STAIOT IoT systems workflow issue

[Ian_P]

Hello I am trying to complete my first IOT system deployment on STAIOT Craft using a STEVAL WINBX1 as the target. I have set up the project, created a dataset and model and am following the workflow to set up an E2E system using ST Azure. I get all the way through the workflow and it looks like its going to be successful but at the last step it errors (as below) and so i cannot use the dataset nor inference tabs on the subsequent page.

Setting inference mode... Error: [Errno 2] No such file or directory: '/app/workspace/ai_configuration.json'

I have tried to do this using different SD cards, on Edge, on Chrome, making sure pop ups, redirects and downloads are all set in the browser.

Can you please help guide me through?

[Davide Aliprandi]

Hello @Ian_P,

Thank you for choosing ST products and ST AIoT Craft web platform, I hope you will enjoy experimenting with our ecosystem.

As a matter or fact, the process of creating IoT systems deployments with user own datasets is not yet completed, but we are in the process of adding it soon.

Will provide you with more info on this very soon, will keep you posted.

Regards,

Davide

[Ian_P · Accepted solution]

Hello Davide, and many thanks for your reply.

I had maybe erroneously thought that the workflow would have been completed prior to being circulated.

As a note it's rarely possible to enjoy products that don't work and if I might offer some advice ... making out as though they are, isn't genuine. I would add caution in pretending and would counsel that you should raise this at an elevated forum in ST. If developers can't get a good experience, then neither can consumers. This is central to stock price.

I expect to fall on deaf ears with my reply but in the event you all dont become 'company sided' that you at least reflect. Thank you... its a shame its incomplete.

Regards

Ian

[Davide Aliprandi]

Hi @Ian_P,

Thank you for your feedback, this is valuable for us to better shape the UI/UX of the tool.

I have gone through the whole following flow apparently with no issues:

1. Create a project from scratch, with a model for the STWIN.box board and the MLC v1 intelligent sensor, linked to an empty dataset with three classes;
2. Flash the sensor board and log some data for each defined class with the board connected to the PC;
3. Train the model with default options by using the collected logs;
4. Validate the model with the board connected to the PC by clicking on the Try MLC yellow button;
5. Create an IoT system based on the project and the model just created;
6. Deploy the IoT system by flashing the sensor board and the Raspberry Pi gateway, and deploying the needed resources in cloud (order may change);
7. Validate the model with the board connected to the gateway by activating the inference mode;
8. Collect some more data for each class by activating the datalogging mode and upload them to cloud;
9. Re-train the model with default options by using all the collected logs (old plus new ones);
10. Validate the re-trained model with the board connected to the gateway by activating the inference mode.

Everything seems to work fine when following this flow. If you still experience the issue you described, please elaborate on the steps that led to the error. Thank you.

Let me also add that when configuring a new IoT system (step 6.), you can skip flashing the SD card in case you did it already; on top of this, it is always needed to generate and copy the staiotcraft_config.ini file on the SD card and reboot the gateway.

Concerning the sensor board, be sure to flash it not only when logging data from the PC (step 2.), but also when configuring the IoT system (step 6.). In particular, if you flash it in step 6., you can use it even in step 2. without flashing it again, but the opposite does not hold; this happens because IoT systems require two firmware to be flashed on the board, while from PC only a firmware is needed.

Regards,

Davide

[Ian_P]

Hello Davide, and many thanks for your detailed reply. I have re-run the as you suggest by starting a completely fresh project and i note when i log in that the Web app seems to have been updated to a later version?

I followed the wizard on the website and everything seems to work as expected up to configuring the gateway which is where i hit a roadblock (This is happening earlier than the previous webapp version). I can flash the sensor device successfully (it shows a greenbanner confirming this) and i flash the pi with the gateway firmware (and i can log into the gateway and it appears to boot successfully). So i reach the stage of the wizard to power up both devices and check the deployment status

When I hit the yellow button i get the following screen and it doesnt appear to progress further than this status no matter how long i leave it (the circle keeps revolving and i have left it for hours).

In the previous webapp version i could get the modules to load and the stages to progress until the configuration status showed all green lights)

If i try and connect to the gateway using the yellow button i get the follwing error in the tab that is opened (the site address has a warning saying not secure in both edge and chrome)

Additionally if i connect to the sensor via ST BLE sensor app and check the firmware on the board I only see firmware in slot 1 which is FP-SNS-DATALOG2_Datalog2 v3.1.0 (despite having successfully flashed the device as part of the above wizard).

So in summary i can create a project, capture data, create the MLC, test it . But i cannot get the IoT model to deploy correctly.

Many thanks

Ian