have been presented so far. The video shows that the unwanted mechanical resistance of the servo drive can be so low that the arm can be rotated with a burnt out match. In a different experiment, the reduction servo drive rotates and stabilizes a 6 kg platform with the torque being as low as 3.8 kg/cm. The servo drive was also tested on an Ecilop Easy drone. For a more visual demonstration, an on-board camera with zooming was used.
Force Servo also allows for other usage scenarios. Thanks to Force Servo, the rudder turn angle is automatically limited at high speeds, preventing hypergravity. At low speeds, vehicles retain their maneuverability. It also prevents mechanical overloads in cases of rudder jamming.
Comments
Hi,
Here is another demo:
https://vimeo.com/82082914
The accuracy of the controlled position for the new servo is not discrete and depends only on the accuracy of mechanical parts.
Honestly I think that bit at the end showing force servos usefulness on airplanes is more appealing than on a gimbal. But they're probably going to be expensive (hand modified?) until rctimer or hobbyking "borrows" and mass produces them.
Not sure at this size, but at the sizes I work with, this is no longer true. You loose RPM with torque, but overall you end up with a faster system, with no loss in precision.
I agree with most of what you say Guy, but my day to day job as a lead design engineer, for a machine builder, has lead to a lot of research on servos, in particular direct drive applications.
We just developed a direct drive system for a large laser cutting system, and well working with Koll Morgen, I was able to prove there is enough resolution in a modern servo drive (There AKD for example) to use current as a feed back method. The system could detect the weight of my baseball cap, on the Z axis with a mass of several hundred kg's.
So the norm for direct drive servo's no longer applies, as we have shown, so expect additional direct drive applications in the near future, WITHOUT additional sensors :)
We are using Current feedback now for some real funky stuff (which I can not discuss), but I'm sure others are working on the same.
Still cool stuff Aleksey!
@Guy,
A drive belt system will only add weight and complexity. Therefore there are always tradeoffs between weight/complexity and positional accuracy/stability.
I have worked with torquers in the Aerospace industry where, until recently, money was no object when building direct drive tracking gimbals.
Regards,
TCIII ArduRover2 Developer