Camera Positioning Using Head Movement

[Hell-e-Guy]

Hi Guys,

After finishing my Magnetic Proximity Sensor project I need a new project to keep me up at night.

I saw a cool movie of a camera mounted on a plane that had pan & tilt rotation using the head movement of the pilot on the ground wearing TV goggles.

What kind of sensor can give a signal relative to it's current position?
I thought of a simple gyro sensor that will be integrated but the resolution and drift will be hard to compensate for (I think....).

Any Ideas on how this this done?

Hell-e

[robe_uk]

I have seen this to, Its an the AP section at the minute and there might be a link to the site of the maker.

having read through the site it appears they use gyro and have to activate a reset every 2-5 min, were the user has to look streight ahead and press a reset to allow the system to know the user is looking streight ahead and so adjust the camera to suit.

[green66]

See this (http://www.rc-tech.ch/web/index.php?page=shop.product_details&flypage=shop.flypage&product_id=70&category_id=16&manufacturer_id=0&option=com_virtuemart&Itemid=26&lang=en) for some ideas; converts output of head-mounted gyro to transmitted signal thru pilot's regular r/c Tx.

[mike50]

See this (http://www.rc-tech.ch/web/index.php?page=shop.product_details&flypage=shop.flypage&product_id=70&category_id=16&manufacturer_id=0&option=com_virtuemart&Itemid=26&lang=en) for some ideas; converts output of head-mounted gyro to transmitted signal thru pilot's regular r/c Tx.
I think that is the same system being used by VRflyer, who has posted extensively in the "Electric Plane Talk" forum.

Mike

[Hell-e-Guy]

Hi Guys,
Got hold this week of a couple of working gyro sensors from old LAMA 4 in 1 boards.
Now I can proceed with this project.
What will I need to do in order to use the proportional output of the gyro sensor to give the servo my head angle?
As I understand I will need to do some kind of double integration, but this is where I am stuck...
One integration should give me the head speed and the other should give me the angle.
I am sure many tests should be done until this will even begin to work but I would like to have some equations to get me off in the right direction.
I will work with a Cypress MCU and will write with a C Compiler.
Differential equations I already forgot from my studies...

Any help will be greatly appreciated,
Hell-e

[AndyOne]

I've been looking at heading lock gyros for this application and I think this is the only type that's satisfactory. No calculus involved, in fact I think you need differentiation not integration which is what the heading lock function gives you. Ordinary rate gyros aren't really any good because they will always drift.

Andy.

[Hell-e-Guy]

Hi Andy,
I am pretty sure I will need to integrate the sensor's output. The output is the angular acceleration integrated once with respect to time will give angular velocity , integrated once more with respect to time will give the angular position or angle. Of course drift is the problem here... but that's the fun of trying to make it work as good as possible.

By the way how where you thinking of using heading hold gyros for this purpose? You can buy now pretty good E-Sky HH gyro for around 35-40$.

Hell-e

[HELModels]

You turn your head and the gyro wants to correct a servo to hold heading. Take that correction and send it to the servo gimbaling the camera. Now, how do you get that correction into the TX?

[Hell-e-Guy]

ElectroStorch,
If I understand correctly when you say correction you mean the output from the gyro to the servo...
I will try it tonight, I have a GY401 that's not in use at the moment.

In my TX (Futaba 9ZAP) this is not a problem. In Trainer mode I can give control only to certain channels. I will be in constant trainer mode and feed the TX with the 2 channels for pan and tilt. I think I will need the MCU to reconstruct some fake pulses for the 9ZAP to input the correct channels.

Hell-e-guy

[AndyOne]

Hi Andy,
I am pretty sure I will need to integrate the sensor's output. The output is the angular acceleration integrated once with respect to time will give angular velocity , integrated once more with respect to time will give the angular position or angle. Of course drift is the problem here... but that's the fun of trying to make it work as good as possible.

By the way how where you thinking of using heading hold gyros for this purpose? You can buy now pretty good E-Sky HH gyro for around 35-40$.

Hell-e

OK well my mathematics is pretty rusty now but what you need to do the job are heading hold gyros mounted at right angles on the head-set. You need to feed each of them with a dummy receiver signal say 1.5ms pulse which you may want to adjust to trim the position of the view. From the output from each gyro you will need to take the servo signals to replace two of the pulses in the PPM signal from your transmitter. This is best done on a Futaba TX using the channel selective trainer mode. A micro of some sort will need to generate say 8 pulses the first 6 can be set to a constant 1.5 ms as they will be ignored anyway but they need to be there, put this signal into the trainer input of the TX. The last 2 pulses for channels 7 and 8 will be the new ones introduced by your electronics. Set the trainer mode to accept only channels 7 and 8 , put a rubber band round the trainer switch to hold it on. Then take channels 7 and 8 from your receiver and feed them to your pan and tilt servos.

Andy.

[Hell-e-Guy]

Today I connected my gy240 to a servo and to a servo tester. I placed the gyro on the table and turned it around its axis. The servo didn't follow the gyro very well. Parts of the servo travel where very smooth and accurate and at times very jittery and inaccurate. Unless I am doing something wrong i don't see how I can use this....
I used a sanwa 091 micro servo for the test.
Any ideas???

Hell-e

[jeffs555]

The problem with a heading hold gyro and a 1.5ms input from a servo tester is that it will try to maintain a fixed heading. If you displace it from the fixed heading, it will change the output, but not necessarily linearly. I know with my helis it doesn't take much displacement to move the tail rotor servo to max. Have you tried adjusting the gain way down?

I seem to recall some cheap VR helmets that just used pots with a mechanical linkage between the head and shoulders to control pan and tilt. I think for controlling pan and tilt on a camera, you would want to measure the relation between the head and the rest of the body, not the absolute position of the head in space.

[Hell-e-Guy]

Hi,
I tried my GY401 instead of the GTY240 and when I put it in Digital Servo mode movement improved substantially although I was using an analog servo. I will ask a friend for a good digital servo and test it.

Hell-e

[AndyOne]

I've only tried this with a clubmate's heli but turning it on its axis moves the tail pitch quite linearly enough when the gyro gain is adjusted to a low level. The gyro being used was a fairly expensive Futaba one though, there may be a price component to this.

Andy.

[Hell-e-Guy]

Hi Andy,
I think you are right. Seems like price is the key here. I will test wit a fast and accurate tail servo. Do you know what does the Digital Servo mode on the gyro do?

Hell-e

[AndyOne]

Hell-e,

I'm no expert on gyros but I'm pretty sure the digital mode changes some of the closed loop dynamics of the system. Since in this application there is no closed loop, I can't see it making much/any difference.

Andy.

[rjfon1]

Hi Guys.
Have a look at the UAV forum. There are a couple of concepts there.

Follow this link. It sells gyro modules, and the link goes directly to a dual axis angular rate gyro, for about $70 US. It uses analogue voltages as the output. If you want to reset for drift, I believ you just remove power and it resets to zero. This would work in the same manner as one of the previosly listed devices. Just look ahead and rest. Would it be possible to use a multiplier/divider circuit and something to convert analogue voltages to PPM or Visa Versa? This could then be fed into a transmitter buddy box connection, and the master set for perhaps only those channels, ie: x-axis channel 6, y-axis channel 7?
http://www.sparkfun.com/commerce/product_info.php?products_id=698
Just a thought, as it would do away with having to compile code.

[jeffs555]

Being an angular rate gyro, you would have to integrate the output over time to get an angle, and then convert that to a PPM signal.

Another option would be a magnetometer or electronic compass to measure heading, and an accelerometer to measure tilt. Sparkfun also sells both of these.

[Hell-e-Guy]

Hi,
Tried a good digital servo today. Works much smoother but still not good enough for camera stability. Jeff, good idea!!! I'll do some homework on the digital compass, an accelerometer I already have....

Hell-e