Alternatives to DJI Vision Positioning System???

[gatorlane]

With release of the DJI Inspire and the announcement of the new Phantom 3, I can't help but marvel at the idea of the Vision Positioning System. So much so that I've considered purchasing the Phantom 3 simply because of this particular feature (but let's be honest - the included lightbridge is pretty awesome too).

Despite the awesomeness, I don't want to drop another $1000+ on a quad that I don't need. I do a ton of indoor filming and it can be a real challenge sometimes without GPS. I use a Naza M V2 on my hex and a KK2.1.5 on my quad.

Does anyone know of any alternatives on the market to this Vision Positioning System?

 

[Bartman]

from what i've heard DJI's vision positioning system was copied from an existing open source system that did basically the same thing. maybe it was related to the APM projects? you'd have to do some research to find it but I doubt the basic functionality can be added to any existing flight control system except maybe the Pixhawk if was indeed the APM community that developed it.

the new RTF quad from 3D Robotics is being reported to have the optical positioning system.

 

[rhsant]

Hi guys! I'm new here. I do research on quadcopters, and I've come across other implementations of the Vision Positioning System. The basic idea is not too complex - the downward facing cameras on the P3/P4 find distinctive features on the ground and keep track of them to estimate how much the aircraft is moving (this is done using a technique called optical flow), while the ultrasonic sensors are used to measure height from the ground. If you know how much you're moving, you can feed that back into the flight control to counteract that movement

That being said, DJI's system is very tightly integrated with the flight controller and a hobbyist open-source solution would probably not work as well because of things like delays/latency in receiving camera information. Also the flight control system needs to be tuned to respond to these inputs appropriately.

 

[Av8Chuck]

Your assessment of how it works is good, but I don't agree with your conclusion about open-source versus DJI's implementation of this technology. Although DJI's implementation of this technology is well integrated there execution of this technology is not great. This has limited applicability, doesn't work well over smooth or low contrast objects, doesn't work well above five or six feet and the tight integration is actually a negative if you want to use it in ways that DJI may not have intended.

The latter point is a real problem for DJI products in commercial applications. For commercial use operators need systems that are open and extensible so they can be modified to broaden or meet the mission of the user. DJI products are generally pretty good at doing easy to define tasks and although, as evidenced by third party app development using DJI's DTK, open source and DJI are not mutually exclusive but it is antithetical.

It all depends on what problem your trying to solve. If what your doing fits within the very narrow mission parameters of any DJI then it can be a great tool. But if you need something more, don't hold your breath waiting for DJI to help solve your problem.