Thank you Luis,
having also hints about a preferable solutions for Kurento is useful because we are still in a definition phase for live demo.
Just a note to clarify one of my assertion, obviously the robot is able communicate with external world, but we have to consider three independent channels for information exchange:
· one is carrying ROS messages and enable ROS services to exchange Commands and sensors data (this is based on ROS communication), and it operates through FIROS Robotics component and Context Broker GE; his is a sort of "Instructions plane".
· another is used for the Service Logic management (e.g. to create, delete ROS nodes and configure ROS service spaces), and operate with RCM Robotics component (able to communicate with FIROS); this is a sort of "Management Plane";
· the last one is the communication of a media stream to a Kurento Client toward Kurento Media Server; this is a sort of "Media plane".
All these three information channels should be able to send / receive information/data to one (at least) Web/Application servers but they are independent. Specifically, ROS communications between nodes in Robots and Servers (for commands and data) should be considered "isolated" and independent from a media stream channel toward the Application Server hosting Kurento client and to Kurento Server.
This is not an issue for Kurento, but something we (Robotics Group) have to consider to define a consistent a functional architecture for the Robotics Live demo.
We'll have to investigate if the Application Server used to host the demo application to Command and Manage Robots can be used also to host Kurento Client, if we decide to have the App server external to the Robot.
Best,
Gianmario
Da: Luis López Fernández [mailto:luis.lopez at urjc.es]
Inviato: giovedì 19 febbraio 2015 09:54
A: Bollano Gianmario
Cc: Antonini Roberto; fiware-robotics at lists.fi-ware.org
Oggetto: Re: [Fiware-robotics] Robot characteristics
Hi,
Some comments inline
El 18/02/2015, a las 12:41, Bollano Gianmario <gianmario.bollano at telecomitalia.it<mailto:gianmario.bollano at telecomitalia.it>> escribió:
Hi Luis,
I have found initially of documentation from the FIWARE pages and it was quite complex to understand where to get clear code and examples. Probably the catalog page it's not the simplest page to start with, because it provides lot - for me too many - links for Kurento server and Client at Kurento Github<https://github.com/Kurento>, at NPM<http://npmsearch.com/?q=kurento-client>, at Maven Central<http://search.maven.org/#search|ga|1|kurento>, Bower<http://bower.io/search/?q=kurento-client> and Kurento FIWARE Forge Repository<https://forge.fiware.org/frs/?group_id=7#title_data-kurento> ... Really hard to understand which component may be needed and where to get them.
Information at Kurento.org<http://Kurento.org> is definitely more helpful; without digging deep in the technical details, from the Tutorial it is clear there are three different "usage scenarios" , also represented in the picture in the introducing_kurento<http://www.kurento.org/docs/current/introducing_kurento.html> tutorial ("Kurento Api, Client and Protocol" section).
In my understanding, these 3 scenario mapped to our environment becomes:
1. Robot equipped with Chrome + NodeJS + Bower + Kurento JS Client + Application Code. This allow to have Application server as localhost in the robot and to open the browser for transmitting video in the same environment. Having everything in the Robot allows to communicate directly with a Kurento media server. The Browser endpoint connects to Kurento server.
2. Robot with Chrome only, which connect to on external Linux App server with NodeJS + Bower + Kurento JS Client + Application Code (downloaded and launched in the Robot browser).
3. As 2, but working in Java EE App server and Java Kurento client instead of JS.
I believe the scenario 3 beyond the purposes of the Robotics live demo because it requires larger development effort and Java EE knowledge, however the Robotics developers may better evaluate this issue. The scenario 1. is preferable but we have to check if the Robot has the requirements to host all the Application Server software.
I think you should go to scenario 2 or 3 and use an Application Server external to the robot. You can base the Application Server on Java EE or in Node.js, but definitely you need that Application Server to be external to the robot. Otherwise, you limit too much the use cases you may be supporting. We can discuss about this slowly if you want, but putting the AS in the robot is not the best design decision and there are strong reasons for it:
- You won't be able to put the robot behind a NAT easily and connect from external applications (browsers) to the Application Server.
- You will be putting CPU overhead in the robot
- You will make the application too sensitive to robot network access.
IMHO, you should go to option 2 or 3 and put in a FIWARE Lab node both, the Kurento server and the application server. You have may examples on doing it in Kurento.org<http://Kurento.org> tutorials. It wont' hurt you and you'll benefit from lots of advantages.
The scenario 2 move the Application server outside the Robot and is less resource consuming for the Robot, but it involves further issues concerning communication between Robot and the Application server.
You will just need to put the AS into a well known IP.
Robotics GE adopts ROS environment which already consider a Robot Entity composed of nodes communicating in their way, same of them placed externally to the Robot (in a service Space which runs also in an external server). Moreover, our demo also use an external application server to send and display Robots commands, which interact with Context broker on one side and calls Robotics management APIs, but the Robot ROS environment (between ROS nodes) is isolated (also for Robotics security reasons) and cannot communicate with that external server. This issue should be better investigated because it may be important also for the scenario 1.
If you cannot communicate with the external world you won't be able to see any video.
The talk.io test may be come handy to test if WebRTC support from Chrome is working fine in the Robot, but does not require Kurento. The next steps will need to use a configuration including a Kurento Client (thus using Kurento protocol) and we'll need an instance of Kurento Server. Do you think it would be possible in the next phase to use a FI-LAB instance of Kurento to make some test with a Kurento Client? The solution of installing a local Kurento Server wouldn't probably simplify the communication issues with the Robotics platform, and probably would have less impact for Robotics and Kurento GEs demo. For the same reason we are using an external instance of the Context Broker: one of purpose of the live demo is to show interoperability of different FIWARE GEs in the Cloud.
We are having a lot of problems with the FIWARE Lab. You can launch a Kurento server there and it works sometimes ... but sometimes it doesn't.
We are still defining the Live demo, and at this stage it's not yet completely defined if and how exploiting the Kurento advanced features (beyond the basic video streaming supported by WebRTC). One idea might be working on some pattern or face detection when we use an Infrared camera. This solution would obviously imply the usage of Kurento equipped with some processing function among the ones already developed and already made available with Kurento (no need to have ad-hoc functions, but use of one of the Kurento processing components available).
This is my general understanding and overview; all the other Robotics in the mailing list guys have deeper knowledge of the technical details and specifications of the Robotics Platforms GE components, they can arise more precise proposals or questions.
Thank you,
Gianmario
Da: Luis López Fernández [mailto:luis.lopez at urjc.es<http://urjc.es>]
Inviato: martedì 17 febbraio 2015 19:05
A: Bollano Gianmario
Cc: Antonini Roberto; fiware-robotics at lists.fi-ware.org<mailto:fiware-robotics at lists.fi-ware.org>
Oggetto: Re: [Fiware-robotics] Robot characteristics
Hi
Answers inline
El 17/02/2015, a las 16:47, Bollano Gianmario <gianmario.bollano at telecomitalia.it<mailto:gianmario.bollano at telecomitalia.it>> escribió:
Hi Luis,
I think the questions arisen from Roberto and Davide can be summarized as:
- Assuming to use a chrome browser, do we need a specific "Kurento Client" code (e.g. a JS library for Kurento APIs) to call a Kurento function and transmit a video streaming to the media server?
The short answer is no. The long answer is that it depend on the architecture of your application. Probably, the simplest way for you to start working with kurento is following one of these tutorials:
1. Read first this and pay attention to the role of the application server (node or java based)
http://www.kurento.org/docs/current/introducing_kurento.html
After that, my opinion is that you should go to an architecture based on application server, so the browser does not speak directly with Kurento Media Server. You can find several tutorials of applications basing on such architecture. For example
In java
http://www.kurento.org/docs/current/tutorials/java/tutorial-1-helloworld.html
In JavaScript (for Node.js)
http://www.kurento.org/docs/current/tutorials/node/tutorial-1-helloworld.html
For making something more useful (e.g. making possible to robot to broadcast the video to a number of receivers), take a look to these tutorials:
In java
http://www.kurento.org/docs/current/tutorials/java/tutorial-3-one2many.html
In JavaScript (for Node.js)
http://www.kurento.org/docs/current/tutorials/node/tutorial-3-one2many.html
- Would it be possible to have a simple code snipped to try to transmit a video stream to Kurento and use it to check if it works immediately in our robot (equipped with chrome and linux)?
Install one of the tutorials listed above and open them from the robot using a chrome browser. That should do the job.
The point is not WebRTC, I think the different options for WebRTC clients are clear now, but instead identifying the simplest code (Kurento client library, snippet code, etc) required to make a quick test and transmit a video stream from a Robot to a real Kurento media server.
If you were able to execute the simple peer-to-peer example I provided in a previous mail (basing on talky.io for example), then you won't have problems for interoperating with Kurento.
This will help to see if the Robot environment is sufficiently compliant with WebRTC and Kurento requirements, and the robot have enough resources (CPU, memory) to allow a realtime transmission of a video stream from a webcam.
Again, you can check this with the basic peer-to-peer example based on talky.io without requiring Kurento at all for it.
Best.
Gianmario
Da: fiware-robotics-bounces at lists.fi-ware.org<mailto:fiware-robotics-bounces at lists.fi-ware.org> [mailto:fiware-robotics-bounces at lists.fi-ware.org<mailto:robotics-bounces at lists.fi-ware.org>] Per conto di Luis López Fernández
Inviato: martedì 17 febbraio 2015 16:20
A: Antonini Roberto
Cc: fiware-robotics at lists.fi-ware.org<mailto:fiware-robotics at lists.fi-ware.org>
Oggetto: Re: [Fiware-robotics] Robot characteristics
Hi Roberto,
Please, take into consideration the following information for having an understanding on how to use Kurento:
Kurento is:
- A server infrastructure (it receives streams from clients and "do things" with those streams such as transcoding, processing, recording, distributing, etc)
- A WebRTC capable server (it's capable of receiving WebRTC streams, among other types of streams)
Kurento is not:
- A WebRTC client (it cannot "open a webcam" and send the stream to a remote party)
Hence, in your architecture, you need two ingredients:
- A WebRTC client catching the stream from a Camera and sending the stream using WebRTC standard protocols
- A WebRTC server, for receiving the stream and "do things" with that stream.
The "WebRTC server" part is already solved by Kurento and you'll not need to do much for integrating with it.
The "WebRTC client" part is not solved by Kurento. You'll need to have your very own client and in my last e-mail I just provided different alternatives that you have. Kurento is not one of them because Kurento is not a client.
I hope this clarifies the issues.
Best.
El 17/02/2015, a las 16:12, Antonini Roberto <roberto1.antonini at telecomitalia.it<mailto:roberto1.antonini at telecomitalia.it>> escribió:
Hi Luis,
thanks for having promptly answered my e-mail .
Amongst the solutions you've proposed, it's not clear (at least to me) which of these are using KURENTO GE.
I figured out that you provided us with some code snippet (with apposite KURENTO API usage) to build a client and connect your GE from our robot, I guess these are only a test cases to validate WebRTC technology on the scenario we shared, so I suppose I miss the link between these tests and KURENTO.
Anyway, just to better clarify what we meant to do:
On our side, we would like to use KURENTO javascript clients (through Google Chrome) to connect to your KURENTO media server (using WebRTC technology), is it feasible?
Please, let us know which are the steps (and consequently the complexity) to integrate the components above in our scenario, if we understand well.
BR
Roberto and Davide
Da: Luis López Fernández [mailto:luis.lopez at urjc.es<http://urjc.es>]
Inviato: giovedì 12 febbraio 2015 16:29
A: Antonini Roberto
Cc: Garino Pierangelo; Colombatto Davide; fabio.dibenedetto at consoft.it<mailto:fabio.dibenedetto at consoft.it>; Bollano Gianmario; Angel Hernandez; Iñigo Gonzalez (igonzalez at ikergune.com<mailto:igonzalez at ikergune.com>); Jose Jaime Ariza (jjaime at ikergune.com<mailto:jjaime at ikergune.com>); Fernando Herranz (fherranz at ikergune.com<mailto:fherranz at ikergune.com>); fiware-robotics at lists.fi-ware.org<mailto:fiware-robotics at lists.fi-ware.org>
Oggetto: Re: Robot characteristics
Hi Roberto,
basing on the specifications you are providing, my recommendation is to integrate WebRTC support into the robot in a native way. For this, you have several options.
1. Use the WebRTC Google stack.
* This is the most mature WebRTC stack providing lots of advanced features
* It's the stack used by Chrome
* Can be compiled and used independently on the browser (with some work)
* Only supports VP8 codec. Supporting H264 requires some complex integrations
* Details of it can be found here: http://www.webrtc.org/
2. Use the WebRTC Ericsson Stack
* Less mature but also less heavy
* Supports natively H.264, which is a more performant codec for your use case.
* Details can be found here: http://www.openwebrtc.io/
Here, my choice would be option 1, but you may wish to perform a deeper analysis.
For validating that you webcam works correctly with WebRTC, do the following:
- Provide internet connectivity to the robot
- Install chrome on it
- Open this URL
-- https://talky.io/
- Fill a string in the form (e.g. "aaa") and press the button.
- Accept to share the webcam
- The video should appear
- Open another browser into another computer with Internet connectivity and a webcam
- Open the url https://talky.io/aaa (where "aaa" is the string you filled previously in the form)
- Accept to share the webcam
- You should have a video conference between the two browsers
Best.
-----------------------------------------------------------
Luis López Fernández
Subdirector de Investigación y Relaciones con la Empresa
Escuela Técnica Superior de Ingenieros de Telecomunicación
Universidad Rey Juan Carlos
http://www.etsit.urjc.es
e-mail: luis.lopez at urjc.es<mailto:luis.lopez at urjc.es>
Tf1: +34 914 888 747
Tf2: + 34 914 888 713
El 11/02/2015, a las 13:54, Antonini Roberto <roberto1.antonini at telecomitalia.it<mailto:roberto1.antonini at telecomitalia.it>> escribió:
Hi Luis
As agreed, we start sending to you the characteristics of the robot we have here, that's a turtlebot2 and the PC on board has Ubuntu 12.04 OS (more details available athttp://www.clearpathrobotics.com/turtlebot_2/tech-specs/))
The camera we intend to use is a Logitech c270 (more details available at http://www.logitech.com/it-it/product/hd-webcam-c270).
Let us know if this's enough, we wait for instructions on how to setup a WebRTC video streaming, in order to solve technical issues and test the end-to-end communication.
Thank you.
Davide and Roberto
Questo messaggio e i suoi allegati sono indirizzati esclusivamente alle persone indicate. La diffusione, copia o qualsiasi altra azione derivante dalla conoscenza di queste informazioni sono rigorosamente vietate. Qualora abbiate ricevuto questo documento per errore siete cortesemente pregati di darne immediata comunicazione al mittente e di provvedere alla sua distruzione, Grazie.
This e-mail and any attachments is confidential and may contain privileged information intended for the addressee(s) only. Dissemination, copying, printing or use by anybody else is unauthorised. If you are not the intended recipient, please delete this message and any attachments and advise the sender by return e-mail, Thanks.
<logo Ambiente_foglia2.jpg>Rispetta l'ambiente. Non stampare questa mail se non è necessario.
<logo Ambiente_foglia2.jpg>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.fiware.org/private/fiware-robotics/attachments/20150219/34a757a2/attachment.html>
You can get more information about our cookies and privacy policies clicking on the following links: Privacy policy Cookies policy