3D Robotics Selects Fast RTPS for System Infrastructure

Joined
May 7, 2015
Messages
619
Reaction score
494
Location
Washington, DC
  • Like
Reactions: RichWest
Wow that is very interesting stuff. We'll be owned now...

eProsima Fast RTPS is a high performance publish subscribe framework to share data in distributed systems using a decoupled model based on Publishers, Subscribers and Data Topics.
 
So far this actually looks like a great option for devs too. Unless I missed something, it looks like a great way to implement your own real-time application alongside others.

For example, this could be the solution to having multiple video feeds for processing and viewing that a lot of developers have been looking for. Another example would be to identify other nearby drones and avoid collisions with each other. This move seems promising and looks to pave the way for even more clever innovations with the Solo.
 
  • Like
Reactions: RichWest
So far this actually looks like a great option for devs too. Unless I missed something, it looks like a great way to implement your own real-time application alongside others.

For example, this could be the solution to having multiple video feeds for processing and viewing that a lot of developers have been looking for. Another example would be to identify other nearby drones and avoid collisions with each other. This move seems promising and looks to pave the way for even more clever innovations with the Solo.

what do you think about the messaging format? 3dr is contemplating capt n proto vs protobuf, mavlink 2.0 or CDS. i'm not familiar with those message formats except mavlink, but i did take a look at capt n proto. it seems very flexible and secure.
 
Proposed message definition. Canonical message set, capt n proto for format/definition and rtps for delivery. Mavlink 1 is the current system for canonical message set, format/definition and delivery.

ddbb662c2e5822bd89782104286ea7bd3b383fa3.png
 
I'm not familiar with these frameworks either. Of the frameworks you listed, my only exposure is Mavlink.

A quick look at Cap'n Proto: Introduction makes me a little skeptical of the scheme. It states it is secure but also admits it has not gone under a security review but then again, neither have the others. It also mentions that it just adds new fields at the end and zeros out data that is no longer used and the receiver has to do the bound checks themselves. Yes, compression can be used but that just sounds like a lot of overhead to deal with in the long run rather than just focusing on getting things done and in a clean way much like the philosophy of Golang.

As you said earlier, only time will tell to see how 3DR will implement this.

Disclaimer: This is actually my first time I actually looked into a RTPS model so I have much to learn here.
 
So, this will be in some future Solo Enterprise hardware version? I'm taking it this won't mean much for most of us in the consumer space...
 
So, this will be in some future Solo Enterprise hardware version? I'm taking it this won't mean much for most of us in the consumer space...

Most likely an all software solution. I don't see any mention of additional hardware being required.

In the enterprise world, this framework would potentially allow multiple drones to go out and communicate with each other efficiently to cover a large survey site in parallel, for example.
 
Last edited:
can someone explain this to me like I'm 5?
It's almost like it's written in another language...in layman's please.
 
can someone explain this to me like I'm 5?
It's almost like it's written in another language...in layman's please.

I'm in the same boat but here's my understanding so far.

ELI5: You know how there are several computers in every car and each control a specific function? They all communicate with each other in a modular way. The same architecture is going to be applied with the Solo. End goal is a much cleaner code base and more reliable Solo than it already is.

The current model right now with how all the components in the Solo communicate is starting to get bloated and complicated. It will soon become a bottleneck when trying to accomplish more complicated tasks.

What this framework solves is that each component will become modular and can communicate with anything, anyone, anytime in real time. This opens the doors to a lot of potential.

The first link Frank provided details a lot more.
 
Nor I familiar with it but,
This is some deep and exciting stuff here. Huge possibilities we will see coming to life in the next few years. I would hope with the Mozilla Foundation involvement some security issues are present. "a license developed by the Mozilla Foundation. "
 

Members online

No members online now.

Forum statistics

Threads
13,093
Messages
147,741
Members
16,048
Latest member
ihatethatihavetomakeanacc