Mindboards

Right now I'm just daydreaming, because this project would require about four €50-€100 sensors, but I'm pretty sure it can be done. In fact, I found 2 existing project already.

The dIMU seems the best gyro out there, but from what I gather from the 2 projects I found, it's rather hard to get accurate and drift-free readings. However, my slowstick is pretty much self-stabilizing, so I don't think it needs a rock-solid gyro.

Contrary to what you'd think, the elevator is mostly concerned with velocity, while the motor is mainly concerned with altitude. Thinking about it like this frees us from having to keep the plane level, all we need to do is maintain speed and altitude.

The dGPS module provides these two, plus a means of navigation, although I'm not sure about the accuracy and refresh rate. Alternatives would be a Barometric Sensor and a pressure sensor set up as a Pitot tube, still no clue about accuracy.

A manual override is absolutely required. While it might be possible to do at a hardware level or by messing with the NXTServo, in terms of weight, communicating with the NXT would be ideal. Bluetooth is not up to the job, but maybe NXTBee PRO is?

I read the NXT Rechargeable Battery is lighter than AA batteries, is that true? It's surely bigger.

Might I go completely crazy, I could try to use the ultrasonic sensor to coordinate the last 2-3m to and from the ground. I haven't seen anything that has a longer range.

One last thought: Maybe I can get sponsorship for the sensors? Or just make some money...

If you are planning on using the Lego Ultrasonic Sensor, you ought to do some experiments on the accuracy of that device, as well as the behavior at distances greater than 1 meter. My experiments tended to return unreliable results greater than about .5 meters. The folks that wrote this article saw the same things - http://www.tik.ee.ethz.ch/mindstorms/sa ... e=tests_us.

That is for sure. Nothing is leaving the ground without proper testing. However, one thing I notice with most US tests is that they use relatively small walls/balls/boxes.

Whilst surfin da 'net I have come across autonomous flying devices using NXT & gps.

Try googling 'nxt gps model plane aircraft' or similar and have a look at what has already been done.

Mainly this one, AFAICT: http://diydrones.com/profiles/blog/show ... st%3A32733

They mention using a HiTechnic servo controller that has that manual override part covered, can't find it though.

Something else to think about is weight. Have you thought about how you would power this? I'm not sure if you can get enough power out of the NXT; with the NXT, and necessary batteries, you might reach a weight limit that requires a pretty heft UAV. Just a thought, if you haven't started calculating this out.

The ideal solution would be to power everything, including the NXT, from a 2-cell LiPo. Apparently, the LEGO rechargeable battery is also a LiPo, which pushes that black button inside the NXT, which I suppose has to do with the difference in voltage.

I think a 2-cell 7.2V 2000-3000Mah is quite feasible, although I'm not sure how much power the NXT with all the sensors attached would consume.

Mindsensors confirmed that I could power the servo controller with the BEC from my regulator, yay!

I haven't weighted the NXT sans penlites, but it's not all that heavy, and I have a lot of power to spare. Those huge wings of the slowstick are usually powered by a geared-down speed380, while mine is driven directly by a brushless monster. I normally fly at 1/4 throttle.

For safety, building a gyrocopter could prove interesting. They fly really weird, but the emergency procedure would be simply: shutdown

Another electronics stunt I'm not sure will work out… What if I connected a RC servo PWM signal to the tacho inputs of the motor ports?

In other words, I'm still looking for a lightweight method to have manual control over the airplane.

pepijndevos wrote:...What if I connected a RC servo PWM signal to the tacho inputs of the motor ports?...

You want to connect RC style servos' control pins to the tacho inputs of the motor ports? That would be completely impossible without some major FW changes.

Why don't you use an NXT servo controller? Mindsensors sells one, or you could make your own.

Other way around. I plan to use the Mindsensors controller, but I want to be able to manually override the autopilot. So what I want to do is connect the output of the receiver to the NXT.

In the meantime, I think I figured out this will not work. I don't think the servo output has a phase shift, like the tacho output. I also think it is to fast. I read a cycle takes 1-2ms, and I read the firmware only checks once every ms. But I have no idea how these motor controllers function.

That servo controller of yours looks rather simple to make. Would it be hard to modify that to take servo inputs as well?

It seems your power and jumper(?) are directly connected to the servos, so I assume using a BEC should be fine?