Hello guys,

I need to set up a motion sensed camera movement for my sony handycam. the most hard thing i face is , during stage events, at my college the spokesman, moves around the stage, this makes the camera guy to adjust, the hadycam, to the direction he moved. I am dreaming of a sensor connected to the spokesman, which the handycam detects and the camera's direction is focused on to that specific location.

Need the help from basics , as i'm not too much into circuits

ThanX.

 

Sound like an interesting project. There may be commercial solutions already out there. Rolling your own might get interesting and it might be hard to get suitable results. For instance, if someone is on stage talking and isn't moving around (i.e., his feet are stationary) but he is moving his upper body as he turns to point to the screen or to face a different portion of the audience, does the guy with the camera constantly move the focus of the camera to keep a specific point on this person, say their lapel mike, centered in the frame? If not, would you like him to, or would you suspect that it might result an annoying amount of camera movement for someone watching.

Not to say that you couldn't address some of these things with signal processing tricks, just be prepared for this project to end up getting much more involved than you are probably imagining right now.

 

I think... I would look into using something like OpenCV with a cheap webcam to "track" the person and then use some servos to move the camera around as needed.
Its not a "simple" project at all though.. (unless you are familiar with openCV,etc...)

I'm fairly sure though that some off the shelf webcams have this functionality built in.. Might be worth it to ditch the handycam and use one of those webcams with a laptop or something.

 

What about making a small pin with an infra red LED on it that the speaker wears and then build a mount that sits under the camera with a lens and 3 IR photo transistors. The mount rotates to keep the IR emitted from the pin on the centre photo-transistor, if the right gets more light rotate left, if left rotate right.

It's not too complicated an approach but might be limited on range. The infra red sensors/receivers are generally available but need modulating so that IR to be detected can be separated from background infra red.

 

The same issues still apply.

Do you really want the camera moving constantly just because the speaker has this annoying habit of breathing and thus constantly moving the small pin that they're wearing? What happens when the system loses signal because the speaker raises their arms and covers the pin or when they turn to face the board?

 

Just a thought, and I'm not sure how it can be implemented, but I think there is some info out there on using an Xbox kinect to track movement. In one of my classes a group proposed a project and demonstrated tracking a human body using a program available online. Not sure which program or whether or not it's free but may be close to what you need. The difficult part of the hardware would already be designed and it would just be a matter of implementing code to signal camera movement.

 

The same issues still apply.

Do you really want the camera moving constantly just because the speaker has this annoying habit of breathing and thus constantly moving the small pin that they're wearing? What happens when the system loses signal because the speaker raises their arms and covers the pin or when they turn to face the board?

yes. wat u said regarding the movement is a real trouble. but what if the sensor is being pinned on to this shirt or on to the pocket. and i think there may be no acrobatics from the speaker.

 

yes. wat u said regarding the movement is a real trouble. but what if the sensor is being pinned on to this shirt or on to the pocket. and i think there may be no acrobatics from the speaker.

I never new "breathing" was an acrobatic movement that a speaker could avoid.

But, seriously, I suspect you are likely to find that the small, constant movements that will result from rigid tracking of something the speaker is wearing will be very annoying. It doesn't take much. Also, as I pointed out, there are ways of dealing with it. Just don't dismiss it and then find out down the road that this "simple" project is turning into a time and money sink.

 

What about making a small pin with an infra red LED on it that the speaker wears and then build a mount that sits under the camera with a lens and 3 IR photo transistors. The mount rotates to keep the IR emitted from the pin on the centre photo-transistor, if the right gets more light rotate left, if left rotate right.

It's not too complicated an approach but might be limited on range. The infra red sensors/receivers are generally available but need modulating so that IR to be detected can be separated from background infra red.

i think that would do. but wat i fear is that there will be so much artificial lighting tat can cause the IR rays, from some other source. but i'm not sure though.

 

i think that would do. but wat i fear is that there will be so much artificial lighting tat can cause the IR rays, from some other source. but i'm not sure though.

As was pointed out when this was suggested, you need to modulate the IR signal so that you can discriminate the other sources. There are small, cheap modules that will do this. The problem you have to contend with is the possibility that other devices in the room will be using the same approach at the same modulation rate (there are only a couple common rates and they are close to each other). But there are ways of deconflicting them.

 

ok. So where do i start from. i've got a Handycam with a tripod.

 

ok. So where do i start from. i've got a Handycam with a tripod.

Probably the first thing is to forget about the sensor and the tracking and get something put together that can move the camara in response to signals of some kind.

To do that, you need to start identifying the bounds. Pick the spot that your camera will be located and then determine what envelope the speaker has to work with -- how far can they move, high high or low can they go -- and determine how much range of motion you need the camera to have.

Then the next thing is to decide what kind of mechanism would work best for carrying out the motions. There are almost certainly commercial items out there that you should consider before making your own. But with either approach, one obvious way to consider is something that controls the rotation of the camera about the tripod's vertical axis and a separate something that controls the tilt of the camera about the camera's lateral axis.

 

Just to respond to the point about the speaker making small movements: It is obviously quite possible to introduce some hysteresis either aa a time delay so that the camera only moves after a short time if the speaker continues to be off centre; or use more sensors or space the sensors more widely so that you only respond if the speaker wanders too far off centre. You probably need some combination of these.

For driving the camera it may be worth looking at the drives available for astronomic telescopes although these may move too slowly. That said it amy give you some ideas about how to build one. Getting smooth motion is going to require acceleration/deceleration.

Someone suggested usng a Kinnect - not a bad idea but I believe the range of these is limited. However you could put teh Kinnect on the lecturn or front of the stage and assuming teh speaker does not move mor than a metre or so then it may work

 

My first inkling would be to try a PID controller and adjust the gains so that it is biased against rapid moves, requires a modest error to drive a prompt response, but still slowly gets rid of residual error.

As for the Kinnect -- if it only has a range of a meter or so, then that is almost certainly not good enough. In most situations, any speaker that is so rigid that they never move more than an arm's length from where they started is probably not a speaker that people are going to be paying a lot of attention to. There are exceptions and certainly situations in which a speaker would be firmly tied to a podium, but for case all you have to do is expand the field of view just a bit and use a static camera. This sounds like the opposite situation -- tracking dynamic speakers that exploit a large portion of the stage area.

 

Just to clarify, in my experience Kinnect seems able to track one person something like 1.5 to 2.5m from the device or two people 2.5-3.5m from the device (I haven't measured this, just estimating from room size). Programming it yourself and under the best lighting conditions it may be possible to do better than this

 

What about using PIR sensors? The resolution generally seems quite low – the ones on my external lights only come on if I move a good few feet but I guess a number of them could be mounted along the stage to detect position.

Another idea depending on stage layout would be to put an ultrasonic sensor at the side of the stage and measure the distance to the speaker from one side of the stage (probably needs an array of sensors because the speaker will move forwards/back)

 

Maybe this will work for you:
http://www.roborealm.com/

You can try it for free for 30 days.

I have seen demonstrations of this software and was very impressed. One example was solving a Rubic's cube. The setup used a laptop computer, LEGO Mindstorm and a camera if I remember correctly.

Good luck with your project and please let us know how things go.

 

Try setting up the camera further away from the subject being filmed. (or a better camera). You would have a better chance of keeping the subject in focus regardless of movement...or you could activate the auto-focus feature on the camera.

 

Hello guys,

I need to set up a motion sensed camera movement for my sony handycam. the most hard thing i face is , during stage events, at my college the spokesman, moves around the stage, this makes the camera guy to adjust, the hadycam, to the direction he moved. I am dreaming of a sensor connected to the spokesman, which the handycam detects and the camera's direction is focused on to that specific location.

Need the help from basics , as i'm not too much into circuits

ThanX.

It may be that I'm not understanding your problem entirely, but I have something of a similar situation going on here. In my case, the sensor(s) will be for the purpose of perceiving, tracking and photographing would-be thieves in my fenced-in backyard, where I have an aquaponics garden, a hydroponic garden, and others. A food supply - especially where human thieves are concerned of organically grown food - has already drawn much attention from wildlife, and with the economy as it is and where it's going, human thieves will also be a problem. Just the fact of so much "hockable" equipment in the yard is a temptation.

I'm experimenting with garage door opener and closer lights, that due the fact that continual winds here continually - just about constantly - turn on other, more conventional sensors. Directed along very close to the ground, trees, plants, and more blown by the wind do not trigger the sensors. My arrangement, of course, is much simply than yours, but with an array of such lasers or light beams otherwise, a computer program should be able to track a speaker anywhere he goes. With the beams arranged at different levels and angles, any movement he - or animals makes can be followed.

But, as I said, maybe I don't completely understand what you need. All I need is a picture of the interloper and a warning inside the house that he's there. You seem to need much more.

 

What about making a small pin with an infra red LED on it that the speaker wears and then build a mount that sits under the camera with a lens and 3 IR photo transistors. The mount rotates to keep the IR emitted from the pin on the centre photo-transistor, if the right gets more light rotate left, if left rotate right.

It's not too complicated an approach but might be limited on range. The infra red sensors/receivers are generally available but need modulating so that IR to be detected can be separated from background infra red.

Don't forget the camera will "see" the IR led, and it'll appear as a spot of light pointed at the camera.