What are the advantages and disadvantages or each?

 

The PX4 looks like the very old days when APM had many boards and no case. Just looking at the advertisement I have no idea why I would want to trade in my APM 2.5+ for a PX4?

 

.

 

Views: 48549

Reply to This

Replies to This Discussion

You need to get beyond the glossy brochure, else you are destined to be mugged.

https://pixhawk.ethz.ch/px4

Umm... maybe the modern 32-bit processor, instead of the 15 year old design the APM uses, or the orders of magnitude greater processing power?

ARM cortex based processors are where everything is going.  No doubt about that, 3DR has already said they're switching over for the next version.  You can have one now or you can wait a few years until the APM3 comes out.

The APM3 in a nice neat package with a processional looking case is what I am waiting for. Sort of a super version of the APM 2.5+ would be perfect. I expect that 3DR is working on the design right now. It would be perfect if it had the same footprint as the current APM. It would be nice if 3DR would give us a hint about future APM 3 plans.

 

Actually I am super happy with my APM 2.5+ so I have no motivation to change at this time.

 

.

 

Hi Jake - I can't see how the 32bit powerful processor translate into a better controller? If it is flying, it's flying.

If someone manages to create an Ardupilot compatible quad core 1ghz 64bit processor - does that mean it's a better autopilot than the PX4?

The Ardu* with PX4 seems very lightly documented and used, while the APM 2.5 is very mature and easy to get started with.

I have 3 APMs and am about to buy a 4th, so I'm at the same junction but like Mike I don't understand the need for the PX4, to be honest.

obstacle avoidance, enough power to collect any logs possible for example.

high resolution optical flow processing for precise movement

ability to run 2 instances of code concurrently for failsafe

I fly mostly airplanes (and one quad) so those features are of no interest to me. I am fairly happy with the current hardware. The firmware can always be better but it has come a long way in the past year.

 

I think that in the future higher-powered processors will be the norm. When there are more compact single board autopilots that look professional like the APM 2.5 and appropriate firmware with increased capability is available then I might be interested. Just look at how fast we progressed from the APM1 to the APM 2.5+. Maybe this time next year it will be worth another look.

 

.

The PX4 software is a product of 5 years of development & is light years ahead of the APM which is neither very mature nor easy to get started with according to many posts on this forum.

"If someone manages to create an Ardupilot compatible quad core 1ghz 64bit processor - does that mean it's a better autopilot than the PX4?"

Yes, the more processing power, the better tuning of the sensor biases & fusion. The APM doesn't have the power for this. Nor does it have the power as was said above to handle things such as obstacle avoidance & vision techniques.

If you want something that works out of the box,  don't like to participate in open source projects, & don't care about getting the best performance, you're better off with a Dji which already has excellent performance for what it is.

@J,

There are always exceptions to the rule. Our present ArduRover2 v2.20b code has very good obstacle avoidance capabilites using the Sharp IR sensors. It works well and I have verified its performance on my 6WD rover. In the final update of the original JLN ArduRover2 code we plan to offer the ability to select several different Sharp IR sensors for medium to long range obstacle avoidance.

Just a thougth.

Regards,

TCIII

This is true. There are obstacle avoidance circuits for rovers that have been around for decades, many which require no computer at all. There's a nice book written on this subject: http://en.wikipedia.org/wiki/Braitenberg_vehicle

It's even got a professional looking cover, but more importantly creative & thoughtful content

http://books.google.com/books/about/Vehicles.html?id=7KkUAT_q_sQC

For primitive stuff that does much with little, the master is

http://en.wikipedia.org/wiki/Mark_Tilden

APM is the opposite--does little with much, & then tries to sell it back to those who made it. Don't get me wrong--I'm not knocking your work in particular, I'm sure you've made it do some cool stuff. I'm just pointing out that this forum is in general insular & filled with extremely dumb stuff like Mike Macs comments on what makes a good machine.

I was speaking of the more general purpose vision based systems which is where things are headed due to the now sufficient processing power available.

One of the things which I find most fun about technology, is that new things come frequently, and in the case of vision--frequently^2. If you are more consumer oriented like Mike Mac, shopping for the system that "looks professional", much of that is missed out on unless you've got tons of cash to spend. You are likely to end up with a system that appears professional above all those things which actually matter.

"ability to run 2 instances of code concurrently for failsafe"
That seems great if we can also double on hardware failsafe.

As for optical flow, it still seems amazingly limited in height and types of ground you are flying, so maybe in a room that has 10 ft ceilings and you are flying over a carpet it will work great... for what I have no idea (yet) and also includes the painfully crazy sonar. Outside missions seem tenable at best. Obstacle avoidance too seems far far away, let's see and remain hopeful!  

How about folowing mode, when aircraft recognize you for example by IR beacon or your clothing color. 

More important thing for the moment is motor feedback. When controller really know how fast your motor spinning and how much amps each motor draw in the moment. It is crucial when we speak to failsafe. Hexacopter can change its mode to tricopter in case of one motor failure or increase speed on ramaining motors.

The posibilities are endless, it is only question of processing power and ability to realise it in code.

Reply to Discussion

RSS

© 2016   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service