Bit of a break from last time, I’ve been stuck into a huge new project with Force.com.
The stabilizer has had a few small problems. I took it out to an alpine environment for some testing, filming the guys riding bikes, and the 5 degree air made the wires in the gimbal go very stiff, so stiff that the gimbal couldn’t stabilise because the motors weren’t strong enough.
I then spent a lot of time trying to work out how I can do without the wires. The accelerometer communicates back to the board via I2C bus, and there is a wireless option available, however its infra red and doesn’t have a large angle of operation. I think that lag would also be a big issue. I ordered some softer wires and cable form hong kong, but when they arrived they weren’t that much softer.
The best that I could do was re-route the cables so that they do not interfere with the operation of the gimbal as much. I moved the cables that exit the IMU 90 degrees so they follow the tilt axis. I ended up with this arrangement:
I waited for a sunny day so that the wires could get to room temperature just to be sure.
My mate John came out with me to test. Johns method of testing was to film me riding the bike, whilst he runs alongside the track, once with the camera just by itself in his hand and the second time using the camera stabiliser unit.
Results are here:
Ok Johns non stabilised footage is REALLY shaky, sorry about that. But it does demonstrate potentially how well the rig can stabilise!
This encouraging result has prompted me to make a list of things that are wrong and make a second prototype. Notably, here’s what I found:
1. John mentioned that the rig as it is, is difficult to use. I based this on the rig by Movi, but I think their design is probably great for filming days of our lives, but not great for action sports. John mentioned that one-handed operation will enable the camera to be handled easier and carried into more places, which is a great idea. I’m picking up one of these http://camcaddie.com/sk8/scorpion.php for prototype 2.
2. I’ve noticed from Johns footage that the footage isn’t as silky smooth as I’d like. It is dampening a lot of movement (just look at the before footage) but it could be better. My main concern is that although we correct for pitch and roll, we don’t correct for yaw (left/right movement). Prototype 2 needs to have yaw stabilisation.
3. Upon inspection of the jig, I have found that the mounting system that I have offers a lot of movement within the dampers. Prototype 2 will not have any dampers due to this. As the device is handheld, we are not likely to see a lot of high frequency movement due to the fact humans tend to absorb this naturally.
4. The motors aren’t strong enough! I shouldn’t have to wait for a warm day to get the wires bendy enough to use!
The next post will be about prototype 2. I have found that by increasing the pulse width to the stepper motors I have actually corrected point 4 and made the whole stabilisation effect a LOT stronger. A case of RTFM: increasing the pulse width in the settings will increase the motor strength. I increased the PWM until the motors were on the brink of resonant vibration.
Prototype 2 will see the addition of custom yaw stabilisation, and a whole new, more robust and stronger gimbal made by Tarot RC: