Seeking help from Multi-Rotor Experts for (overly?) ambitious project!

[Dr Flux]

Hi, I'm writing from a small design studio ( www.superflux.in ) based in London. We are currently working on an ambitious project using multi-rotors.
It is a performance event happening in early September at a high profile London museum, where (the plan is)  5+ multi-rotors will  fly autonomously in the museum's courtyard.

While we have experience of working with a wide range of technologies and are already working with some talented designers and technologists, none of us has any real experience of building, configuring and flying multi-rotors. Currently we're testing with an F450 clone and an OpenPilot Revo board, but we also have a 3DR/PX4 PixHawk board to test.

However we're running into quite a lot of unexpected problems:

1. Stabilisation problem

Our first multi-rotor with OP Revo hasn't flown stably yet. It flies, but it's extremely difficult to control with a transmitter (Turnigy 9XR) or with external software which mimics a transmitter. When it's in Altitude Hold, it oscillates even though the parameters are quite 'mild';. It also drifts quite a lot.

As this is our first time building our own multi-rotor, we've been doing lots research, reading how to's, forum posts etc. but our only real reference points for how it should/could be flying are YouTube videos and after weeks trying (and going though countless props and arms) we're still nowhere near.

It would be amazing if someone would be willing to take a look at where we're at either via videos / skype or by coming into our studio in Bermondsey, London.
Ideally we're looking for someone familiar with either PX4/Pixhawk or OpenPilot. And even more ideal if someone with that knowhow would willing to spend send a day or two with us doing a 101, helping us dial things in and pointing out any school boy errors we're making. Whilst we don't have a huge budget we would definitely want to compensate you for your time and expertise.

2. Positioning system

The second problem is a little more technical, and maybe outside of what most people in the community are doing, but... 

As the performance is set outdoors, we're thinking the best approach is to build a positioning system based on GPS.
We managed to get a set of Piksi RTK GPS modules ( http://swift-nav.com/piksi.html ), which claims centimetre accuracy. Although this is an application the units were built  for, we haven't managed to get it working with the multi-rotors yet. We think we need someone with a good knowledge of multi-rotor-friendly GPS and a programming skill preferably with multi-rotor systems.

Again, we're happy to talk about ways in which we can remunerate anyone kind enough to help, for their time and expertise.

Please forgive me for posting this without taking time to properly introduce myself to the forum before asking for help but we've just been really busy trying to figure it all out on our own and now time is getting short and we're realising we're desperately in need of help so any comments, suggestions, advise would be hugely appreciated.

Thanks in advance.

[teslahed]

Your issues with lack of stable flight might be down to the frame and quadcopter hardware and nothing to do with the flight controllers at all.

If you wanted to post pictures and technical details of your multirotor i bet you';d get some suggestions straight away.

[orfordness]

They may also be the tip of the iceberg if you are not legally qualified to perform commercial work with the CAA?

BCCu anyone?

Please don';t wing it and think we won';t need insurance etc....

if you have this all sorted, then I wish you good luck as your learning curve is going to be very steep...


Steve

[quadfather]

... none of us has any real experience of building, configuring and flying multi-rotors.

Legality aside, that';s your problem right there.  Building a stable multi-rotor is a fine art, with many, many parameters to consider, e.g. motors, props, sensors, batteries, AUW, desired performance, etc pp.  The flight controller is just one of the considerations.

Not knowing what you want the quads to do, I would suggest an off-the-shelf product, which removes any guess-work, and take it from there.  The Parrot AR Drone comes with many sensors pre-configured, including GPS and Ultrasound, and has an open source API which you could use to control the quad. /me dons asbestos suit for incoming flames

The majority of people here use quads for FPV or aerial photography, and will design accordingly.  If you want a quad ballet you will need an entirely different configuration.

My £0.02

[Jumpy07]

They may also be the tip of the iceberg if you are not legally qualified to perform commercial work with the CAA?

BCCu anyone?

Please don';t wing it and think we won';t need insurance etc....

if you have this all sorted, then I wish you good luck as your learning curve is going to be very steep...


Steve

If its indoors.. bit of a grey area tbh.. and prob not of interest to CAA

[atomiclama]

They say in the court yard and using GPS so outside.

[Jumpy07]

They say in the court yard and using GPS so outside.

Sorry missed that.. suppose they could string some netting above courtyard..  GPS in a courtyard will be  though if walls on all sides..

[barneyg]

If you want that kind of precision then you would probably be better going with camera and external processing system.

[kilby]

The SBAS should make it considerably more accurate than what we are used to but centimetre accurate may be a bit of an exaggeration.

The normal way of doing this is cameras and ir illumination as GPS can be a bit unreliable (to say the least)

[HowDoIfly]

The SBAS should make it considerably more accurate than what we are used to but centimetre accurate may be a bit of an exaggeration.

The normal way of doing this is cameras and ir illumination as GPS can be a bit unreliable (to say the least)

I thought SBAS has been enabled for a while now? But it still won';t give centimetre accuracy.

"RTK gives you centimeter accurate relative positions between the two receivers. To get a position that is centimeter accurate with respect to the earth, one unit must be used as a stationary "base station." To get an absolute position (in a coordinate system such as ECEF), a CORS base station must be used or the base station must be placed at a known geodetic location. A list of these locations in the US can be found here.

If both receivers are in motion, RTK will give you the relative position between the two but not fixed position relative to the environment."

"Piksi is accurate to 2cm horizontally on short baselines (distance between the rover and base station receivers). Vertical precision is typically 2-3 times worse than horizontal precision for GPS receivers, so we expect 6cm accuracy on short baselines. Statistical testing of the accuracy will be available on the website. Due to the nature of RTK, accuracy degrades at a rate of 1mm horizontal and 3mm vertical for each km between the base and rover. "