Design Methodologies for an embedded control

[NapDynamite]

Dear all,

My students are on the verge of designing an autonomous air vehicle which could take-of, land, stablize, follow waypoints and return. Since it's autonomous it means no human-in-the-loop (except for downlnking data to monitor orientation and tracking). We have to finalise the "build or buy" aspect for this system and there has been a discussion going on around my peers on which methodology to adopt. I'll just be brief here and will cut to the chase.

In order for this vehicle to fly autonomously, we'd need a flight controller. Either we can design one or buy one. Here are the caveats:

1. If we "buy" one, we have PIXHAWK, runing a built-in flight controller inside ARM processor with a real-time OS (NuttX). Ofcourse the flight control we'd be needing might have augmented controls such as adaptive control, parameter estimation and to do that in Pixhawk we would need to "re-program" it. 

2. The committee suggested that we could use an "off-board" computer talking to pixhawk wirelessly. This off-board computer (SBC, PC, Intel Stick etc.) might be running another layer of control system(outer-loop either in MATLAB/LabVIEW) and sending control commands to pixhawk which, in this configuration, is running as an inner-loop control. This method will not burden students to custom code pixhawk flight controller. 

4. Visualizing this topology sitting on top of the veicle will  add to the weight and dynamics. 

5. Good thing about this topology is that it's not very expensive.

Now my part:

1. Isn't using a computer with a more robust computer, an overkill?

2. I've suggested using myRIO and LabVIEW to program the entire flight control. Students already have designed projects using these tools. 

3. myRIO along with LabVIEW software package comes around for $800-$1000 depending upon university discounts. while the retail price of Pixhawk is $200 and software is open-source free.

4. Designing custom controller in pixhawk may take months considering the organized code architecture and no experience in text-based language. This means more-time to market. 

5. Designing one in myRIO or sbRIO will take less time to prototype and less time to market. 

I hope I'm able to make my point and need sugestions and comparisons on both mthodoligies. 

[Blokk]

I am not in a state to comment on the overall comparison or give advice for the project, but I do not understand this part:

"myRIO along with LabVIEW software package comes around for $800-$1000 depending upon university discounts. while the retail price of Pixhawk is $200 and software is open-source free."

I think myRIO costs around 300 EUR, and LabVIEW comes free for universities under Academic/Student licence if there is one. Of course it is a different story, if you wanna sell your software product. As much as I know, Academic and Student licences are not allowed to be used to create commercial products. But you can ask your local NI reps about these matters, I am not sure...

[elcalverado]

NapDynamite,

I don't have any experience with Pixhawk, so I can't say much as to what it would take to re-program it to get it to the point that you want it. From what I understand, the Pixhawk is a fully developed flight-controller and autopilot. To get the myRIO to the point where it can perform the same as the Pixhawk would take a ton of development. I'm not sure of your timeframe, but I am willing to bet that customizing the Pixhawk would be way less labor-intensive than using the myRIO. But again, I have never seen the Pixhawk software, so I don't know what it takes to program it.

Like you said, no matter what controller you use, you are going to add weight to the aircraft. However, the Pixhawk weighs 36 grams while the myRIO is 193 grams. I believe the current academic price for the myRIO is $500 US.

You said that the vehicle is going to be autonomous. Do you need to be able to communicate with it from the ground mid-flight? It doesn't make sense to me to use a second computer to do the flight control. I agree that it should be done on the vehicle because doing so on the ground would introduce possible lattency and communication problems. It does make sense, though, to use a PC on the ground to send waypoint updates, monitor flight status, etc.

J. Calvert

Applications Engineer

National Instruments