QuadCopters

yourmoo

Re: QuadCopters

Postby yourmoo » Sat Oct 23, 2010 7:52 am

Hi Michael,

Thanks for your reply, I can understand the power system you are using now it make sense to me. I thought of taking the same approach where the brains and muscle have 2 saparate power source too.

HAHAHAHAHA funny one on the payload question. i was actually thinking of using a smaller Brushless motor. so need to know to the trust from a smaller motor will be sufficient for stable flight :lol:

I have 2 more question i hope you don't mind helping me with.

1. Notice you have Phidgets Voltage sensor on your quadcopter. How does it apply to the quad copter? or what is the reason to use the voltage sensor?

2. I also notice that your ESC does not get it's power source from the 1061 board. I notice most ESC will need to get the power source from the red/black/white wire but yours the red wire was removed. Is this a special ESC or you redirecting the power source to the battery?

Thanks for your time, And i love your quad copter good engineering.

sashikumar

RBBM
Phidgetly
Posts: 19
Joined: Mon Nov 23, 2009 6:03 am
Contact:

Re: QuadCopters

Postby RBBM » Sun Oct 24, 2010 1:47 pm

Hi Sashikumar

The voltage sensor is only there for observing purposes. We wanted to observe the voltage drop during operation and to define a threshold at which the software cuts power to prevent the drone from falling due to low voltage.

Concerning the ESC, it is powered by the battery packs. The red wire can provide power to the receiver in conventional RC objects, not the other way round.
When you use the Phidgets Servo Controller, it is able to power the servos it is connected with via the red wire (using the 12V DC input). The white one provides PWM signals to control the orientation of the servos. As our ESCs also respond to the same PWM signals, we can use them with the Servo controller, only without the power connection. In fact we don't need the 12V input at the Servo controller and so it is solely powered via the USB connector.

Concerning the power system, I hope you understood my description. It is right, we divide the whole system into two subsystems (electronics and engines) but these are connected to a single battery pack parallelly. The battery pack in our case consists of two LiPo packs that are connected parallely themselves. So we use one energy source for everything. The advantage of this configuration is that a voltage drop at one LiPo pack does not result in a deactivation of one of the two subsystems or, a much worse case, in a one-side energy loss at the engines.

Just for information, you write about brains and muscles, what are you planning to build?

I hope that was helpful

Greetings from Lower Bavaria
Michael (RBBM)

yourmoo

Re: QuadCopters

Postby yourmoo » Sun Oct 24, 2010 9:02 pm

Hi Michael,

Once again your explanations gives clarity to my mind.What i meant when i say brains and muscle is actually the brains (phidgets SBC and extensions) and Muscle (the brushless motor).

I am actually planning to build a Quad rotor also. But i am going to fix in more sensory equipments like 4 ir sensors around the machine for 360 degree proximity sensoring. sonar top and bottom for landing and take off accuracy inside buildings and maybe a usb camera for sight.

I am going to make this as a framework for future developments like find it self through maze. Then i am going take what i have learn from the quad copter and implement in other vehicles like hover craft, hexapods and etc. As now i am only a beginner i shall start with the sensory first move to mechnical and then intelligence.

My software is going to be in Java, but it looks like C++ is also needed may take some time with that. But i will endure.

i also hope to put up a website for my project progress, i will let you know once it is up.

Again thanks for your information.

sashikumar

yourmoo

Re: QuadCopters

Postby yourmoo » Mon Nov 01, 2010 8:08 am

Hi Michael,

Has your quad copter made it's maiden flight?

Sashikumar

RBBM
Phidgetly
Posts: 19
Joined: Mon Nov 23, 2009 6:03 am
Contact:

Re: QuadCopters

Postby RBBM » Mon Nov 01, 2010 11:05 am

Hi Sashikumar

you cannot really call it a maiden flight, but it did a liftoff from the table where it was standing. As you can see in the video on our homepage, we secured it with wires to stabilize the roll and pitch oscillations. So we proved the ability to lift its own weight, but it is far away from a stable flight.

This is because we encountered a big problem with the accelerometer values we want to use to stabilize it. There is a significant amount of noise due to vibrations and because of this noise our PID elements delivered a wrong feedback. The result was an extremely instable flight behaviour.

Our task in recent days was to record live IMU data and analyse it. The first solution ist to filter it with a digital low-pass filter to extract the real data. The next task would be to test the filter on the SBC. Finally there will be an additional Kalman filter that uses both accelerometers and gyros to calculate the orientation.

I attached a small plot to this comment to visualize the problem. The comments are in german, but I hope you can interpret it. The plot shows the raw data from the IMU (red) and two stages (first stage green, second blue) of filtering.

Greetings from Lower Bavaria
Michael (RBBM)
Attachments
27langProcessedPitch95.png
(30.58 KiB) Not downloaded yet

User avatar
Patrick
Lead Developer
Posts: 3078
Joined: Mon Jun 20, 2005 8:46 am
Location: Canada
Contact:

Re: QuadCopters

Postby Patrick » Mon Nov 01, 2010 12:13 pm

Here at Phidgets, we're very interested to see this working. The idea of IMU/Helicopter was one of our motivations for the Spatial 3/3/3, but we haven't had a chance to implement our own yet.

-Patrick

yourmoo

Re: QuadCopters

Postby yourmoo » Mon Nov 01, 2010 9:44 pm

Hi Michael,

For the vibration i read some where hollow boom for the motor may be a contributor, consider dampening the booms maybe? as for the calculation i will need to read up about it to see if i can help in anyway. sorry as this is the only knowledge i can offer for now.

Hi Patrick,

I was wondering if there is any plan to have a smaller SBC? where the USB uses pins instead of connectors? this might reduce the weight more...just a suggestion.

sashikumar

RBBM
Phidgetly
Posts: 19
Joined: Mon Nov 23, 2009 6:03 am
Contact:

Re: QuadCopters

Postby RBBM » Tue Nov 02, 2010 7:25 am

Yes, and USB 2 would be great too...

Concerning the damping of the vibrations, we already disconnected the sensors from the frame by mounting the sensor area onto four rubber absorbers (30mm). The result can be seen on the plot which is attached to my previous post. The noise is not that big. We are optimistic about our approach to filter it so far. The low pass filter works and needs to be tested on the SBC. The next step will be to record the filtered data and test the PID elements.

Concerning the Kalman filter, I need to read about it myself but my time is limited due to an upcoming exam. After that I will create a description of all filter mechanisms at the website.

Greetings from Lower Bavaria
Michael (RBBM)

User avatar
Patrick
Lead Developer
Posts: 3078
Joined: Mon Jun 20, 2005 8:46 am
Location: Canada
Contact:

Re: QuadCopters

Postby Patrick » Tue Nov 02, 2010 12:10 pm

I agree that USB 2.0 would be nice - but for working with Phidgets totally unnecessary, it comes in handy when you want Video/USB WiFi/Audio/etc. Remember, that if the PhidgetSBC does not meet your requirement - virtually any ARM based SBC that runs linux and has USB Host should be able to work with Phidgets - it's just a matter of cross compiling the library. I believe that gumstix has a board with USB Host, and their board are tiny.

The PhidgetSBC is mostly a convenience factor for our users - easy Networked Phidgets, with the option for embedded code for the more advanced users, and we are quite proud of it :-).

-Patrick

RBBM
Phidgetly
Posts: 19
Joined: Mon Nov 23, 2009 6:03 am
Contact:

Re: QuadCopters

Postby RBBM » Mon Nov 15, 2010 2:58 am

Hi Patrick

taking a gumstix board would be a non-sufficient solution for building a Phidgets-based quadrotor :-)

By the way, we managed to filter the incoming IMU data digitally by creating a FIR filter of 99th order and applying it to the recorded data. You can see the result in the attached file. The filter is built using the window method. For the first version by applying a rectangular window. We will test other versions too, especially concerning the performance on the SBC, but for a first try and for the first manually-built digital filter the result is good.

The plot also shows a third curve. This is our reference curve to evaluate the time delay our filter builds up. It is created by applying a fourier transformation to the raw data and cutting it at low frequencies. That means low frequencies are not touched and the ones beyond a certain thershold are simply set to complex ZERO.

Comments are welcome

Greetings from Lower Bavaria
Michael (RBBM)
Attachments
RollComparison.png
(26.29 KiB) Not downloaded yet


Return to “Robotics”

Who is online

Users browsing this forum: No registered users and 1 guest