Openag-ui web app build diary


I was using a non-Docker brain, which appears to be part of the issue. Just installed a Docker-brain and am able to access the UI.

The question remains for me though how to access the UI if I’m using the non-Docker/developer-brain.

Thanks for the help everyone!


Have you gotten any further?
I put your code in a software_module and loaded it to the database, and I am not even seeing the rostopic. The software_module_type is there (from the default.json).
Since this is running on the Raspberry, it shouldn’t need to be flashed (to the Arduino). Was there anything you did to get it up and running (besides possibly a reboot)?


@serein I’m impressed you got the camera working… I didn’t know that code was hooked up yet! Can you share the code you used to get it up and running?


@webbhm, @gordonb

Well, that was a long time ago and I don’t recall details out of my head. Let me hook it up again and refresh my memories. Will do this over a weekend.

And just to manage expectations correctly - I was not able to see webcam output in OpenAg UI. The only thing I got is a ROS topic /camera_1/image_raw with some image data posted.


A missing piece of the documentation (at this time) is rostopic management - how to easily connect messages to services. I think this is being done through json files, but it would be nice if this all got moved into the database. First step is to get image data to a rostopic, then persist the rostopic to the database, and finally get the database info passed to the UI (likely an API with curl).
At the moment I will be happy to see how Raspberry side python (camera) scripts get managed, and have their output to to rostopic. I suspect that python scripts are a counterpart to sensors on the Android.


@serein yeah, I don’t expect the UI to work yet (because I think you’re the first to have the backend for images all hooked up). However, once the backend is working, I think the UI should be an easy fix.


@gordonb, am I correct that the problem with images has been fixed already?

I finally was able to find some time to play with recent version of openag_brain (arduino-less setup WIP PR) and see that areal images are being correctly saved to environmental_data_point

They are not displayed in UI though. Since, I assume, UI is hardcoded to some time-lapse accumulation ("http://raspberry/img.jpg?timestamp=1491509454029" class="timelapse--video")


@gordonb @webbhm

How is this coming? I’ve got sensors and actuators up and am interested in connecting ROS topics. How is this done? Where does one set them up?




You would need to add code into the openag_brain repo, mainly under firmware/lib and edit personal_food_computer_v1.yaml, personal_food_computer_v1.launch, and maybe personal_food_computer_var_types.yaml.


Cool! Thank you @spaghet

I see var_types in the openag_brain repo but not in the docker build. Will you be generating a new docker container soon?




The camera is still WIP and the UI displays an image from http://<ip>/img.jpg which is a hack based on fswebcam saving directly to /var/www/html/img.jpg

I tried working on getting the camera image to work on Monday, it looks like there is a ROS topic of sensor_msgs/Image that is being published but I was having difficulty trying to get that to something accessible through the UI (mjpg, etc.)


I think we’re working on getting a build process set up so we can use something like TravisCI to automatically create new docker images and publish to dockerhub.

I’d say since you’re doing a lot of developer work right now, you would probably benefit from doing a local install of openag_brain and pulling the latest changes as they get pushed to github.


Yep, I’m gonna do that. I can contribute a PR on the PFC1 bom build as well. Let’s me get that set up and I’ll ping you back.

P.s. I use Travis every day to publish docker containers to a regionally redundant ECS on Amazon. While we don’t need a ECS deployment, getting the containers to ECR or DockerHub is easy peasy. I trigger builds/publish when we merge PRs to the develop master branches.

If you need help, let me know.



Oh, is that PR already there? I must have missed it. Sorry, I’d be the person to contact regarding the gro-* repos since they aren’t officially maintained anymore but I still try to look at PRs and stuff.

TravisCI is pretty easy to use, I’ve been using it to deploy our wiki theme

I think we had automated builds on openag_python as well, but not on openag_brain yet. We’re working on it though!



No PR yet. I’ll post it after I get things cleaned up.