Hi, I am the Lead Coach/Mentor 1, but when I login FIRST Login | Identity Server 4 , it is saying " No roles have been assigned to this account or no teams are associated with the given roles." Can you help to resolve? Many thanks.
We have "No roles have been assigned to this account or no teams are associated with the given roles."
Welcome to the forums! Have you perhaps logged in and completed the new season tasks for your Dashboard account? After the season rollover, all accounts went into “new season” mode and so all agreements need to be re-signed before your account is considered “complete” and your “role” is validated. That’s the biggest thing people forget to do each season, and is the easiest thing to validate.
If you have, which team is this for?
Thanks for the help. Yes, I finished those registrations. The Team number is 21316
Greetings, and Welcome to FIRST Tech Challenge.
We created the FTC-ML Toolchain to allow teams to learn how to use customized TensorFlow Machine Learning and incorporate it into their robot programming to enhance their capabilities. We launched the FTC-ML Toolchain mid-season during FREIGHT FRENZY and at that time all registered teams received allocation quota to use within the tool. Existing teams can use any unused allocated quota during the offseason - this is because we do not (but more aptly cannot) roll over the season database to the upcoming season until kickoff. The reason for this is our requirement to support existing teams and events for offseason events - if we rolled over prior to kickoff we would lose any team and event info from the FREIGHT FRENZY season (we can only support one active season at a time).
While this is great for existing teams, unfortunately it negatively affects new rookie teams. As a rookie team, your team will not officially exist in our system until kickoff day (September 10) and new team quota allocations will not be issued until that time as well. Therefore, unfortunately the only way to use the tool prior to kickoff is through the self-hosting feature within your own Google Cloud account. This unfortunately is not optimal, as this can be technically challenging if you don’t have previous experience using the Google Cloud.
The GitHub repository for the open-source machine learning toolkit that we host for teams is:
Basic instructions on how to host an instance of the Toolchain yourself is on that github page if you’re up for the challenge. I deeply apologize for this inconvenience, we will look into ways to support rookies using the system “earlier” in the future, but for right now I’m afraid we have no other option.
Thanks. I see. Is the Google cloud the colab?
No, there is no python programming required for training TensorFlow models using our system. We do not use Google CoLab at all.
Google Cloud Services is Google’s cloud hosting system - Amazon has Amazon Web Services (AWS), Microsoft has Microsoft Azure, and others like DigitalOcean provide similar cloud services. Our FIRST Machine Learning Toolchain is written specifically for the Google Cloud Services, and anyone with a google/gmail account gets $300 worth of Google Cloud Services for free, certainly enough to last you a month or two of relatively heavy model use.
I’ll deny I ever said this if you bring it up (haha), but many people create a new Google account and host the service in that new Google account while they’re familiarizing themselves with making custom TensorFlow models. Once you’ve exhausted the $300 credit, you don’t get charged anything unless you want to continue - hopefully by that time you’ll receive your allocation from our hosted service, but to be honest the $300 credit is WAAAY more than any single team gets in hosted allocation. Hosting your own instance of the fmltc project is the way to go if you are able.
Thanks so much. We will give a try.
Hi Danny, after we generate the model, can you point me, what Java code to use the model? The one example comes with the FTC app seems not be able to call the model file.
This is Allison Hua. I work in Coach xlnxzh’s team.
I tried to set up the FIST Machine Learning Toolchain. After running through all the steps required, I could not launch the default service at the recommended URL:
Below is the error message:
My project ID is ftc2131622
There is one setting I am not sure if I have set it up correctly or not, which is the ORIGIN setup in the env_variables.ymal file.
What I have is:
I referenced the information from this page to come up the ORIGIN setting:
When I initially set up this URL in the env_variables.yaml, I believe I was not able to fine this URL from the recommended page.
What do you think my issue is? Is this URL valid? Could it be some permission issue? What would you recommend me to try to overcome this issue?
Let me know if you need more information from me to help diagnosing this issue.
Thank you very much for any comments you may have!
Hi Danny, do you think my URL is in the correct format? thanks!