2017 Bridge Connection Issues

[Bradley Boxer]

Today I successfully flashed, re-configured, and re-imaged the bridge (using tool 17.2) and RoboRio to 2017 versions.

When the bridge reinitialized and I connected to it through Ethernet, I was unable to see anything on the Bridge's network. I could ping the bridge at 10.3.3.1, and the bridge gave my computer an address at 10.3.3.94, but when I attempted to go to the bridge's webpage in firefox, I received an error.

This happens both through Ethernet and Wi-Fi. It also happens with both the OM5P-AN and OM5P-AC, on an English version of Windows 10, using both firefox and edge. Restarting the computer and robot does not work.

Does anyone have any ideas? I'm really stumped as to why I wouldn't be able to connect to the bridge's http server or see anything on the network.

Thanks in advance.

[Mark McLeod]

There is no Bridge webserver. It is no longer part of the FRC firmware version.
That's a major change from last season.

[Bradley Boxer]

Oh, I did not realize that. That clears a lot up. Regardless, do you know why the bridge might not be able to see the RoboRio on the network?

[Mark McLeod]

Depends on what the other things are and how they are setup.
The radio probably sees everything, but it's your PC that cannot communicate with the other things, and that's usually because of mismatched IP settings.

Is the radio configured as an AP or as a Bridge?

So, take the roboRIO for instance (I assume that's one of the "things").
First check what IP it has been assigned. The easiest way to do that is connect via USB (so as the leave the Ethernet connection alone). Use a web browser to go to 172.22.11.2. The roboRIO web interface will show what Ethernet IP has been assigned. See if it is compatible with your PC IP setting.

[Bradley Boxer]

Thank you for the help. The issue wasn't with the bridge at all, and I was confused because I thought it had a web interface again this year. Setting a static address for the RoboRio at 10.3.3.2 worked beautifully.

With everything working so far we decided to test our luck and try to get a camera feed on the Smart Dashboard.

We reset our Axis Camera to factory settings and reconfigutred it to a static address of 10.3.3.31. Sure enough, the camera started returning some errors I've never seen before, and the feed refuses to show up on the Smart Dashboard. We are running the Axis Camera Sample example code on our Rio (java), with "axis-camera.local" substituted for our camera's ip, "10.3.3.31".

The errors are:

ERROR: select() to 10.3.3.31 port 80 error 113 - No route to host (TCPConnector.cpp:167)
ERROR: connect() to 10.3.3.31 port 80 error 101 - Network is unreachable (TCPConnector.cpp:175)
ERROR: serve_Rectangle - Too many simultaneous data streams (MjpegServerImpl.cpp:403)

Thanks again in advance for the help.

[Storcky]

We're having a similar problem trying to display the result of our Grip Pipeline on the SmartDashboard.

CS: ERROR: serve_Blur: Too many simultaneous client streams (MjpegServerImpl.cpp:403)

Did you solve it?

Our vision code (in robotInit()):

    	UsbCamera camera = CameraServer.getInstance().startAutomaticCapture();
        camera.setResolution(640, 480);
        CvSource outputStream = CameraServer.getInstance().putVideo("Blur", 640, 480);
        visionThread = new VisionThread(camera, new GripPipelineWithContours(), pipeline -> {
            if (!pipeline.filterContoursOutput().isEmpty()) {
                Rect r = Imgproc.boundingRect(pipeline.filterContoursOutput().get(0));
                synchronized (imgLock) {
                    centerX = r.x + (r.width / 2); }
                System.out.println(centerX);
                outputStream.putFrame(pipeline.filterContoursOutput().get(0));
            }
        });
    	visionThread.start();

[Storcky]

Figured it out. We had GRIP running which was also connected to the webcam.

However, now we're running into a different problem.

CS: ERROR: Blur: PutFrame: 2-channel images not supported (CvSourceImpl.cpp:125)

The CameraServer code doesn't appear to have much documentation on the types of Mats allowed. The filterContoursOutput returns ArrayList<MapOfPoint>. Any suggestions on how to display the results of the pipeline on the SmartDashboard for debugging purposes? Ideally, the output would be the same as the final stage of the GRIP pipeline.

Current pipeline:
blur (Mat) -> hsvThreshold (Mat) -> findContours (ArrayList<MatOfPoint>) -> filterContours (ArrayList<MatOfPoint>)

[bkeeney]

One thing that we found today is that if your GRIP filters are too complex this error can happen too.

One of the other mentors was playing around and had created an extremely complex GRIP filter set and we processed about 30 frames and the RIO came back with this error.

Simplified the filters a bit and now the RIO runs like a champ.