Bitcraze Forums

Hi,
I've recently seen this video https://www.youtube.com/watch?v=w2itwFJCgFQ where at minute 12.24 is shown a particular mode:
the quadcopter remains fixed in a point and then it's possible to move it like if it is in a viscous fluid.
Can someone help me in realizing this mode for the crazyflie.
Thanks

Hi,
The first thing you will need is a way to position the Crazyflie in space, and that is the hardest thing to realise this demo. In this TED talk he seems to be using a vicon system to stabilise his copters in space. For the Crazyflie you could use kinect positioning as a starting point, you can see this thread that is describing a kinect stabilisation system for Crazyflie: http://forum.bitcraze.se/viewtopic.php?f=6&t=800.
BR

arnaud wrote:Hi,
The first thing you will need is a way to position the Crazyflie in space, and that is the hardest thing to realise this demo. In this TED talk he seems to be using a vicon system to stabilise his copters in space. For the Crazyflie you could use kinect positioning as a starting point, you can see this thread that is describing a kinect stabilisation system for Crazyflie: http://forum.bitcraze.se/viewtopic.php?f=6&t=800.
BR

Thanks for the reply,
and something like this https://www.youtube.com/watch?v=0_Uuf2c ... wRqmCglzy5 (min 0.02)
it is possible to realize?

Well, it seems like further along in the video they say that that mode is just turning on the motors automatically after you throw it. I suppose the visual effect they displayed could be done by using the barometer to lock the Crazyflie at a certain height.

otto wrote:Well, it seems like further along in the video they say that that mode is just turning on the motors automatically after you throw it. I suppose the visual effect they displayed could be done by using the barometer to lock the Crazyflie at a certain height.

Perfect. But the datas (altitude) that I get from the barometer have a very big drift even when the crazyflie is still.
So, do you know some algorithm to get a more precise altitude.
Thanks

We have been playing with throwing and flying the Crazyflie for a long time though we where flying it manually, this is a nice effect to do it with the barometer . It definitely looks possible to do with the Crazyflie.

Do you observe much drift even when the Crazyflie is on a table or only when it is flying?
For flying, one low-tech way we have used to stabilise the gyro is to stuck some foam on top of it. The foam that is present in the Crazyflie box works well, I usually cut a cube of it and stuck it under the battery on top of the pressure sensor.

When standing still on a table, what we have observed with the barometer is big steps that are caused by people opening doors and windows (during the winter I could detect someone opening the entrance door of the building one floor bellow ...). I guess this can be solved by correlating acceleration and pressure information and ignoring steps that are only in the pressure and not in the acceleration. Though this work has not been started yet.

arnaud wrote:We have been playing with throwing and flying the Crazyflie for a long time though we where flying it manually, this is a nice effect to do it with the barometer . It definitely looks possible to do with the Crazyflie.

Do you observe much drift even when the Crazyflie is on a table or only when it is flying?
For flying, one low-tech way we have used to stabilise the gyro is to stuck some foam on top of it. The foam that is present in the Crazyflie box works well, I usually cut a cube of it and stuck it under the battery on top of the pressure sensor.

When standing still on a table, what we have observed with the barometer is big steps that are caused by people opening doors and windows (during the winter I could detect someone opening the entrance door of the building one floor bellow ...). I guess this can be solved by correlating acceleration and pressure information and ignoring steps that are only in the pressure and not in the acceleration. Though this work has not been started yet.

Thanks for the reply,
I observe much drift when it is on a table.
I'll try to implement the second solution